See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/244989364

Parental meta-emotion: Validity of the Emotion-Related Parenting Styles Self-

Test and Emotion-Related Parenting Styles

Thesis · September 2012

CITATIONS READS

3 2,093

1 author:

Shawna Scott
Saskatchewan Health Authority
8 PUBLICATIONS 14 CITATIONS

SEE PROFILE

All content following this page was uploaded by Shawna Scott on 03 June 2014.

The user has requested enhancement of the downloaded file.

 PARENTAL META-EMOTION:

VALIDITY OF THE EMOTION-RELATED PARENTING STYLES SELF-TEST AND

EMOTION-RELATED PARENTING STYLES

by

Shawna Alysia Scott

A Thesis
Submitted to the Faculty of Graduate Studies
through the Department of Psychology
in Partial Fulfillment of the Requirements for
the Degree of Master of Arts at the
University of Windsor

Windsor, Ontario, Canada

2012

© 2012 Shawna A. Scott

 Parental Meta-Emotion:

Validity of the Emotion-Related Parenting Styles Self-Test and Emotion-Related

Parenting Styles

by

Shawna Scott

APPROVED BY:

______________________________________________
Dr. K. Babb
Department of Psychology

______________________________________________
Dr. B. Barrett
Department of Social Work and Women’s Studies

______________________________________________
Dr. J. Hakim-Larson, Advisor
Department of Psychology

______________________________________________
Chair of Defense
 DECLARATION OF ORIGINALITY

I hereby certify that I am the sole author of this thesis and that no part of this

thesis has been published or submitted for publication.

I certify that, to the best of my knowledge, my thesis does not infringe upon

anyone’s copyright nor violate any proprietary rights and that any ideas, techniques,

quotations, or any other material from the work of other people included in my thesis,

published or otherwise, are fully acknowledged in accordance with the standard

referencing practices. Furthermore, to the extent that I have included copyrighted

material that surpasses the bounds of fair dealing within the meaning of the Canada

Copyright Act, I certify that I have obtained a written permission from the copyright

owner(s) to include such material(s) in my thesis and have included copies of such

copyright clearances to my appendix.

I declare that this is a true copy of my thesis, including any final revisions, as

approved by my thesis committee and the Graduate Studies office, and that this thesis has

not been submitted for a higher degree to any other University or Institution.

iii
 ABSTRACT

Parents’ meta-emotion philosophy (Gottman, Katz, & Hooven, 1996) includes their

thoughts and feelings about emotions. The Meta-Emotion Interview (Katz & Gottman,

1986), used to evaluate meta-emotion philosophy, has been found to be related to

emotion socialization practices. Based on the interview, long and short form Likert-type

measures have been developed (see Gottman & DeClaire, 1997; Hakim-Larson, Parker,

Lee, Goodwin, & Voelker, 2006; Paterson, Babb, Camodeca, Goodwin, Hakim-Larson,

Voelker, & Gragg, 2012). The purpose of the present study was to evaluate the construct

validity of the questionnaires in conjunction with the original interview. Archival data

included 33 mothers with at least one child between ages 3 to 5. When mothers’ scores

on the questionnaires correlated with dimensions scores on the interview, the coefficients

were in the expected directions, suggesting further evidence for the construct validity of

the long and short forms. Additional findings and study implications are discussed.

iv
 DEDICATION

My thesis is dedicated to the primary emotion socializers in my life: my parents.

Mom, you are a source of solace and unconditional love. Dad, you always bring humour

into my life when I need a boost. To my siblings Ryan, Stephanie, Jamie, Christopher,

and Tyler: as the eldest of six children, I am supposed to be the leader of the pack. One

day, I hope you will all realize that you are the ones who inspire me. Dereck, you have

encouraged me every step of the way. My life would certainly be quite different without

you, and I thank you for bringing out the best in me. To my friends who have stood by

my side as my cheerleaders: ―Go ninja, go ninja, go!‖

v
 ACKNOWLEDGEMENTS

I thank Dr. Julie Hakim-Larson for being a remarkable supervisor and teacher

throughout this research process. I appreciate her for providing me with support and

positivity at just the right moments. I would also like to express appreciation to my

committee members, Dr. Kimberley Babb and Dr. Betty Barrett, for their invaluable

guidance and sincere desire for this study to succeed. Next, I would like to thank the

research assistants of the Emotional Competence Research Group. Sandra, Sinead,

Ashley, Brianne, Silvia, Kristen, Gillian, Erin, Aman, and Melissa all remained dedicated

to the noble task of transcribing dozens of lengthy interviews. If it were not for all of

you, I would still be listening to audio cassette recordings at this very moment. I am also

greatly appreciative of Melissa’s assistance in data checking. I would like to express my

gratitude to members of the Cuddly Cohort of Unconditional Love. We certainly know

how to survive the impossible.

vi
 TABLE OF CONTENTS

DECLARATION OF ORIGINALITY .............................................................................. iii

ABSTRACT ....................................................................................................................... iv

DEDICATION .....................................................................................................................v

ACKNOWLEDGEMENTS ............................................................................................... vi

LIST OF TABLES ............................................................................................................. ix

LIST OF FIGURES ........................................................................................................... xi

CHAPTER

I. INTRODUCTION......................................................................................................1

II. REVIEW OF LITERATURE

The Process of Emotion Socialization ..................................................3
The Construct of Meta-Emotion in Parents, Children,
and Adolescents ...........................................................................4
Parental Meta-Emotion Philosophy ......................................................5
Differentiating Emotion-Related Parenting Styles from Other
Parenting Styles ...........................................................................8
The Meta-Emotion Interview ...............................................................9
A Self-Test: What Style of Parent are you?........................................12
Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L) ...14
Emotion-Related Parenting Styles (ERPS) .........................................16
Testing the Construct Validity of the Long Form (ERPSST-L) and
Short Form (ERPS) ...................................................................20
Study Rationale ...................................................................................21
Hypotheses ..........................................................................................23

III. METHOD
Participants .........................................................................................29
Procedure ............................................................................................29
Measures .............................................................................................31

IV. RESULTS
Overview of Results ...........................................................................48
Data Screening ....................................................................................48

vii
 Preliminary Analyses ..........................................................................49
Main Analyses ....................................................................................55
Additional Analyses............................................................................63
Examples from Mothers’ and Fathers’ Meta-Emotion Interviews .....69

V. DISCUSSION
Construct Validity of the Emotion-Related Parenting Styles Self-Test
– Likert (ERPSST-L) ................................................................74
Construct Validity of the Emotion-Related Parenting Styles (ERPS) 76
Limitations and Directions for Future Research .................................78

REFERENCES ..................................................................................................................86

APPENDICES...................................................................................................................95

APPENDIX A
Inter-Rater Reliability for Coding of Meta-Emotion Interviews
(MEIs) for Coding Training, as Measured by Intra-Class
Correlation Coefficients ............................................................95

APPENDIX B
Inter-Rater Reliability for Coding of Meta-Emotion Interviews
(MEIs) for Present Study, as Measured by Intra-Class
Correlation Coefficients ............................................................96

APPENDIX C
Comparison of Scores on Meta-Emotion Interview Dimensions
between High and Low Emotion Coaching Groups, based on
Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-
L) Scores ...................................................................................97

APPENDIX D
Comparison of Scores on Meta-Emotion Interview Dimensions
between High and Low Emotion Coaching Groups, based on
Emotion-Related Parenting Styles (ERPS) Scores ...................98

VITA AUCTORIS .............................................................................................................99

viii
 LIST OF TABLES

Table 1 Correlations between the Emotion-Related Parenting Styles Self-Test –

Likert (ERPSST-L) and the Emotion-Related Parenting Styles (ERPS)
adapted from Paterson et al. (2012)……...................................................18

Table 2 Hypothesized Relations between the Emotion-Related Parenting Styles

Self-Test – Likert (ERPSST-L) and the Meta-Emotion Interview............25

Table 3 Hypothesized Relations between the Emotion-Related Parenting Styles

(ERPS) and the Meta-Emotion Interview..................................................27

Table 4 Participant Demographics..........................................................................30

Table 5 Scoring of Meta-Emotion Interview Dimensions......................................32

Table 6 Means and Standard Deviations for Dimensions on the Meta-Emotion

Interview……............................................................................................40

Table 7 Scoring of Subscales from Emotion-Related Parenting Styles Self-Test –

Likert (ERPSST-L)....................................................................................42

Table 8 Means, Standard Deviations, and Cronbach’s Alpha Coefficients for the
Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L)..........43

Table 9 Scoring of Subscales from the Emotion-Related Parenting Styles

(ERPS)………...........................................................................................45

Table 10 Means, Standard Deviations, and Cronbach’s Alpha Coefficients for the
Emotion-Related Parenting Styles (ERPS)………....................................47

Table 11 One-Way Analysis of Variance (ANOVA) Comparing Respondents and

Non-Respondents on Demographic Variables and Emotion-Related
Parenting Styles Self-Test – Likert (ERPSST-L) Scores……………...…51

Table 12 Zero-Order Correlations between Child Age, Child Sex, Family Income,
and Scores on the Meta-Emotion Interview, Emotion-Related Parenting
Styles Self-Test – Likert (ERPSST-L), and Emotion-Related Parenting
Styles (ERPS)............................................................................................53

Table 13 Zero-Order Correlations between Meta-Emotion Interview Child

Dimensions and Emotion-Related Parenting Styles Self-Test – Likert
(ERPSST-L)……………………………………………………………...57

ix
Table 14 One-Tailed Partial Correlations between Meta-Emotion Interview Child
Dimensions and Emotion-Related Parenting Styles Self-Test – Likert
(ERPSST-L) Subscales Controlling for Child Age, Child Sex, and Family
Income…………….……………...............................................................58

Table 15 Zero-Order Correlations between Meta-Emotion Interview Child

Dimensions and Emotion-Related Parenting Styles (ERPS).....................61

Table 16 One-Tailed Partial Correlations between Meta-Emotion Interview Child

Dimensions and Emotion-Related Parenting Styles (ERPS) Subscales
Controlling for Child Age, Child Sex, and Family Income.......................62

Table 17 Two-Tailed Partial Correlations between Meta-Emotion Interview Parent

Dimensions and Child Dimensions, Controlling for Child Age, Child Sex,
and Family Income………………………………....................................66

Table 18 Two-Tailed Partial Correlations between Maternal Experience and the

Meta-Emotion Interview Child Dimensions, Emotion-Related Parenting
Styles Self-Test – Likert (ERPSST-L) Subscales, and Emotion Related
Parenting Styles (ERPS) Subscales …......................................................68

x
 LIST OF FIGURES

Figure 1 A Depiction of the Main Variables Relevant to the Current Study........22

xi
 CHAPTER I

INTRODUCTION

Parents teach their children how to express, cope with, and respond to emotions in

everyday interactions. Parental meta-emotion philosophy is a construct that encompasses

thoughts and feelings about emotions and has been found to translate into emotion

socialization practices. The Meta-Emotion Interview (Katz & Gottman, 1986) has been

the principal way to measure this construct. In this interview, parents describe their past

and current experiences with the emotions of sadness, anger, and fear; goals in teaching

their children about these emotions; and understanding how their children express and

cope with emotions. This interview produces continuous scores on dimensions including

parents’ awareness, acceptance, and regulation of their own emotions, as well as the

parents’ awareness, acceptance, coaching, and regulation of their children’s emotions.

Because the Meta-Emotion Interview is time intensive to administer and score,

three questionnaire versions were developed based on the framework by Gottman, Katz,

and Hooven (1996): the true-false version (Gottman & DeClaire, 1997), the long form

Likert scale called the Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L;

Gottman & DeClaire, 1997, modified by Hakim-Larson, Parker, Lee, Goodwin, &

Voelker, 2006), and the short form Likert scale called the Emotion Related Parenting

Styles (ERPS; Gottman & DeClaire, 1997, modified by Paterson, Babb, Camodeca,

Goodwin, Hakim-Larson, Voelker, & Gragg, 2012).

The ERPSST-L is an 81-item, 5-point Likert-type self-report measure of parental

meta-emotion. Relative to the Meta-Emotion Interview, the ERPSST-L is time-efficient.

Each item on the ERPSST-L describes one of the four emotion-related parenting styles

1
originally identified by Gottman, Katz, and Hooven (1996): emotion coaching, laissez-

faire, dismissing, and disapproving. The ERPS is a 20-item short-form Likert-type

questionnaire that is a subset of the items in the ERPSST-L. The ERPS produces

continuous scores on four emotion-related parenting styles: emotion coaching, parental

acceptance of negative emotion, parental rejection of negative emotion, and feelings of

uncertainty/ineffectiveness in emotion socialization.

To date, these two measures have been evaluated in terms of internal validity,

convergent validity, and some preliminary construct validity. The purpose of the present

study was to test the construct validity of both the ERPSST-L and the ERPS. Construct

validity would be demonstrated if scores on the ERPSST-L and ERPS correlate with

scores on the Meta-Emotion Interview in the directions predicted by meta-emotion

theory.

The following sections in the present paper include reviews of the process of

parental emotion socialization, the meta-emotion construct, and parental meta-emotion

philosophy. Next, emotion-related parenting styles will be distinguished from other

parenting styles. The Meta-Emotion Interview and three measures of parental meta-

emotion will be described, followed by the objectives, rationale, and hypotheses of the

present study.

2
 CHAPTER II

REVIEW OF LITERATURE

The Process of Emotion Socialization

It is well-documented that socialization plays a substantial role in the emotional

development of children. For young children, parents are considered to be the most

essential socializers of emotional and social competence. Interest in emotion

socialization was largely influenced by Haim Ginott (1965), who emphasized that

socializers can teach children to understand emotions by using empathy and respectful

communication (as cited in Gottman & DeClaire, 1997, p. 34).

This emotion socialization process is thought to be bidirectional, in that children

can influence parenting behaviours just like parenting behaviours can influence children

(Saarni, 1999). For example, a mother may adjust her interactions with her child based

on her perceptions of the child’s temperament (Eisenberg, 1996). Moreover, a mother

may adjust her interactions with her child based on how she perceives her child’s

tendencies in dealing with emotions (Fabes, Eisenberg, Karbon, Bernzweig, Speer, &

Carlo, 1994). To illustrate this bidirectional process, Fabes et al. (1994) assessed

emotional, physiological, and prosocial interpersonal reactivity between parent-child

dyads during a storytelling task. Child age differences were found in which mothers of

the young children were more likely to attempt to induce a positive mood in the children

in order to minimize their unpleasant responses. Because younger children are often

thought to have emotional skills that are less advanced compared to older children, child

age influenced parenting behaviours. It is also relevant to note that this tendency

3
primarily occurred when mothers of young children believed that the child would likely

become emotionally aroused in the storytelling task. Additionally, this finding by Fabes

et al. (1994) suggested that parental attitudes and beliefs do play a role in a parent’s

shaping and reactivity to the emotional experiences of children.

Emotion socialization, which is thought to be shaped by the attitudes, culture, and

beliefs of parents, can be direct or indirect (Eisenberg, Spinrad, & Cumberland, 1998),

and the recognition of the need to examine these factors has been increasing (Dunsmore,

Her, Halberstadt, & Perez-Rivera, 2009). Direct socialization involves the behaviours of

the socializer that reflect his or her cognitions and goals related to emotions (Eisenberg,

Cumberland, & Spinrad, 1998). The three main ways parents directly socialize emotional

development in children include parental reactions to children’s emotion, parental

expressiveness, and parent-child discussion of emotion (Eisenberg, Cumberland, &

Spinrad, 1998). On the contrary, indirect socialization involves exchanges and

behaviours that affect child’s emotional experience, expression, and understanding;

however, it is not a direct reflection of the socializer’s beliefs and goals related to

emotion (Eisenberg, Spinrad, & Cumberland, 1998). Thus, emotion socialization occurs

through both direct and indirect pathways.

The Construct of Meta-Emotion in Parents, Children, and Adolescents

Meta-emotion encompasses feelings, cognitions, and actions related to the

experience of emotions (Gottman et al., 1996). Measuring this construct can be useful in

better understanding how one responds to negative emotions in self and others. A

negative emotion is not necessarily one that is bad or maladaptive. The term negative

emotion is used to describe the emotions that are typically unpleasant (e.g., sadness,

4
anger, and fear). Positive emotions are those that are typically pleasant (e.g., happiness).

In meta-emotion philosophy, all emotions are described as potentially adaptive.

Assessing parental meta-emotion philosophy in response to negative emotions provides a

way to understand how parents react during potentially stressful situations based on their

own traits, the nature of the situation, and traits of the child (Hakim-Larson, Dunham,

Vellet, Murdaca, & Levenbach, 1999). Emotional intelligence is a construct often

discussed in meta-emotion theory. Emotional intelligence is often described as one’s

ability to experience and express emotions conscientiously and in a controlled manner

(Jäger &Bartsch, 2006). Jäger and Bartsch (2006) pointed out that there is a need to

determine the role of meta-emotion in the self-awareness and self-control of emotions.

Parental Meta-Emotion Philosophy

One of the factors influencing emotion socialization is thought to be one’s meta-

emotion philosophy. Meta-emotion philosophy represents parents’ attitudes toward

emotion and their style of communicating emotions with their children (Gottman et al.,

1996). Gottman and colleagues (1996) defined parental meta-emotion philosophy as ―an

organized set of feelings and thoughts about one’s own emotions and one’s children’s

emotions‖ (p. 243). Parental meta-emotion philosophy is an important consideration

when examining a parent’s verbal and nonverbal emotion socialization practices and

behaviours. For example, research findings suggest that maternal meta-emotion

philosophy is associated with socialization behaviour when mothers interact with their

children ages 4-5 (Gottman et al., 1996). Katz, Gottman, and Hooven (1996) posited that

the exploration of parental meta-emotion philosophy can aid in understanding the relation

between parenting behaviours and children’s physiological regulation and adjustment.

5
 Eisenberg (1996) brought up an important question: how does a parent’s meta-

emotion philosophy relate to his or her parenting behaviour? Hakim-Larson and

colleagues (2006) noted that meta-emotion is a combination of belief about the

acceptability of emotions and belief about active emotion socialization. These two

dimensions can better explain the construct of meta-emotion. Under the framework of

meta-emotion theory, one may recognize how a parent’s understanding and awareness of

emotions can translate into socialization practices (Hakim-Larson et al., 2006; Katz,

Maliken, & Stettler, 2012).

Gottman et al. (1996) also suggested that a parent’s emotional awareness and

coaching can relate to his or her parenting behaviours, but may also lead directly to child

outcome. For instance, Gottman and colleagues (1996) found that children of emotion

coaching parents at age five were predicted to be rated as socially competent by teachers

at age eight.

Gottman’s theoretical model of parental meta-emotion philosophy has produced

four emotion-related parenting styles: emotion coaching, laissez-faire, dismissing, and

disapproving. These emotion-related parenting styles are related to child outcome, as

described in the following sections.

Emotion coaching parenting. Emotion coaching is the most positive meta-

emotion philosophy in terms of parent-child interaction and child outcomes. Emotion-

coaching parents are high in emotional awareness, acceptance, regulation (Gottman &

DeClaire, 1997), and coaching (Gottman et al., 1996) of their children’s emotions. For

emotion coaching parents, emotion is socialized by emotional display, empathic listening,

labelling and validating emotions, offering guidance for emotion regulation, and by

6
teaching problem-solving skills (Gottman & DeClaire, 1997). Characteristically, this

parenting style is used by parents who have a healthy relationship with their spouse.

These parents also feel comfortable with their own emotions and support their children,

through positive parenting, in their exploration and expression of emotions. According to

Gottman and his colleagues (1996), outcomes for emotion-coached children are positive

in that they experience less stress and illness, have better self-regulation skills, higher

levels of academic achievement, and more positive relationships with peers. Children of

parents who adopt an emotion-coaching parenting style tend to develop strong emotion

regulation and social skills (Gottman & DeClaire, 1997).

Laissez-faire parenting. Parents with a laissez-faire meta-emotion philosophy

are typically high in emotional awareness and acceptance but low in emotional regulation

and coaching of their children’s emotions (Gottman & DeClaire, 1997). Compared to

emotion coaching parents, little guidance on emotion regulation is used by laissez-faire

parents. Laissez-faire parents rarely set limits on behaviour and are unlikely to teach

children how to solve socio-emotional problems.

Emotion dismissing parenting. Parents who are dismissing of emotion believe

that negative emotions are harmful (Gottman et al., 1996). Such parents are low in

emotional awareness, acceptance, coaching, and regulation (Gottman & DeClaire, 1997).

They would much rather avoid addressing negative emotions at all. Children of emotion

dismissing parents may face difficulty in solving socio-emotional problems and may

learn that emotions such as sadness, anger, and fear are wrong to experience and express,

leading to a difficulty with emotion regulation (Gottman & Declaire, 1997).

7
 Emotion disapproving parenting. This is considered the harshest of the four

styles in that criticism and punishment may be used when the child expresses

disapproved emotions. Parents using this style are low in emotional awareness,

acceptance, regulation, and coaching (Gottman & DeClaire, 1997). It is important to note

that a disapproving parenting style can lead to particular difficulties for children.

Children may be less emotionally and socially competent (Gottman & DeClaire, 1997)

and tend to have elevated anxiety and poor emotion regulation (Lagacé-Séguin & Coplan,

2005).

Differentiating Emotion-Related Parenting Styles from Other Parenting Styles

Gottman and his colleagues (1996) emphasized how emotion-related parenting

styles differ from general parenting styles. Baumrind (1971) established four general

parenting styles: authoritarian, authoritative, permissive, and uninvolved, which are

determined by various combinations of warmth and control. In contrast to the four

general parenting styles, emotion-related parenting styles determine how parents set

guidelines related to the emotional experiences of the child. Eisenberg (1996) supported

the notion that there is, for example, a distinction between parental derogation (related to

a general parenting style) and parental expression of anger (related to an emotion-related

parenting style). For example, an emotion-coaching parent may inhibit parental

negativity in response to a child’s negative emotion. This is not the same as an

authoritative parenting style which emphasizes a disciplinary style. In essence, an

emotion-related parenting style describes a parent’s response to a child’s emotional

experience, while a parent’s disciplinary style describes a parent’s response to a child’s

behaviour. Similarly, there is a distinction between parental scaffolding-praising (e.g.,

8
establish a structured environment and provide praise and approval in response to a

child’s appropriate actions) and warmth (positivity; Gottman et al., 1996).

The Meta-Emotion Interview

The Meta-Emotion Interview (Katz & Gottman, 1986) is a semi-structured, audio-

taped interview that begins by evaluating parents’ past and present experience with

sadness, anger, and fear. Parents are then asked to describe their children’s experience

with those emotions. Next, parents look at a list of emotions and discuss an emotion they

prefer, an emotion they dislike the most, as well as the emotion with which the child has

the most difficulty. Finally, parents verbally summarize what they are trying to teach

their children about emotions, in general. This interview often has been used by

researchers in the last decade due to its ability to generate rich data on parental meta-

emotion philosophy.

In literature on parenting, the Meta-Emotion Interview has been applied in various

settings and has been used to assess associations between emotion-related parenting

styles and child outcome of social skills and adjustment. For example, the Meta-Emotion

Interview was used to evaluate emotion socialization processes and child outcome in

African-American families with school-age children (Cunningham, Kliewer, & Garner,

2009). Though maternal meta-emotion philosophy was associated with emotional

understanding for boys and girls, some components of emotional understanding differed

based on child gender. Emotional understanding was a mediator between maternal

emotion socialization and the internalizing behaviour of boys. For girls, emotional

understanding was a mediator between maternal emotion socialization and social skills.

Additionally, they found that emotion regulation mediated emotion socialization in boys’

9
adjustment, but not for girls. The Meta-Emotion Interview also has been used to assess

the relation between meta-emotion philosophy and child outcome in families with

domestic violence (e.g., Katz, Hunter, & Klowden, 2008; Katz & Windecker-Nelson,

2006). One interesting finding from the study by Katz and colleagues (2008) was that

emotion coaching in mothers may actually function as a buffer for children exposed to

intimate partner violence. Specifically, children of emotion coaching mothers reacted to

peer provocation in a more adaptive, less negative manner when compared to children of

mothers low in emotion coaching. In another study, the Meta-Emotion Interview was

used to evaluate the emotion regulation of children, as well as the emotion socialization

of mothers who physically maltreat their children and mothers who do not physically

maltreat their children (Shipman, Schneider, Fitzgerald, Sims, Siwsher, & Edwards,

2007). Shipman and colleagues (2007) found that mothers who physically maltreated

their children tended to use less emotion coaching when their children displayed negative

emotion in a mother-child interaction task as compared to non-maltreating mothers.

According to Shipman et al. (2007), mothers who physically maltreated their children

may view negative emotion as useless. In turn, children’s experience of negative

emotions may be invalidated and fewer adaptive emotion regulation strategies may be

used by the children.

Construct validity of the Meta-Emotion Interview. The Meta-Emotion

Interview has been described as the gold-standard measure of the construct of parental

meta-emotion philosophy. It is important to describe what exactly is meant by construct

validity because it is a term frequently misused in the literature. According to Haynes

(2001), construct validity ―comprises the evidence and rationales indicating the degree to

10
which data from an assessment instrument measures the targeted construct; includes all

evidence bearing on the measure and encompasses all types of validity‖ (p. 239). In

other words, construct validity demonstrates how well variables of interest represent the

construct (Cherulnik, 2001). According to Clark and Watson (1995), construct validity is

a primary goal when developing scales.

Meta-emotion also has been studied by examining parents’ scores on the Meta-

Emotion Interview and peer relations among children with conduct problems (Katz &

Windecker-Nelson, 2004). Even for children who are aggressive, more positive play

with peers occurred when mothers were higher in emotion awareness and coaching.

Meta-emotion also has been examined in families with older children and

adolescents. In one study, the Meta-Emotion Interview was used to examine the relation

between maternal meta-emotion philosophy, adolescent affect, and adolescent

temperament (Yap, Allen, Leve, & Katz, 2008). Yap et al. (2008) found that maternal

meta-emotion philosophy was related to maternal emotion socialization behaviours

during mothers’ interactions with their adolescent-aged children. Further, the Meta-

Emotion Interview was used in a study of maternal meta-emotion philosophy in the

families of adolescents with depressive symptomatology (Katz & Hunter, 2007). The

sample contained thirty dyads of adolescents and their mothers. Results suggested that

adolescents of mothers who scored high in acceptance of their own emotions were more

likely to score lower in depression symptomatology, lower in externalizing problems, but

higher in self-esteem. Overall, these findings suggest that the Meta-Emotion Interview

has been a useful tool to study parental meta-emotion.

11
 As interest in meta-emotion philosophy was generated, Gottman and DeClaire

(1997) wrote a parenting book called Raising an Emotionally Intelligent Child: The Heart

of Parenting and presented an 81-item true/false self-report measure. By completing this

measure – entitled ―A Self-Test: What Style of Parent are you?‖ – parents could assess

and determine their primary emotion-related parenting style. This measure was later

converted into a long form Likert-type questionnaire known as the Emotion-Related

Parenting Styles Self-Test – Likert (ERPSST-L; Gottman & DeClaire, 1997, modified by

Hakim-Larson et al., 2006) and also a short form Likert-type questionnaire called

Emotion-Related Parenting Styles (ERPS; Gottman & DeClaire, 1997, modified by

Paterson, Babb, Camodeca, Goodwin, Hakim-Larson, Voelker, & Gragg, 2012). The

construct validity of the long form and the short form has yet to be tested in conjunction

with the Meta-Emotion Interview. In summary, meta-emotion philosophy also can be

measured in a true/false questionnaire, a long form Likert-type questionnaire, and a short

form Likert-type questionnaire. These measures are individually described below.

A Self-Test: What Style of Parent are you?

This is an 81-item true/false self-report measure (Gottman & DeClaire, 1997) that

is also referred to as the Emotion-Related Parenting Styles Self-Test – True/False

(ERPSST-T/F). Each statement represents one of the four emotion-related parenting

styles: emotion coaching, laissez-faire, dismissing, and disapproving. Examples of items

on this measure include, ―I think it’s good for kids to feel angry sometimes‖ and ―I think

sadness is okay as long as it’s under control‖ (p. 42-48). An average score for each scale

is calculated. A parent’s primary emotion-related parenting style is the one that results in

the highest average score. Scoring produces a continuous score on each of the emotion-

12
related parenting styles. All parents have the four emotion-related parenting styles to

some extent, but it is a matter of degree. Typically, average Likert-type scores for each

subscale are calculated and used in analyses.

Hakim-Larson et al. (2006) tested the psychometric properties of the ERPSST-

T/F measure on a sample of 89 mothers and 11 fathers of children ages 2 to 6. Internal

consistency was found to be from .33 to .87 over the four parenting styles, with the

laissez-faire scale as the weakest. Social desirability was found to relate to variables, and

was controlled for in analyses. Test-retest reliability after two to three months was good,

suggesting that a parents’ primarily endorsed style is somewhat stable over time.

Evidence for test-retest reliability and internal consistency reliability were found (Lee,

Hakim-Larson, & Voelker, 2000). Hakim-Larson and colleagues (2006) found that the

emotion coaching parenting style was endorsed most often (in 91 out of 100 parents from

the first administration) and the remaining nine were laissez-faire. Lee (1999) used a

confirmatory factor analysis to test the construct validity of the ERPSST-T/F measure.

Construct validity would have been supported if four parenting style constructs were

represented by the ERPSST-T/F. However, Lee (1999) found the model fit of the

ERPSST-T/F to be poor to mediocre. As described by Clark and Watson (1995),

dichotomous response formats have been criticized extensively in the literature due to

their tendency to be less reliable and less stable than ones with multiple choices. Though

the ERPSST-T/F was useful in the sense that it was quick to administer, it required

reconstruction due to its poor psychometric properties. This led to the development of

the Likert-scale version of the Emotion-Related Parenting Styles Self-Test.

13
Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L)

The ERPSST-L (Gottman & DeClaire, 2007, modified by Hakim-Larson et al.,

2006, reproduced with permission of Simon & Schuster) was developed by converting

the true/false measure (ERPSST-T/F) into a 5-point Likert-type scale (1 = always false, 2

= mostly false, 3 = somewhat true/false, 4 = mostly true, and 5 = always true). All 81

items from the ERPSST-T/F remained on this scale to measure parents’ typical style of

teaching their children about emotion. Average Likert-type scores for each subscale are

calculated, in which a higher score indicates greater endorsement of that emotion-related

parenting style. A parent’s primary emotion-related parenting style is the subscale that

results in the highest average.

Psychometric properties of the ERPSST-L. Psychometric properties of the

ERPSST-L were assessed by Hakim-Larson and colleagues (2006) using a community

sample of 21 mothers and 10 fathers of children ages 3 to 6. Preliminary findings

demonstrated evidence of adequate to very good internal consistency reliability (α = .72

to .91), showing improvement over the ERPSST-T/F. After controlling for social

desirability and parent gender, Hakim-Larson and colleagues (2006) found support for

convergent validity of the ERPSST-L with self-report measures of self-expressiveness

(positive dominance, positive submissive, negative dominance, and negative submissive;

Halberstadt Cassidy, Stifter, Parke, & Fox, 1995), attitudes toward children’s emotional

expressiveness (Saarni, 1985), and ability to cope with negative emotions (Fabes,

Eisenberg, & Bernzweig, 1990). Partial correlation analyses in the study by Hakim-

Larson et al. (2006) produced statistically significant positive correlations between scores

on the emotion coaching subscale and positive expressiveness and expressive

14
encouragement; positive correlations between scores on the laissez-faire subscale and

expressive encouragement; positive correlations between scores on the dismissing

subscale and the self-reported distress, use of punishment, and emotion-minimization;

negative correlation between scores on the dismissing subscale and expressive

encouragement; and positive correlations between a disapproving parenting style and

distress reactions, punitive reactions, and minimization reactions.

The ERPSST-L has been a useful measure to researchers interested in meta-

emotion philosophy. The ERPSST-L was used as a measure in a study published by

Mills, Freeman, Clara, Elgar, Walling, and Mak (2007) to examine parents’ proneness to

shame, use of psychological control, overprotective behaviour, and critical, rejecting

behaviour. The sample included 198 mothers and fathers of preschoolers. Mills et al.

(2007) used a principal components analysis and produced the expected four components

of the ERPSST-L. Because they wanted a measure of parents’ negative approach to the

child, they only used the disapproval scale for their main analyses. They found that this

scale had a significant, positive correlation with measures of spousal overprotection, self-

criticism, criticism towards spouse, guilt, shame, worry about danger, worry about

discomfort, and anger reactivity.

Similarly, another group of researchers decided to select only the disapproving

scale of the ERPSST-L, producing alphas of .88 for mothers and .85 for fathers (Walling,

Mills, & Freeman, 2007). In this study of parenting cognitions and parental use of

psychological control, Walling et al. (2007) found that for fathers of girls the disapproval

of negative emotions predicted a parent’s use of guilt/shame induction. This finding did

not occur for fathers of boys. In addition, they also found that maternal and paternal

15
sensitivity to hurtful messages and disapproval of negative emotions were related to an

increased use of psychological control (i.e., parental intrusion and manipulation of their

children’s feelings, thoughts, and perspectives on the parent-child relationship).

In summation, the ERPSST-L can be used in a number of ways to examine

specific emotion-related parenting styles. Recently, the ERPSST-L was transformed into

a short-form measure called the Emotion-Related Parenting Styles (ERPS). Despite the

utility of the ERPSST-L, no study to date has tested the construct validity of the

ERPSST-L or the ERPS in conjunction with the original Meta-Emotion Interview.

Emotion-Related Parenting Styles (ERPS)

The ERPS (Gottman & DeClaire, 1997, modified by Paterson et al., 2012) is a 20-

item short-form questionnaire. Items on the ERPS are a subset of selected items from the

long form ERPSST-L. The ERPS produces scores on four different emotion-related

parenting styles: emotion coaching, parental rejection of negative emotion, parental

acceptance of negative emotion, and feelings of uncertainty/ineffectiveness in emotion

socialization.

The sample used by Paterson et al. (2012) consisted of 107 mothers of children

without a developmental disability and 107 mothers of children with a developmental

disability who completed the ERPSST-L. Psychometric properties of this short-form

measure were satisfactory in both samples, with Cronbach’s alphas ranging from .70 to

.80. The authors found support for convergent validity in that the ERPS subscales

correlated in the expected directions with subscales of the Coping with Children’s

Negative Emotions Scale (CCNES) and subscales of the Parent Attitude toward

Children’s Expressiveness Scale (PACES). The relation between scales on the ERPSST-

16
L and ERPS, as found by Paterson and colleagues (2012) are found in Table 1. The

ERPS was later revised to include gender-neutral language, removing any pronouns from

the questionnaire. A description of each of the four ERPS subscales follows.

Emotion coaching. Like the emotion coaching subscale from the ERPSST-L, the

emotion coaching subscale of the ERPS is used to assess a parent’s acceptance of his or

her child’s emotional expression and desire to teach the child about emotions (e.g.,

―When my child is sad, I try to help the child explore what is making him or her sad‖).

Paterson et al. (2012) found a positive correlation between the emotion coaching subscale

and that of the ERPSST-L, r = .75, p < .001. Thus, descriptors of emotion coaching

appear to remain quite consistent with the emotion coaching subscale of the ERPSST-L.

Parental rejection of negative emotion. This scale measures the degree to

which parents reject their children’s experience of negative emotions (e.g., ―When my

child gets angry, my goal is to get him or her to stop‖). Paterson et al. (2012) found

evidence for collapsing the dismissing and disapproving styles to produce the parental

rejection subscale. They found that this subscale correlated with the dismissing (r = .68,

p < .001) and disapproving (r = .71, p < .001) subscales of the ERPSST-L.

Parental acceptance of negative emotion. Parents who endorse items on this

scale tend to accept negative emotions but provide little guidance in helping the child

work through those emotions (e.g., ―I think it’s good for kids to feel angry sometimes‖).

Paterson et al. (2012) reported that parental acceptance positively correlated with the

ERPSST-L’s emotion coaching (r = .67, p < .001) and laissez-faire (r = .32, p < .001)

subscales, and negatively correlated with dismissing (r = -.27, p < .001) and disapproving

(r = -.31, p < .001) parenting styles.

17
Table 1

Correlations between the Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-

L) and the Emotion-Related Parenting Styles (ERPS) adapted from Paterson et al. (2012)

ERPS Subscales

Emotion Parental Parental Uncertainty/

Coaching Acceptance Rejection Ineffectiveness

ERPSST-L Emotion Coaching .75*** .67*** -.10 -.24***

Subscales Laissez-Faire .06 .32*** -.01 .38***

Dismissing -.18* -.27*** .68*** .47***

Disapproving -.18** -.31*** .71*** .45***

Note. Permission to reproduce these coefficients was granted by S. Denham, editor of

Early Education and Development (personal communication, September 13, 2012).

Permission to reproduce nonsignificant coefficients from the original data set was granted

by A. Paterson (personal communication, September 14, 2012). *p < .05, **p < .01,

***p < .001.

18
 Feelings of uncertainty/ineffectiveness in emotion socialization. This subscale

is unique to the ERPS in that parents who endorse this subscale typically feel uncertain or

ineffective with regards to handling their child’s experience of negative emotions (e.g.,

―When my child is angry, I’m not quite sure what he or she wants me to do‖). This

uncertainty/ineffectiveness subscale negatively correlated with the ERPSST-L’s emotion

coaching subscale (r = -.24, p = .001), but positively correlated with the laissez-faire (r =

.38, p < .001), dismissing (r = .47, p < .001), and disapproving (r = .45, p < .001)

subscales (Paterson et al., 2012).

Aside from Paterson et al. (2012), there has been one other attempt to adapt the

true/false self-test (ERPSST-T/F) into a psychometrically-sound short-form measure.

Lagacé-Séguin and Coplan (2005) produced a 22-item, 5-point Likert-type self-report

measure called the Maternal Emotional Styles Questionnaire (MESQ). Parents rated their

level of agreement on statements on a scale ranging from 1 (strongly agree) to 5 (strongly

disagree). This measure was found to be internally consistent, correlated with the Meta-

Emotion Interview, and established convergent validity with parental goals that are

parent-centred and empathetic (Lagacé-Séguin & Coplan, 2005). Their measure

produced a two-factor structure to assess emotion-coaching and emotion-dismissing

philosophies. There are two known published studies that included the MESQ as a

measure. In the first, Lagacé-Séguin and Gionet (2009) found that parental meta-

emotion, as assessed by the MESQ, and temperament were predictors of coping skills for

adolescents aged 10-13 years. In a second study that used the MESQ, Baker, Fenning,

and Crnic (2010) examined relations among various parental emotion socialization

behaviours, including reactions to children’s negative emotions in a sample of parents (88

19
mothers and 76 fathers) of 8-year-old children. They found that paternal emotion

coaching attitudes, which were associated with the social competence of the children,

predicted their reactions to child emotion, emotional expression in the family, and use of

an emotion-coaching approach. The 20-item ERPS differs from the 22-item MESQ in

several ways. First of all, the ERPS assesses four theoretical meta-emotion philosophies,

while the MESQ assesses only emotion-coaching and emotion-dismissing philosophies.

Additionally, the development of the ERPS began with a smaller pool of items (81) than

the MESQ (over 100).

Testing the Construct Validity of the Long Form (ERPSST-L) and Short Form

(ERPS)

The purpose of the present study was to test the construct validity of both the

ERPSST-L (long form) and the ERPS (short form) in conjunction with the original Meta-

Emotion Interview. Though the ERPSST-L and ERPS scales may be useful to

researchers, there is a need to determine if scores on these scales correlate with scores on

the Meta-Emotion Interview in the expected directions. In order to test the construct

validity of the long form questionnaire, scores on the ERPSST-L were compared to

dimension scores of the Meta-Emotion Interview. The Meta-Emotion Interview

dimensions include parents’ awareness, acceptance, and regulation of their own

emotions, as well as the parents’ awareness, acceptance, coaching, and regulation of their

children’s emotions. Meta-emotion theory was used in formulating hypotheses regarding

the anticipated direction of correlation. Finding these predicted relationships would show

that construct validity has been established (Crano & Brewer, 2002), and the ERPSST-L

would be considered a valid measure of meta-emotion. For hypothetical constructs, like

20
parental meta-emotion philosophy, it is optimal to examine whether scores on the new

measure conform theoretically to the target construct (Smith, 2005).

The next aim of the present study was to evaluate the construct validity of the

short form ERPS by comparing subscale scores to the dimension scores of the Meta-

Emotion Interview. If scores correlate in the expected direction, it would provide support

for the construct validity of the ERPS as a short-form measure of parental meta-emotion.

Study Rationale

A summary of all measures used in the present study and their subscales is

displayed in Figure 1. Hakim-Larson and colleagues (2006) sought to develop and

evaluate the ERPSST-L because ―such self-report measures take less time and fewer

resources to administer and score than lengthy, structured interviews, and could

potentially facilitate relevant research on parenting meta-emotion‖ (p. 231). Hakim-

Larson et al. (2006) also stated ―future studies on the ERPSST-L will need to include a

comparison of parents’ scores on the four self-report scales to the scores obtained from

the coding of the meta-emotion interview as originally developed by Gottman and his

colleagues‖ (p. 248). The objective of the present study is to meet this very need.

There is a need for a time-efficient, valid measure of meta-emotion, as the Meta-

Emotion Interview can take up to 135 minutes to complete. Audio-recorded, semi-

structured interviews are also susceptible to data loss due to technical difficulties and

insufficient prompting by interviewers, as was the case in a study that used the Meta-

Emotion Interview (DeOliveira, Moran, & Pederson, 2005). In addition to these

problems, data analysis can be time and cost prohibitive. As found in the present study,

transcribing one Meta-Emotion Interview can take anywhere from five to twelve hours.

21
 Meta-Emotion Interview “A Self-Test: What Style of Parent are
A parent is interviewed. You?”
Experiences related to sadness, Emotion-Related Parenting Styles Self-Test
anger, and fear pertaining to – True/False (ERPSST-T/F)
self and child are explored. A parent completes an 81-item dichotomous
______________________ measure
_______________________
Parent Dimensions
◦ Awareness ◦ Emotion coaching
◦ Acceptance ◦ Laissez faire
◦ Regulation ◦ Dismissing
◦ Disapproving
Child Dimensions
◦ Awareness
◦ Acceptance Emotion-Related Parenting Styles Self-Test
◦ Coaching – Likert (ERPSST-L)
◦ Regulation A parent completes a measure with 81-items
on a 5-point Likert-type scale
_______________________

◦ Emotion coaching
◦ Laissez faire
◦ Dismissing
◦ Disapproving

Emotion-Related Parenting Styles (ERPS)

A parent completes a subset of 20 items from
the ERPSST-L; 5-point Likert-type scale
_______________________

◦ Emotion coaching
◦ Parental rejection of negative emotion
◦ Parental acceptance of negative emotion
◦ Feelings of uncertainty/ineffectiveness in
emotion socialization

Figure 1. A depiction of the main variables relevant to the current study.

22
 Further, Meta-Emotion Interview coders are required to complete extensive

training, which involves the use of manuals and audio tapes for approximately 20 hours

(Cunningham et al., 2009). Once practice tapes are completed and adequate inter-rater

reliability has been established, the Meta-Emotion Interviews need to be coded; this can

also be a lengthy process. Cunningham and colleagues (2009) reported that each

interview took 45 to 60 minutes to code. By using a validated paper-pencil questionnaire,

meta-emotion researchers will save time and resources. Validating self-report measures

addresses an unmet need in the study of meta-emotion philosophy.

If the ERPSST-L and ERPS are found to have good construct validity, they may

be useful in addressing parental strengths related to emotion socialization practices.

Because these are quick self-report measures, they may also be useful in pre and post-test

for interventions in family therapy and in promoting positive parenting.

Hypotheses

Hypotheses related to group differences: Child gender. In a study by

Cunningham and colleagues (2009), the maternal emotion socialization process did not

differ for mothers of boys and mothers of girls in an African American sample.

However, they found that emotion socialization practices related to emotion regulation

for boys but not for girls. Thus, in the present study it was expected that main study

variables would differ significantly for mothers of boys and mothers of girls.

Hypotheses related to the construct validity of the ERPSST-L long form.

This hypothesis pertains to the expected direction of correlations between the four Meta-

Emotion Interview child dimension scores (parent’s awareness of child’s emotions,

23
acceptance of child’s emotions, parent coaching of child’s emotions, and parent’s account

of child’s regulation of emotion) and the four ERPSST-L subscale scores (emotion

coaching, laissez-faire, dismissing, and disapproving). Each parent receives a score on

each of these variables, with a higher score indicating greater endorsement. On specific

construct validity analyses, the focus was on the Meta-Emotion Interview child

dimensions, as opposed to the Meta-Emotion Interview parent dimensions, because

ERPSST-L focused on parenting behaviours when it comes to children’s emotions, not

specifically on the parent’s own management of his or her emotions. Due to the

specificity of the expected direction of correlations, Table 2 displays a summary of the

major hypotheses for this study regarding the relation between the Meta-Emotion

Interview and the ERPSST-L. These hypotheses are all based on meta-emotion theory

because Meta-Emotion Interview scores do not directly produce emotion-related

parenting styles, like the ERPSST-L does. Gottman, Katz, and Hooven (1997) stated

that the Meta-Emotion Interview can be assessed qualitatively to determine a parent’s

meta-emotion philosophy by examining content for statements that describe the

categories of emotion-related parenting styles. As pointed out by Shine and Wampler

(1997), this qualitative classification of parenting styles from the Meta-Emotion

Interview dimensions has never been explained in the literature and specific procedures

are not available. Instead of classifying emotion-related parenting styles within the Meta-

Emotion Interview, the relation between Meta-Emotion Interview dimensions and

emotion-related parenting style scores were hypothesized using meta-emotion theory.

24
Table 2

Hypothesized Relations between the Emotion-Related Parenting Styles Self-Test – Likert

(ERPSST-L) and the Meta-Emotion Interview

ERPSST-L Subscales
Emotion Laissez- Dismissing Disapproving
Coaching Faire
Awareness Positive Positive Negative Negative
of Child’s Correlation Correlation Correlation Correlation
Emotions (Gottman & (Hakim- (Gottman & (Hakim-
DeClaire, Larson et al., DeClaire, Larson et al.,
1997, p. 63; 2006, p. 1997, p. 50; 2006, p. 231)
Hakim- 230) Hakim-
Larson et al., Larson et al.,
2006, p. 230) 2006, p. 231)
Meta- Acceptance Positive Positive Negative Negative
Emotion of Child’s Correlation Correlation Correlation Correlation
Interview Emotions (Gottman & (Hakim- (Gottman & (Gottman &
Dimensions DeClaire, Larson et al., DeClaire, DeClaire,
1997, p. 63; 2006, p. 1997, p. 50) 1997, 1997, p.
Hakim- 230) 51)
Larson et al.,
2006, p. 230)
Account of Positive Negative Negative Negative
Child’s Correlation Correlation Correlation Correlation
Regulation (Gottman & (Hakim- (Gottman & (Gottman &
of DeClaire, Larson et al., DeClaire, DeClaire,
Emotions 1997, p. 52; 2006, p. 1997, p. 56) 1997,
Hakim- 230) p. 51)
Larson et al.,
2006, p. 230)
Coaching Positive Negative Negative Negative
of Child’s Correlation Correlation Correlation Correlation
Emotions (Gottman et (Gottman & (Gottman & (Gottman &
al., 1996, p. DeClaire, DeClaire, DeClaire,
244) 1997, p. 50) 1997, p. 50) 1997, p. 51)

25
 Hypotheses related to the construct validity of the ERPS short form. It was

expected that the ERPS would establish construct validity in conjunction with the original

Meta-Emotion Interview by correlating in directions predicted by meta-emotion theory

(refer to Table 3). Consistent with the hypothesized findings for the ERPSST-L long

form, it was expected that the emotion coaching subscale of the ERPS would show a

positive correlation with the Meta-Emotion Interview child dimensions just as was

expected for the emotion coaching dimension of the ERPSST-L. Second, it was expected

that the parental rejection of negative emotion subscale would show a negative

correlation with each of the Meta-Emotion Interview dimensions. Third, because

parental acceptance of negative emotion was found to correlate highly with the ERPSST-

L’s emotion coaching subscale (Paterson et al., 2012), it was expected that parental

acceptance would show a positive correlation with the Meta-Emotion Interview child

dimensions, just as the emotion coaching subscale did. Further, acceptance of emotion is

a large component of emotion coaching. Finally, it was expected that the feelings of

uncertainty/ineffectiveness in emotion socialization subscale would negatively correlate

with child dimensions of the Meta- Emotion Interview. The rationale for this hypothesis

is that parents high in uncertainty/ ineffectiveness may feel incompetent with regards to

emotion socialization and may, therefore, avoid or struggle with being involved in

experiences related to the emotional awareness, acceptance, coaching, and regulation of

their children.

Overview of hypotheses. As a general hypothesis, it was expected that the

ERPSST-L and ERPS would demonstrate construct validity by correlating in the

expected directions with the Meta-Emotion Interview subscales. This was expected

26
Table 3

Hypothesized Relations between the Emotion-Related Parenting Styles (ERPS) and the Meta-Emotion Interview

ERPS Subscales
Emotion Parental Parental Rejection Uncertainty/
Coaching Acceptance Ineffectiveness

Meta- Awareness of Child’s Positive Positive Negative Negative

Emotion Emotions Correlation Correlation Correlation Correlation
Interview
Dimensions Acceptance of Child’s Positive Positive Negative Negative
Emotions Correlation Correlation Correlation Correlation

Account of Child’s Positive Positive Negative Negative

Regulation of Emotions Correlation Correlation Correlation Correlation

Coaching of Child’s Positive Positive Negative Negative

Emotions Correlation Correlation Correlation Correlation

27
because meta-emotion theory is used to code the Meta-Emotion Interview, and the

ERPSST-L and ERPS specify emotion-related parenting styles encompassed in meta-

emotion theory. In having a small sample size, it is possible for there to be nonsignificant

correlations between some variables. This is likely to happen on scales that have low

internal consistency.

It has been found that mothers and fathers differ in parenting practices and

emotional expression (e.g., Katz, Gottman, & Hooven, 1996). Hakim-Larson et al.

(2006) found that mothers were more likely to report higher self-expressiveness and

expressive encouragement than fathers. Fathers were significantly more likely to adopt a

dismissing parenting style. Additionally, Gottman et al. (1996) found that mothers

reported greater emotional awareness and coaching than fathers. Due to having a small

sample size in the present study, the focus on the present study was on maternal meta-

emotion philosophy.

28
 CHAPTER III

METHOD

Participants

Archival data collected from the fall of 2000 to the fall of 2002 were used for the

present study (Hakim-Larson, 2000; Fostering emotional competence in preschool

children, University of Windsor internal research grant). Participants were recruited from

the Psychology Department participant pool (i.e., a group of undergraduate students who

may elect to be research participants in studies approved by a Research Ethics Board) at a

mid-size university in southwestern Ontario. Spouse pairs were excluded from the

present study, as well as participants with an inaudible Meta-Emotion Interview. The

final sample included 36 parents (33 mothers, 3 fathers). Mothers (Mage = 30.97 years,

SD = 5.83) were used in all analyses but fathers were only incorporated as additional

analyses in order to inform considerations for future studies. All mothers had a child who

was between ages 3 and 5 years old (Mage = 3.91 years, SD = .84). In the event that a

parent had more than one child within this age range, the parent was asked to report on

the oldest child that fit into the study design. Complete information on participant

demographics is in Table 4.

Procedure

Parents first completed a consent form and were asked to bring home a package

containing a background information form, two counterbalanced measures, and items for

a storytelling task not used in the present study. The two counterbalanced measures

included the Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L; Gottman

& DeClaire, 1997, modified by Hakim-Larson et al., 2006) and a questionnaire on

29
Table 4

Participant Demographics

Mothers Fathers
(n = 33) (n = 3)
Frequency (% Frequency (%
Feature Description of total) of total)
Age of Parent Mage = 30.97 Mage = 33.67
years years
SD = 5.83 SD = 6.81
Minage = 21 Minage = 26
Maxage = 45 Maxage = 39
Marital Status Common-Law 5 (15.2%) 1 (33.3%)
Married 17 (51.5%) 2 (66.7%)
Separated/Divorced/Widowed 3 (9.1%) 0 (0%)
Single, never married 8 (24.2%) 0 (0%)
Ethnicity Caucasian 25 (75.8%) 3 (100%)
Middle Eastern 1 (3%) 0 (0%)
Native/Aboriginal 2 (6.1%) 0 (0%)
Other 2 (6.1%) 0 (0%)
No response 3 (9.1%) 0 (0%)
Annual Family Less than $10,000 1 (3.0%) 0 (0%)
Income $11,000 to 20,000 9 (27.3%) 0 (0%)
$21,000 to 30,000 2 (6.1%) 1 (33.3%)
$31,000 to 40,000 0 (0%) 0 (0%)
$41,000 to 50,000 3 (9.1%) 0 (0%)
$51,000 to 60,000 2 (6.1%) 0 (0%)
$61,000 to 70,000 2 (6.1%) 0 (0%)
Greater than $70,000 12 (36.4%) 2 (66.7%)
No response 2 (6.1%) 0 (0%)
Employment Currently employed 18 (54.5%) 3 (100%)
Status Not currently employed 15 (45.5%) 0 (0%)
Birthplace Canada 30 (90.9%) 3 (100%)
Outside Canada 3 (9.1%) 0 (0%)
Highest level of Some college/university or less 13 (39.4%) 2 (66.7%)
education College/university graduate or 20 (60.6%) 1 (33.3%)
more
Use of Yes 8 (24.2%) 1 (33.3%)
counselling No 25 (75.8%) 1 (66.7%)
services for self
Age of Target 3 years 13 (39.4%) 0 (0%)
Child 4 years 10 (30.3%) 2 (66.7%)
5 years 10 (30.3%) 1 (33.3%)
Sex of Target Female 16 (48.5%) 2 (33.3%)
Child Male 17 (51.5%) 1 (66.7%)

30
reasons for reading that was not used in the present study. Parents returned the materials

to the researchers or a researcher picked up the materials from the family’s home.

Next, participants were invited back to the university to complete the Meta-

Emotion Interview (Katz & Gottman, 1986). Interviewers included five members of the

University of Windsor’s Emotional Competence Research Group who were trained in

administering the Meta-Emotion Interview. The duration of this interview varied from

one to two hours, and participants were offered $10 in compensation for participating in

the entire study. Those parents who were students at the university received bonus points

for an eligible undergraduate course. Finally, parents received a debriefing form.

Measures

This section begins with a description of the Meta-Emotion Interview, the long

form, and the short form measures used in the present study. Scoring procedures and the

psychometric properties from past studies are discussed for each measure. The inter-rater

reliability, mean scores and standard deviations, and internal consistency reliability for all

measures are described.

Meta-Emotion Interview (Katz & Gottman, 1986). This is a semi-structured

interview completed by a parent. There are seven dimensions of meta-emotion evaluated

by the interview. The first three dimensions pertain to the parent: awareness of emotions,

acceptance of emotions, and regulation of emotions. The remaining four dimensions

pertain to the child: parent’s awareness of child’s emotions, parent’s acceptance of

child’s emotions, parent’s coaching of child’s emotions, and parent’s account of child’s

emotional regulation. All seven dimensions are further described below. Table 5

contains the range of possible scores for Meta-Emotion Interview dimensions.

31
Table 5

Scoring of Meta-Emotion Interview Dimensions

Name of Number of Items Per Range of Possible Range of Possible

Dimension Emotion Scores Per Emotion Combined Scores

(Sadness and Anger)

Awareness of 9 9 to 36 18 to 72

child’s emotions

Acceptance of 13 13 to 65 26 to 130

child’s emotions

Coaching of 11 11 to 55 22 to 110

child’s emotions

Regulation of 9 9 to 45 18 to 90

child’s emotions

Note. Most dimension items are scored on a scale from 1 to 5 (1 = strongly disagree, 2 =

disagree, 3 = neutral, 4 = agree, 5 = strongly agree, and DK = don’t know). Some items

could only be rated on a scale from 1 to 4, while others could be rated on a scale from 1

to 5.

32
 Dimensions of the Meta-Emotion Interview that pertain to the parent. Three of

the seven dimensions of the Meta-Emotion Interview pertain to the parent. They are

parental awareness, acceptance, and regulation of his or her own emotions.

Parent’s awareness of his/her own emotions. There are 12 items on this

dimension to examine the degree to which a parent is aware of his or her own emotional

processes (e.g., ―Parent is descriptive of their experience of this emotion‖). This is an

important dimension in that Gottman and colleagues (1996) stated, ―we found that only

people who are aware of emotion and can differentially talk about the nuances of emotion

and emotion intensity find emotional expression to be acceptable‖ (p. 267). Those who

are high in this dimension consciously experience emotions, while those who are low in

this dimension might prefer to avoid noticing negative emotions. Such parents might see

the ―passage of time‖ as a means of resolving issues of sadness or anger (p. 267).

Parent’s acceptance of his/her own emotions. This dimension assesses a parent’s

attitudes toward emotion regarding their own level of comfort accepting emotions. There

are 17 items on this dimension (e.g., ―Parent feels comfortable with their expression of

this emotion‖).

Parent’s regulation of his/her own emotion. This dimension evaluates a parent’s

ability to control negative emotions. This can be indicated by a parent’s use of

remediation techniques. There are 12 items on this dimension. (e.g., ―This emotion is

difficult to get over‖). Difficulty in emotion regulation has been associated with physical

and mental health problems and difficulties with marital relations (Gottman, Katz, &

Hooven, 1997).

33
 Parent’s responses to child’s emotions. There are four dimensions of the Meta-

Emotion Interview that pertain to the child. They are parent’s awareness, acceptance, and

coaching of the child’s emotions, and parent’s perception of the child’s ability to regulate

emotions.

Parent’s awareness of child’s emotions. There are 9 items on this dimension

(e.g., ―Parent knows cause of child’s emotion‖). This dimension evaluates a parent’s

ability to observe and decode the emotions of the child.

Acceptance of child’s emotions. This dimension contains 13 items (e.g., ―Parent

wants child to know it’s OK to have this feeling‖). It measures a parent’s responses (both

direct and indirect) to his/her child’s emotional expression. For instance, responses can

be physically soothing, verbal, or even analytical.

Parent coaching of child’s emotions. This dimension contains 11 items that tap

into a parent’s ability to show respect for and comfort the child during emotional

experiences, as well as teach age-appropriate strategies for the child to soothe his or her

own emotion (e.g., ―When child is upset, parent talks about situation, emotion‖).

Additionally, these parents aim to teach their children about the world of emotions. The

ability to soothe oneself physiologically is crucial in one’s ability to develop empathy

(Gottman et al., 1996).

Parent’s account of child’s regulation of emotion. From this dimension, a

researcher can evaluate a parent’s ability to recognize his or her child’s ability to get over

the emotion. To score high on this dimension, a parent may recognize that his or her

child can self-regulate the emotion and/or identify remediation strategies that are

34
effective for the child. There are 9 items on this dimension (e.g., ―This emotion is

difficult for the child to get over‖).

Psychometric properties of the Meta-Emotion Interview. Convergent validity of

the Meta-Emotion Interview has been found to be adequate. For example, emotion

coaching parents are more likely to use scaffolding and praising but less derogation than

parents of other styles (Gottman, Katz, & Hooven, 1997). Internal consistency, as

measured by Cronbach’s alpha, has been found to be moderate (Katz & Windecker-

Nelson, 2004). Katz and Windecker-Nelson (2004) found the inter-rater reliability to

range from r = .57 to r = .82 among dimensions, while Cunningham and colleagues

(2009) obtained an overall inter-rater reliability of r = .72.

Transcription of the Meta-Emotion Interviews. Interviews were transcribed

verbatim into Microsoft Word, as accurately as possible, by a team of research assistants.

With the exception of one study (Cunningham et al., 2009), the transcribing of the Meta-

Emotion Interview prior to coding has rarely been reported in the literature. Typically,

other researchers only use the audio-recording of the interview when coding (e.g., Katz et

al., 2008). Transcription was beneficial because the transcripts included line numbers

that were used to document where codes occurred. Line numbers were useful in

resolving coding discrepancies. When interviews were transcribed in the present study,

no personal information (e.g., real names or identifying information about the

participants) were included in the transcript. To confidentially document the dialogue

between the interviewer and the interviewee, general descriptions were used instead of

names [e.g.,―(name of son’s teacher‖)].

35
 Coding of the Meta-Emotion Interviews. After receiving ethics clearance to use

archival data, the author completed the Meta-Emotion Coding System Coding Training

by Katz, Mittman, and Embry (n.d.). The Meta-Emotion Coding System Coding

Training Manual (Katz, Mittman, & Embry, n.d.) was used to assist in making coding

decisions, and coding sheets were used to record scores. The coding system contains a

booklet of the items of all of the Meta-Emotion Interview dimensions, descriptions of

these dimensions, and possible codes for each item. Coding training involved listening to

and coding sample Meta-Emotion Interviews that were audio-recorded on cassettes.

Interview transcripts were unavailable for coding training. The duration of training was

approximately 20 hours, which is consistent in the literature (e.g., Cunningham et al.,

2009).

When coding the interviews from the present study, the coder first listened to the

audio-taped interview in order to code items related to the parent’s tone of voice (e.g., in

terms of interest in the questions being asked, hesitation, and uncertainty). Next, the

coder coded each interview by listening to the tape and following along with the

transcript. The Meta-Emotion Interview was scored on sadness, anger, and the

combination scores of sadness and anger. Interview questions related to fear were

excluded because the ERPSST-L only contains items related to sadness and anger.

During coding of the Meta-Emotion Interviews, it is possible that a code may not

be applicable on an item (e.g., if the interviewer never asks the corresponding question).

In this case, a DK (don’t know) code would be provided. It is important to note that for

the Meta-Emotion Interview, scoring instructions require that ―don’t know‖ (DK)

responses are given the average score for the dimension, and the total score is adjusted.

36
For example, if the total score for Awareness of Child’s Emotions is 22 but the parent

responded with DK for 2 of the 9 items, the score is calculated as follows: 22 + (22/7)(2)

= 28.28. It is important to note that even though dimensions may have the same number

of items it is possible for the range of scores to differ due to the computation of DK

responses. For example, the awareness of child’s emotions dimension has 9 items but the

scale only ranges from 1 to 4, making the maximum score 36. In contrast, the regulation

of child’s emotions dimension has 9 items but the scale can range from 1 to 5, making the

maximum score 45.

The coder was blind to information about the parents, their children, and their

scores on other measures. However, the coder was aware of participant gender based on

the context of the interview and the audio recorded voices of the participants.

Inter-rater reliability for the Meta-Emotion Interview. In both the coding

training and in the coding of the Meta-Emotion Interviews from the present study, inter-

rater reliability was calculated. Gottman and colleagues (1996) and a research assistant

in Katz’s lab (A. Maliken, personal communication, July 11, 2012) recommended using

Pearson’s correlations to calculate inter-rater reliability. However, the Intraclass

Correlation Coefficient has an advantage over Pearson’s correlations in that it accounts

for both rating differences and the correlation between raters (McGraw & Wong, 1996).

A two-way random Intraclass Correlation Coefficient was selected (i.e., both rater effects

and item effects are random) in order to compare ratings between the primary researcher

and each of the other raters on each dimension score. An Intraclass Correlation

Coefficient ranges from 0.00 to 1.00, and an Intraclass Correlation Coefficient of .70 or

higher is considered to be adequate.

37
 Inter-rater reliability: Meta-Emotion Interview coding training. The primary

researcher coded fifteen Meta-Emotion Interviews from the Coding Training System

(Katz, Mittman, & Embry, n.d.). To calculate inter-rater reliability for the coding

training, dimension scores computed by the primary coder were first compared to scores

in the training manual (Katz, Mittman, & Embry, n.d.). The primary coder’s scores also

were compared to training scores computed by two independent coders who were trained

in the same procedure. The first author of the Coding Training System stated that

reliability tapes should be completed until coders are confident they understand the

dimensions being coded (L. F. Katz, personal communication, June 13, 2011). Results of

Intraclass Correlation Coefficients for training interviews (Appendix A) indicate that the

inter-rater reliability between the primary researcher and the developers of the Meta-

Emotion Interview was .90 on average, ranging from .68 to .97. Between the primary

coder and these two other independent coders, the Intraclass Correlation Coefficient on

the training interviews was .71 on average, ranging from .34 to .97.

Inter-rater reliability: Meta-Emotion Interview coding of data from the present

study. With an adequate inter-rater reliability established in coding training (i.e.,

Intraclass Correlation Coefficient exceeding the .70 minimum on the training data), the

coding of the Meta-Emotion Interviews from the present study began. Interviews with

mothers (n = 33) and fathers (n = 3) were coded. To calculate inter-rater reliability for

the present study, scores computed by the primary coder were compared to scores

computed previously by two other trained, independent raters. In the present study, there

were five batches of seven to eight randomly-ordered interviews, and two interviews

from each batch were tested for inter-rater reliability. Inter-rater reliability, calculated on

38
30.30% of the sample, was found to be reliable (average r = .80, with a range of .13 to

1.0). The poor inter-rater reliability of .13 was on the dimension for parent awareness of

sadness. Refer to Appendix B for complete findings regarding inter-rater reliability

analyses for study data.

In the case of a discrepancy between two coders’ dimension scores, item codes

were compared in order to reach a final decision. For subsequent analyses, dimension

scores from the primary coder were used, as per a recommendation from a research

assistant from Dr. Katz’s lab (A. Maliken, personal communication, July 11, 2012).

Also, the primary coder in the present study had the highest inter-rater reliability with the

Gottman lab.

Means and standard deviations for the Meta-Emotion Interview. Findings from

the present study are reported in Table 6. In order to allow for the recognition of

distinctive profiles, DeOliveira and colleagues (2005) recommended examining parent

dimensions and child dimensions separately; additionally, they recommended that the

dimensions be examined differentially by emotion type. Thus, in the present study the

child and parent dimensions were analyzed separately. Emotions of sadness and anger

were examined separately; they also were examined together, referred to as ―combined,‖

in order to stay consistent with the literature. The means and standard deviations of the

Meta-Emotion Interview dimensions were not available from other research publications

because summary scores have often been used. These summary scores represent

aggregate variables, combining parent dimension scores with their corresponding child

dimension scores (e.g., an overall awareness score is created by adding scores on parents’

awareness of their own emotions with parent’ awareness of their children’s emotions).

39
Table 6

Means and Standard Deviations for Dimensions on the Meta-Emotion Interview

Name of Measure Subscale Mean (SD)

Parent Awareness
Sadness 44.15 (3.86)
Anger 44.92 (2.86)
Combined 89.07 (5.93)
Parent Acceptance
Sadness 59.92 (5.88)
Anger 57.24 (6.40)
Combined 117.16 (9.49)
Parent Regulation
Sadness 42.05 (4.07)
Anger 40.92 (5.75)
Combined 82.97 (7.78)
Child Awareness
Sadness 34.51 (1.75)
Anger 34.19 (2.15)
Combined 68.70 (3.44)
Child Acceptance
Sadness 46.81 (3.83)
Anger 41.63 (5.49)
Combined 88.44 (8.08)
Child Regulation
Sadness 32.79 (3.82)
Anger 27.68 (5.08)
Combined 60.47 (7.72)
Child Coaching
Sadness 42.34 (2.39)
Anger 38.03 (4.60)
Combined 80.37 (6.07)
Note. ―Combined‖ refers to the summed scores of sadness and anger for that dimension.

40
Thus, dimension scores in the present study could not be compared to dimension scores

from other research findings.

Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L; Gottman &

DeClaire, 1997, modified by Hakim-Larson et al., 2006). This measure contains 81 items

related to parental meta-emotion about sadness and anger, measured on a 5-point Likert-

type scale (1 = always false, 2 = mostly false, 3 = somewhat true/false, 4 = mostly true,

and 5 = always true).

Each statement on the ERPSST-L represents one of four parenting styles: emotion

coaching (23 items), laissez-faire (10 items), dismissing (25 items), and disapproving (23

items), and higher scores represent greater endorsement of that parenting style. Table 7

depicts the range of possible scores for the ERPSST-L.

Means, standard deviations, and internal consistency reliability for the

ERPSST-L. Means, standard deviations, and internal consistency reliability also were

calculated for the ERPSST-L (refer to Table 8). Each mother’s average Likert-type score

for each subscale was used in later analyses. To calculate an average Likert-type score,

the item responses for each individual scale were summed and then divided by the

number of items for that scale.

In a study by Hakim-Larson et al., (2006), 91/100 parents were classified as

having a predominant style of emotion coaching using the ERPSST-T/F version in their

analysis. The scale with the highest average score for each participant was designated

their primary emotion-related parenting style. Based on this method of classification,

81.8% (n = 27) of the mothers in the current study sample using the ERPSST-L were

classified as predominately emotion coaching, 15.2% (n = 5) were classified as

41
Table 7

Scoring of Subscales from the Emotion-Related Parenting Styles Self-Test – Likert

(ERPSST-L)

Name of Subscale Number of Range of Possible Raw Range of Possible

Items Scores Scores Per Item

Emotion Coaching 23 23 to 115 1 to 5

Laissez-Faire 10 10 to 50 1 to 5

Dismissing 25 25 to 125 1 to 5

Disapproving 23 23 to 115 1 to 5

Note. On the ERPSST-L, parents rated their level of agreement on a scale ranging from 1

(always false), 2 (mostly false), 3 (somewhat true/false), 4 (mostly true), and 5 (always

true).

42
Table 8

Means, Standard Deviations, and Cronbach’s Alpha Coefficients for the Emotion-Related

Parenting Styles Self-Test – Likert (ERPSST-L)

Name of ERPSST- Number of Average Likert- Subscale Mean Alpha

L Subscale Items type score (SD) Coefficient

Emotion coaching 23 3.83 88.08 (10.91) .89

Laissez-faire 10 3.36 33.64 (3.81) .59

Dismissing 25 2.55. 63.65 (9.26) .79

Disapproving 23 2.10 48.34 (12.02) .90

Note. Each mother’s average Likert-type score for each subscale was used in later

analyses.

43
predominately laissez-faire, and 3% (n = 1), had an equal score on emotion coaching and

laissez-faire. Thus, no mother in this sample reported a predominately dismissing or

disapproving style, although some mothers had scores on these scales that were higher

relative to that of others in the sample.

Converting scores on the ERPSST-L to scores on the Emotion-Related

Parenting Styles (ERPS). Though participants did not directly complete the ERPS,

subscale scores may be computed by extracting corresponding scores from the 20 items

on the ERPSST-L that are contained in the ERPS. To calculate an emotion-coaching

subscale score on the ERPS, scores on items 34, 75, 29, 35, and 64 from the ERPSST-L

(corresponds to items 3, 6, 8, 15, and 19 on the ERPS, respectively) were summed. To

calculate a parental rejection of negative emotion subscale score, scores on items 3, 11,

14, 41, and 66 (corresponds to items 1, 4, 10, 11, and 14 on the ERPS, respectively) were

summed. To calculate a parental acceptance of negative emotion subscale score, scores

on items 38, 31, 73, 72, and 39 of the ERRPSST-L were summed. This corresponds to

items 2, 5, 9, 12, and 16 on the ERPS, respectively. Finally, a score on the feelings of

uncertainty/ ineffectiveness in emotion socialization subscale score was calculated by

summing scores on items 53, 77, 48, 76, and 78 of the ERPSST-L (corresponds to items

7, 13, 17, 18, and 20 on the ERPS, respectively). As explained in Table 9, each ERPS

subscale score can range from 5 (low endorsement of that parenting style) to 25 (high

endorsement of that parenting style). Average scores on each subscale were calculated

and used in analyses. On both the raw scores and average subscale scores, a higher score

indicated greater endorsement of that emotion-related parenting style.

44
Table 9

Scoring of Subscales from the Emotion Related Parenting Styles (ERPS)

Name of ERPS Subscale Number of Range of Possible Range of Possible

Items Raw Scores Scores Per Item

Emotion Coaching 5 5 to 25 1 to 5

Parental Rejection 5 5 to 25 1 to 5

Parental Acceptance 5 5 to 25 1 to 5

Uncertainty/Ineffectiveness 5 5 to 25 1 to 5

Note. On the ERPS, parents rated their level of agreement on a scale ranging from 1

(always false) to 5 (always true).

45
 Means, standard deviations, and internal consistency reliability for the ERPS.

Finally, means, standard deviations, and internal consistency reliability were also

calculated on the ERPS. Refer to Table 10 for this information. Each mother’s average

Likert-type score for each subscale was used in later analyses. To calculate an average

Likert-type score, the item responses for each individual scale were summed and then

divided by the number of items for that scale. Mothers were classified according to

ERPS scale with the highest score; this represented their predominant emotion-related

parenting style. When examined using the ERPS, 78.8% (n = 26) of the mothers were

primarily emotion coaching, 15.2% (n = 5) were primarily in the parental acceptance of

negative emotion group, and 6.1% (n = 2) were equally emotion coaching and accepting.

46
Table 10

Means, Standard Deviations, and Cronbach’s Alpha Coefficients for the Emotion-Related

Parenting Styles (ERPS)

Name of ERPSST-L Number Average Subscale Alpha

Subscale of Items Likert-type Mean (SD) Coefficient

score

Emotion coaching 5 4.20 21.02 (2.50) .75

Parental rejection 5 2.23 11.14 (3.36) .78

Parental acceptance 5 3.66 18.28 (3.80) .81

Uncertainty/ineffectiveness 5 2.20 10.98 (2.68) .67

Note. Each mother’s average Likert-type score for each subscale was used in later

analyses.

47
 CHAPTER IV

RESULTS

Overview of Results

Analyses are divided into five main sections. The first section includes data

screening procedures related to missing data and statistical outliers. Next are the

preliminary analyses, including the assessment of attrition in the sample, testing for

assumptions to be used in the main analyses, and the identification of control variables.

The third section for main analyses consists of testing the construct validity of the

ERPSST-L long form, and testing the construct validity of the ERPS via zero-order

correlations and partial correlations. The fourth section includes additional analyses,

such as the comparing high emotion coaching and low emotion coaching groups,

examining the relation between Meta-Emotion Interview parent dimensions and child

dimensions, examining the role of maternal experience, and comparing results by

emotion (sadness and anger). The final section ends with examples from mothers’ and

fathers’ Meta-Emotion Interviews. Only the final section contains data pertaining to

fathers.

Data Screening

Missing data. Prior to the main analyses, variables were examined in order to

identify missing data. Missing data were not found in items for the Meta-Emotion

Interview dimensions. However, missing data points on ERPSST-L items were identified

on five cases, also creating missing data points on the ERPS. When data points are

missing at random, as was the case in the present data set, one option is to use a mean

substitution for the missing values (Tabachnick & Fidell, 2012). Mean substitutions were

48
used for each missing data point, as overall assumptions and testing outcomes were not

impacted.

Outliers. Data were then examined for outliers, using a z-score of 3.29 as a

cutoff, as recommended by Field (2009). One outlier (z = -3.89) was identified for the

awareness of child sadness dimension, and another outlier (z = -3.49) was identified for

the awareness of child (combined sadness and anger) dimension. Both outliers were from

the same case. With outliers included in the data set, kurtosis was elevated on these

dimensions (z = 3.81 and z = 4.23, respectively) and skewness was within the acceptance

range. Upon further inspection, it was determined these outliers were sampled from the

target population and the case’s other scores did not indicate a pattern of a response set.

In order to reduce the impact of this variable, a score change was implemented, as

recommended by Tabachnick and Fidell (2012) and Field (2009). To implement the

score change for each variable, the two outliers were substituted with a raw score that

was one unit smaller than the lowest score on that variable. In doing so, the impact of

this outlier was reduced.

Preliminary Analyses

Attrition. Mitchell (1985) emphasized the importance of comparing

nonrespondents and respondents. If a special characteristic was related to the

respondents but not the nonrespondents, or vice versa, it may indicate that the sample is

not representative or that a confounding variable might be present. Thus, attrition was

assessed in the present study because 49 parents completed the ERPSST-L and

questionnaire package but 10 did not return for the Meta-Emotion Interview (note that

two Meta-Emotion Interview tape were inaudible and spouse pairs were removed,

49
reducing the sample size to 36 mothers and fathers). Thus, ―respondents‖ refers to

participants who completed the study questionnaires and returned to complete the Meta-

Emotion Interview. ―Non-respondents‖ refers to participants who completed the study

questionnaires but not the Meta-Emotion Interview. The presence of statistically

significant differences between these groups may warrant looking further into

confounding variables.

A one-way Analysis of Variance (ANOVA) was conducted to determine if

respondents differed from nonrespondents in terms of demographics (i.e., parent age,

parent sex, child age, number of children, marital status, and race) and scores on the

ERPSST-L subscales. In order to use this test, assumptions of homogeneity of variance,

normality, interval-level data, and independence of observations were met (Field, 2009,

p. 113). Findings, as shown in Table 11, indicate that no statistically significant

difference was found between respondents and nonrespondents in terms of demographics

and scores on the ERPSST-L (all ps > .05).

Assumptions. After using score substitution to resolve the problem with two

outliers, skewness and kurtosis values were converted into z-scores by dividing each

skewness and kurtosis value by its respective standard error, as recommended by Field

(2009). For small to moderate samples, an alpha level of .001 (i.e., an absolute value of

3.29) can be used to indicate a normal distribution (Tabachnick & Fidell, 2012).

Skewness and kurtosis values all met the assumption of normality with the exception of

the Meta-Emotion Interview parent awareness of sadness dimension (skewness z = -

3.85). Across all other variables, the range of skewness was from z = -2.93 to z = 1.80;

the range of kurtosis was from z = -1.57 to z = 2.64. Frequency histograms were visually

50
Table 11

One-Way Analysis of Variance (ANOVA) Comparing Respondents and Non-Respondents

on Demographic Variables and Emotion-Related Parenting Styles Self-Test – Likert
(ERPSST-L) Scores

Respondents Non- ANOVA Result η2

(n = 39) Respondents
M(SD) (n = 10)
M(SD)
Parent Age 31.18(5.67) 29.8(3.52) F(1, 47) = .54, p = .468 .01
Parent Sex 1.87(.34) 1.70(.48) F(1, 47) = 1.71, p = .197 .04
Child Age 4.00(.83) 4.30(.82) F(1, 47) = 1.05, p = .311 .02
Child Sex 1.54(.51) 1.50(.53) F(1, 47) =.05, p = .832 .01
n Children 2.02(.90) 1.80(.63) F(1, 47) =.55, p = .462 .01
Marital Status 2.13(1.78) 1.70(1.25) F(1, 47) =.51, p = .479 .01
Race or ethnicity 2.18(2.36) 1.80(2.53) F(1, 47) = .20, p = .657 .01
ERPSST-L Scores
Emotion Coaching 3.79(.48) 3.92(.36) F(1, 47) =.62, p = .435 .01
Laissez-faire 3.33(.37) 3.28(.39) F(1, 47) =.17, p = .686 .04
Dismissing 2.55(.35) 2.37(.33) F(1, 47) = 2.17, p = .147 .01
Disapproving 2.12(.49) 2.10(.34) F(1, 47) =.02, p = .892 .01
Note. ―Respondents‖ refers to participants who completed the study questionnaires and
returned to complete the Meta-Emotion Interview. ―Non-respondents‖ refers to
participants who completed the study questionnaires but not the Meta-Emotion Interview.

51
inspected for normality. In general, histograms represented the normal bell curve;

however, the histogram for Meta-Emotion Interview parent awareness of sadness was

negatively skewed. On nearly all dimensions for the Meta-Emotion Interviews, skewness

values were negative, indicating that data contained scores in the higher range. In dealing

with a well-educated sample, high scores on dimensions related to meta-emotion

philosophy, such as parent awareness of sadness, were anticipated. One option to repair

the skewness of the parent awareness of sadness variable was to use a data

transformation. However, data transformation may make interpretation difficult in other

planned analyses (Tabachnick & Fidell, 2012), particularly so when only transforming

one variable. Analyses proceeded without using a data transformation but with the

acknowledgement that the assumption of normality was violated on that variable.

Identification of control variables. Zero-order correlational analyses were first

used to test if hypothesized control variables (i.e., child age, child gender, and annual

family income) relate to scores on the Meta-Emotion Interview, Emotion-Related

Parenting Styles Self-Test – Likert (ERPSST-L), and Emotion-Related Parenting Styles

(ERPS). Though child age and gender were not significantly related to ERPSST-L scores

in a study by Hakim-Larson et al. (2006), they were checked in the present sample. In

order to preserve the sample size for correlational analyses, cases were excluded on a

pairwise (analysis-by-analysis) basis. Complete findings are located in Table 12, and

primary findings are now discussed for each control variable.

52
Table 12

Zero-Order Correlations between Child Age, Child Sex, Family Income, and Scores on
the Meta-Emotion Interview, Emotion-Related Parenting Styles Self-Test – Likert
(ERPSST-L), and Emotion-Related Parenting Styles (ERPS).

Potential Control Variable

Child Age Child Sex Family Income†

Meta- Parent Awareness

Emotion Sadness .17 .11 -.15
Interview Anger .40* .16 .25
Variables Combined .30 .15 .02
Parent Acceptance
Sadness .03 -.21 .42*
Anger .18 .05 .13
Combined .14 -.10 .35
Parent Regulation
Sadness -.22 .12 .30*
Anger -.25 .04 .06
Combined -.30 .10 .20
Child Awareness
Sadness .08 -.22 .16
Anger .20 -.12 .32
Combined .16 -.20 .27
Child Acceptance
Sadness .05 .12 -.13
Anger .08 -.01 .09
Combined .08 .05 .01
Child Regulation
Sadness -.07 .25 .04
Anger -.10 .24 .04
Combined -.10 .28 .05
Child Coaching
Sadness .15 .10 -.25
Anger .18 .08 .22
Combined .20 .10 .06
ERPSST-L Emotion Coaching .34 .15 -.15
Variables Laissez Faire -.12 .16 -.53**
Dismissing -.05 .21 -.20
Disapproving .08 -.15 -.24
ERPS Emotion Coaching .31 .09 .01
Variables Parental Rejection .12 -.02 -.04
Parental Acceptance .26 .19 -.20
Uncertainty/Ineffectiveness -.08 .03 -.11
Note. † = A sample size of 31 was used for that variable. * p < .05, ** p < .01.

53
 Child age. Results of two-way correlational analyses indicated that child age was

positively related to maternal awareness of her own anger, r(31) = .40, p = .021 and was

trending towards significance with regards to the total emotion coaching score from the

ERPSST-L, r(31) = .34, p = .055.

Child sex. Independent sample t-tests were used to determine if mean scores on

main variables differed by child sex. On average, daughters were rated higher in terms of

their ability to regulate sadness and anger (M = 62.63, SD = 5.69) as compared to sons (M

= 58.44, SD = 8.93). This difference was not statistically significant, t(31) = 1.62, p =

.121; however, it nearly represented a medium-sized effect, r(31) = .28.

Family income. When data were collected, annual family income was reported

categorically by income brackets. Examination of frequencies for annual family income

indicated that income bracket could be split at $51, 000. Thus, a high income group (n =

16) and a low income group (n = 15) were formed based on this median split.

Independent samples t-tests were used to determine if mean scores on main

variables differed by family income bracket. On average, mothers of a higher income

bracket were better able to accept their own sadness (M = 61.07, SD = 5.08) than mothers

of a lower income bracket (M = 56.30, SD = 6.16), t(29) = -2.36, p = .025, effect size of

r(29) = .42. Mothers of a higher income bracket reported being better able to regulate

their own sadness (M = 43.70, SD = 2.89) than mothers of a lower income bracket (M =

40.35, SD = 4.59), t(29) = -2.45, p = .021, effect size of r(29) = .30. On average, mothers

of a higher income reported lower scores on the ERPSST-L laissez-faire scale (M =

31.74, SD = 2.65) than mothers of a lower income (M = 35.49, SD = 4.12), t(29) = 3.03, p

= .005, effect size of r(29) = -.53.

54
 In summary, child age, child sex, and family income correlated with some of the

main variables in the present study. Those three variables, therefore, were controlled for

in subsequent partial correlation analyses.

Main Analyses

The purpose of the present study was to test the construct validity of the long

form and short form questionnaires by comparing subscale scores to those of the

interviews. Meta-emotion theory was used to determine expected direction of correlation

between variables. Correlational analyses are considered appropriate for continuous data

that are normally distributed. In 1955, correlation matrices were identified by Cronbach

and Meehl as appropriate for testing construct validity. They stated, ―If two tests are

presumed to measure the same construct, a correlation between them is predicted‖

(Cronbach & Meehl, 1955). The use of correlations to test the construct validity of a

measure has been used in the literature (e.g., Paterson et al., 2012). It is highly

recommended that researchers report effect sizes and confidence intervals in addition to

the correlation coefficients in testing for construct validity (e.g., Thompson et al., 2005).

Construct validity of the ERPSST-L long form. The first purpose of the

present study was to test the construct validity of the ERPSST-L long form. Construct

validity would be demonstrated if scores on the long form correlate with scores on the

Meta-Emotion Interview in the expected directions. It was hypothesized that the emotion

coaching scale would positively correlate with all four child dimensions (i.e., awareness

of child’s emotions, acceptance of child’s emotions, account of child’s regulation of

emotions, and coaching of child’s emotions). It was expected that the laissez-faire scale

would positively correlate with awareness and acceptance, but would correlate negatively

55
with regulation and coaching. Finally, it was anticipated that the dismissing and

disapproving scales would negatively correlate with the child dimensions.

To compare the Meta-Emotion Interview child dimensions to the scales on the

ERPSST-L, zero-order correlations were conducted as shown in Table 13. Greater use of

a dismissing style was related to lower ratings on child dimensions related to the

acceptance of emotions; the coaching of child anger, r(31) = -.44, p = .005; and the

regulation of child anger, r(31) = -.30, p = .048. Greater use of a disapproving style was

negatively related to the acceptance of child anger, r(31) = .44, p = .005 and the

regulation of child anger, r(31) = -.41, p = .009. Statistically significant relations were

not found between the Meta-Emotion Interview dimensions and the ERPSST-L emotion

coaching and laissez-faire emotion-related parenting styles.

As previously explained, zero-order correlations revealed that some scores on the

ERPSST-L correlated with child age, child sex, and family. The removal of the effect of

these three variables was abbreviated and referred to as a subscripted x. Thus, one-tailed

partial correlations were computed to examine the ERPSST-L scores in relation to scores

on the Meta-Emotion Interview while controlling for child age, child sex, and family

income (refer to Table 14). These analyses were one-tailed because a specific direction

was anticipated for each analysis based on literature. A statistically significant relation

between emotion coaching and the child’s regulation of anger (r-a) was found, rEC,r-a.x(26)

= .35, p = .035. In contrast, mothers who scored high on the dismissing (DI) scale tended

to rate low on child regulation of anger, rDI,r-a.x(26) = -.37, p = .026. Similarly, mothers

who scored high on the disapproving (DA) scale tended to have their children rate low on

their ability to regulate anger, rDA,r-a.x(26) = -.37, p = .028. When statistically significant,

56
Table 13

Zero-Order Correlations between Meta-Emotion Interview Child Dimensions and Emotion-

Related Parenting Styles Self-Test – Likert (ERPSST-L)

ERPSST-L Subscales (Average)

Child Emotion Laissez-faire Dismissing Disapproving
Dimensions coaching
Awareness
Sadness .14 -.03 -.02 .07
Anger .22 -.20 -.27 -.17
Combined .21 -.14 -.18 -.07
Acceptance
Sadness .07 -.05 -.33* -.17
Anger .15 -.08 -.53*** -.44**
Combined .14 -.08 -.52*** -.39*
Regulation
Sadness -.02 -.14 .01 .06
Anger .27 .07 -.30* -.41**
Combined .17 -.02 -.19 -.24
Coaching
Sadness .29 .13 -.27 -.15
Anger .01 -.18 -.44** -.23
Combined .11 -.09 -.44** -.23
Note. Pairwise deletion; one-tailed. * p < .05, **p < .01, ***p < .001.

57
Table 14

One-Tailed Partial Correlations between Meta-Emotion Interview Child Dimensions and

Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L) Subscales Controlling
for Child Age, Child Sex, and Family Income

ERPSST-L Subscales (Average)

Child Emotion Laissez-faire Dismissing Disapproving
Dimensions coaching
Awareness
Sadness .19 .05 .04 .05
Anger .26 -.01 -.20 -.15
Combined .26 .02 -.10 -.06
Acceptance
Sadness .02 -.15 -.40* -.22
Anger .16 -.03 -.53** -.46**
Combined .12 -.09 -.55*** -.42*
Regulation
Sadness .01 -.18 -.04 .15
Anger .35* .07 -.37* -.37*
Combined .23 -.04 -.27 -.17
Coaching
Sadness .20 -.01 -.37* -.26
Anger -.03 -.10 -.45** -.20
Combined .05 -.08 -.47** -.25
Note. ―Combined‖ refers to the summed scores of sadness and anger for that dimension.
* p < .05, **p < .01, ***p < .001.

58
the dismissing and disapproving scales of the ERPSST-L negatively correlated with the

Meta-Emotion Interview child dimensions in the expected direction. Additionally,

mothers who scored high on the dismissing scale tended to score low on coaching of

child sadness (c-s), rDI,c-s.x(26) = -.37, p = .028; low on coaching of child anger, rDI,c-a.x(26)

= -.45, p = .009; and low on the coaching of both sadness and anger, rDI,c-sa.x (26) =-.47, p

= .006. Higher ratings on the Meta-Emotion Interview acceptance dimensions were

related to lower scores on the dismissing scales of the ERPSST-L. No significant

correlations found between emotion coaching (EC) and acceptance of child sadness (ac-s),

rEC,ac-s.x(26) = .02, p = .460; child regulation of sadness (r-s), rEC,r-s.x(26) = .01, p = .491;

and the coaching of child anger (c-a), rEC,c-a.x(26) = -.03, p = .447. No significant

correlation was found between the laissez-faire scale of the ERPSST-L and the Meta-

Emotion Interview dimensions. Though there were some nonsignificant correlations, all

statistically significant correlations in the data were in the anticipated direction.

Construct validity of the ERPS short form. It was expected that both the ERPS

emotion coaching scale and the parental acceptance of negative emotion scale would

positively correlate with ratings on all child dimensions of the Meta-Emotion Interview

(i.e., awareness of child’s emotions, acceptance of child’s emotions, account of child’s

regulation of emotions, and coaching of child’s emotions). Next, it was expected that the

parental rejection of negative emotion scale and the uncertainty/ineffectiveness scales

would negatively correlate with ratings on the Meta-Emotion Interview child dimensions.

The ERPS would demonstrate construct validity by correlating, in the expected

directions, with ratings on the Meta-Emotion Interview subscales.

59
 Zero-order correlations (Table 15) revealed that some scores on the ERPS

correlated with scores on the Meta-Emotion Interview. For example, greater parental

rejection of negative emotion was associated with lower scores on emotion coaching of

child anger, r(31) = -.30, p = .047. On the other hand, greater parental acceptance of

negative emotion was related to greater coaching of child sadness, r(31) = .32, p = .036.

One-tailed partial correlations were computed to examine the ERPS scores in

relation to scores on the Meta-Emotion Interview while controlling for child age, child

sex, and family income (refer to Table 16). As done previously, the removal of the effect

of child age, child sex, and family income is represented by a subscripted x. Hypotheses

were partially supported. It was found that mothers who reported greater use of emotion

coaching (ec), as measured by the ERPS, tended to be rated high in child regulation of

combined sadness and anger (r-sa), rec,r-sa.x(26)= .35, p = .035. Mothers who scored high in

parental rejection (pr) tended to score low in the coaching of child anger, rpr,c-a.x(26) =

-.32, p = .050. Additionally, mothers who scored high on parental acceptance (pa) tended

to score high in awareness of child anger, rpa,aw-a.x(26) = .36, p = .032. Mothers who

scored high in uncertainty/ineffectiveness (ui) reported that their children had difficulty

regulating anger, rui,r-a.x(26) = -.36, p = .028. Further, there was a marginal negative

relation between uncertainty/ineffectiveness and child regulation of combined sadness

and anger, rui,r-sa.x(26)= -.30, p = .058. In contrast to this, both mothers who were high in

emotion coaching and high in acceptance of their children’s emotions were rated higher

scores on their children’s ability to regulate anger ,rec,r-a.x(26) = .34, p = .037 and rpa,r-

a.x(26) = .33, p = .042, respectively. Between the Meta-Emotion Interview and the ERPS,

all statistically significant correlations were in the expected directions.

60
Table 15

Zero-Order Correlations between Meta-Emotion Interview Child Dimensions and

Emotion-Related Parenting Styles (ERPS)

ERPS Subscales (Average)

Child Emotion Parental Parental Uncertainty/
Dimensions coaching Rejection Acceptance Ineffectiveness
Awareness
Sadness .12 .22 .08 .18
Anger .18 -.18 .27 -.06
Combined .17 -.01 .21 .05
Acceptance
Sadness .01 -.20 .27 -.18
Anger .22 -.19 .27 -.17
Combined .16 -.22 .31* -.20
Regulation
Sadness .19 .18 -.20 -.11
Anger .28 -.14 .28 -.33*
Combined .28 -.01 .08 -.27
Coaching
Sadness .17 -.10 .32* -.24
Anger -.06 -.30* .23 -.08
Combined .02 -.26 .30* -.15
Note. Pairwise deletion; one-tailed. * p < .05.

61
Table 16

One-Tailed Partial Correlations between Meta-Emotion Interview Child Dimensions and

Emotion-Related Parenting Styles (ERPS) Subscales Controlling for Child Age, Child
Sex, and Family Income

ERPS Subscales (Average)

Emotion Parental Parental Uncertainty/
Child coaching Rejection Acceptance Ineffectiveness
Dimensions
Awareness
Sadness .14 .22 .14 .20
Anger .16 -.21 .36* -.02
Combined .17 -.01 .29 .09
Acceptance
Sadness -.02 -.22 .22 -.19
Anger .22 -.19 .30 -.16
Combined .14 -.23 .31 -.20
Regulation
Sadness .23 .27 -.22 -.12
Anger .34* -.11 .33* -.36*
Combined .35* .03 .11 -.30
Coaching
Sadness .12 -.15 .22 -.28
Anger -.12 -.32* .25 -.05
Combined -.05 -.29 .27 -.14
Note. ―Combined‖ refers to the summed scores of sadness and anger for that dimension.
*p < .05.

62
Additional Analyses

After conducting the study’s main analyses, follow-up analyses were conducted.

The first analysis involved examining emotion coaching dichotomously (high vs. low).

The second additional analysis involved examining the relation between Meta-Emotion

Interview child dimensions (i.e., awareness, acceptance, regulation, and coaching) and

Meta-Emotion Interview parent dimensions (i.e., awareness, acceptance, and coaching).

Third, a maternal experience variable was constructed and tested in order to explore other

possible parent-related factors involved in emotion socialization.

Emotion coaching by high and low groups. It was unexpected that the emotion

coaching subscale of the ERPSST-L and ERPS would have zero correlation with many

Meta-Emotion Interview child dimensions. To further explore the nature of the zero-

correlation, two emotion coaching groups were created: relatively high emotion coaching

(using an absolute cutoff of 1SD above the mean) and relatively low emotion coaching

(using an absolute cutoff of 1SD below the mean). These cutoff standards were created

for emotion coaching scores according to the ERPSST-L and then for emotion coaching

scores according to the ERPS.

High and low emotion coaching group differences using ERPSST-L cutoffs.

Cases with an ERPSST-L emotion coaching score 1SD or greater above the mean (i.e., a

score equal to or higher than 98.99) were included in the high emotion coaching group (n

= 6); cases with an emotion coaching score 1SD or greater below the mean (i.e., a score

equal to or lower than 77.17) were included in the low emotion coaching group (n = 4).

63
 Independent samples t-tests for all mother-child dyads were conducted to test for

differences between ERPSST-L high and low emotion coaching groups on the Meta-

Emotion Interview parent and child dimensions. Results as shown in Appendix C

indicate that no statistically significant group differences were found (all ps > .05).

High and low emotion coaching group differences using ERPS cutoffs. Those

relatively high in emotion coaching (n = 8), according to the ERPS, had a score equal to

or greater than 23.52. Those relatively low in emotion coaching (n = 4) had a score of

18.52 or lower.

Independent samples t-tests for all mother-child dyads were conducted to test if

differences between ERPS high and low emotion coaching groups were present on any of

the Meta-Emotion Interview parent and child dimensions. Findings were similar to that

of the ERPSST-L, in that no statistically significant differences were found between low

and high in emotion coaching in terms of Meta-Emotion Interview dimension scores (all

ps > .05). Refer to Appendix D for complete results.

Relation between Meta-Emotion Interview parent dimensions and Meta-

Emotion Interview child dimensions. Analyses were conducted in order to determine if

scores on the Meta-Emotion Interview parent dimensions relate to scores on the Meta-

Emotion Interview child dimension after controlling for child age, child gender, and

annual family income (together, the removal of the effect of these three variables is

indicated by a subscripted x). Hierarchical Multiple Regression Analyses requires 15

cases per predictor (Field, 2009), and the sample size for the present study was just below

this mark. Being that there would be limited power in such an analysis, other methods of

testing this research question were sought.

64
 Partial correlation analyses were used to examine the relation between parenting

factors (Meta-Emotion Interview parent dimensions) and parents’ perceptions of their

children’s emotional experiences (Meta-Emotion Interview child dimensions). As shown

in Table 17, it was found that mothers who were highly aware of their own sadness,

anger, and the combination of the two were also highly aware of their children’s sadness,

anger, and combined. Additionally, mothers who were highly accepting of their own

sadness (M-a-s) were also highly aware of sadness in their children, rM-as,aw-s.x(26) = .54, p

= .003. Third, mothers who scored high in the ability to regulate their own anger (M-r-a)

also scored high in coaching of their child’s anger, rM-r-a,ca.x(26) = .40, p = .033. Due to

the absence of statistically significant correlations in some instances, the expectation that

Meta-Emotion Interview child dimensions would positively correlate with Meta-Emotion

Interview parent dimensions was only partially supported.

The role of maternal experience. To better examine maternal factors that may

be involved in one’s emotion-related parenting style, a maternal experience proxy

variable was created. This variable incorporated the age of the mother’s oldest child

(range = 1 to 22) in order to estimate how many years the mothers have taken on this

role. A score of one was assigned if the oldest child was ages 0 to 6, a score of two was

assigned for ages 7 to 12, a score of three was assigned for ages 13-17, and a score of

four was assigned for ages 18-22. Number of children (range = 1 to 4) was incorporated

in the maternal experience variable, whereby one point was scored for each child in the

family. Maternal experience scores were computed by adding scores on the two variables

previously described (range = 2 to 8). The role of maternal experience was assessed in

order to better delineate parent-related factors in the meta-emotion process. This variable

65
Table 17

Two-Tailed Partial Correlations between Meta-Emotion Interview Parent Dimensions and Child Dimensions, Controlling for Child
Age, Child Sex, and Family Income

Parent Dimensions
Parent Awareness Parent Acceptance Parent Regulation
Sadness Anger Combined Sadness Anger Combined Sadness Anger Combined
Child Dimensions
Awareness of Child
Sadness .58***. .50** .61*** .54** .16 .44* -.11 -.08 -.12
Anger 58*** .56** .64*** .34 .04 .23 -.17 .17 .04
Combined .66*** .60*** .71*** .49** .11 .37 -.16 .06 -.04
Acceptance of Child
Sadness .01 .13 .07 .17 .01 .11 .01 .06 .05
Anger .27 .27 .30 .19 .26 .30 .02 .27 .23
Combined .19 .24 .24 .21 .18 .25 .02 .21 .18
Child Regulation
Sadness .17 .14 .18 .17 .03 .12 .20 -.34 -.16
Anger .01 .03 .02 -.01 .21 .15 .18 .24 .28
Combined .09 .09 .10 .08 .16 .16 .23 -.01 .11
Coaching
Sadness .26 .27 .30 -.02 .16 .10 -.15 .34 .19
Anger .38* .35 .41* .11 .26 .24 -.17 .40* .23
Combined .38* .36 .42* .07 .25 .22 -.18 .43* .25

Note. ―Combined‖ refers to the summed scores of sadness and anger for that dimension. * p < .05, **p < .01, ***p < .001.

66
is unique in that it does not appear on the Meta-Emotion Interview and provides another

aspect of maternal factors that may relate to one’s emotion-related parenting style. By

using this method, mothers who scored relatively high on maternal experience (e.g., a

mother of four children, with the oldest child being 22 years) could be compared to

mothers scoring relatively low on maternal experience (e.g., a mother of one child who is

3 years old) in terms of meta-emotion and emotion-related parenting styles. In the

present study, overall scores on maternal experience ranged from 2 to 8. The average

maternal experience score was 3.48 (SD = 1.60).

Partial correlation analyses were conducted between maternal experience and the

Meta-Emotion Interview dimensions, ERPSST-L subscales, and ERPS subscales after

removing the effect of child age, child sex, and family income. Findings, as shown in a

column of Table 18, suggest that the relation between maternal experience(me) and

regulation of maternal sadness (re-s) was statistically significant, rme,re-s.x(26) = -.41, p =

.029. All other partial correlations were not statistically significant.

Additionally, a two-tailed correlational analysis was conducted in order to

determine if maternal experience scores related to maternal age. Findings suggest that

greater maternal age was related to greater maternal experience scores, r(31) = .66, p <

.001.

Though not tested directly in the present study, cohort effects in terms of maternal

age may be related to maternal level of experience. As such, partial correlation analyses

were conducted (as shown in Table 18) between maternal experience and the Meta-

Emotion Interview dimensions, ERPSST-L subscales, and ERPS subscales after

removing the effect of child age, child sex, family income, and maternal age (ma).

67
Table 18

Two-Tailed Partial Correlations between Maternal Experience and the Meta-Emotion

Interview Child Dimensions, Emotion-Related Parenting Styles Self-Test – Likert
(ERPSST-L) Subscales, and Emotion Related Parenting Styles (ERPS) Subscales

Maternal Experience
Controlling for child age, Controlling for child age,
child sex, and family child sex, family income,
income and maternal age
Parent Awareness
Sadness .19 .21
Meta- Anger .10 .06
Emotion Combined .17 .17
Interview Parent Acceptance
Variables Sadness -.17 -.26
Anger -.33 -.14
Combined -.34 -.25
Parent Regulation
Sadness -.41* -.24
Anger -.19 -.08
Combined -.36 -.18
Child Awareness
Sadness .27 .28
Anger .28 .26
Combined .31 .31
Child Acceptance
Sadness -.36 -.56**
Anger -.08 -.10
Combined -.22 -.32
Child Regulation
Sadness .01 .05
Anger -.02 .02
Combined -.01 .04
Child Coaching
Sadness .04 .14
Anger .04 .13
Combined .05 .15
ERPSST-L Emotion Coaching .11 .05
Variables Laissez Faire .20 .19
Dismissing -.04 -.06
Disapproving -.11 -.09
ERPS Emotion Coaching .22 .15
Variables Parental Rejection -.19 -.16
Parental Acceptance -.01 -.12
Uncertainty/Ineffectiveness .15 .23
Note. ―Combined‖ refers to the summed scores of sadness and anger for that dimension.
*p < .05, **p < .01.

68
Findings suggest that greater maternal experience (me) was associated with a lower

acceptance of child sadness, rme,ac-s.x,ma(23) = -.56, p = .003. All other partial correlations

were not statistically significant.

Differential responses based on emotion type: Sadness and anger. In the

present study, statistically significant findings were found between the Meta-Emotion

Interview and ERPSST-L and ERPS. These findings involved Meta-Emotion Interview

dimensions related to child anger, maternal use of dismissing and disapproving emotion-

related parenting style subscales of the ERPSST-L, and the parental rejection of child

emotions subscale of the ERPS. To begin, the Meta-Emotion Interview contained an

equal number of questions related to sadness and anger. This was not the case with the

ERPSST-L and ERPS. Sadness-specific content (i.e., items that contained the words

―sad‖ or ―sadness‖) were found on 32/81 items, whereas anger-specific content (i.e.,

items that contained the words ―anger,‖ ―angry,‖ or ―mad‖) were found on 48/81 items on

the ERPSST-L. A greater number of items related to anger also were found, with 9/20

related to sadness and 11/20 related to anger. Because more questions on the ERPSST-L

and ERPS pertained to anger, it is possible that the questionnaires assess anger better than

sadness. A second consideration is that children’s display of anger is often met with

greater maternal invalidation as compared to children’s display of sadness (Shipman et

al., 2007). Thus, it should not be surprising that in the present study many dimensions of

child anger related to maternal use of dismissing and disapproving styles.

Examples from Mothers’ and Fathers’ Meta-Emotion Interviews

Many parents provided unique perspectives and insight regarding meta-emotion

and the study of meta-emotion as a whole. In coding the Meta-Emotion Interviews, some

69
response patterns were acknowledged and were deemed notable for representing the

construct. In the transcription of the Meta-Emotion Interviews, ―P‖ referred to parent

and ―I‖ referred to interviewer. Notable perspectives for mothers will first be discussed

followed by that for fathers.

Mothers. In responding to their children’s negative emotions, many mothers

demonstrated a preference for their children to be soothed before getting involved. This

corresponds to Meta-Emotion Interview item E10 of the child acceptance dimension: ―P

prefers child to be soothed before P gets involved.‖

P: And so I’ll tell him ―Okay if you don’t want to talk right now. Because you

are so hyper. You’re so angry. Go. Go calm down. Then come back and talk‖

Some mothers received a low score in showing the child respect towards his or

her emotional experiences. Under the child coaching dimension, this refers to Meta-

Emotion Interview item F1: ―P shows respect for child’s experience of emotions.‖

P: When I see him pouting, I always say oh my god, is this my son? He looks

so ugly, his face looks so ugly. He’s (cautious?) and cute. You know. He likes

that, so he’s like (stretches?) his face. He goes, ―Mom, this is your son. I’m

not ugly.‖ You know. I’m like okay! And that’s the extent of it, really.

P: I just tell him he has an ugly face. (Laughs). And I tell him it’s not worth

70
 it. Like what are you angry about, not getting a toy? Or not being able to

watch a movie? Like that’s really, really… it’s not good.

Fathers. When it came to the child dimensions (i.e., awareness, acceptance,

coaching, and regulation), the fathers described a variety of their strategies and

approaches in dealing with their child’s sadness and anger.

One item on the dimension for child acceptance of emotion states ―P uses a mental

(analytical) approach to C’s emotions,‖ in which a low score is provided if the parent

uses this approach. Two of the three fathers whose interviews were scored used a

rational approach in dealing with their children’s sadness, while all of the fathers used a

rational approach in dealing with their children’s anger. The fathers, therefore, lost

points on the acceptance dimension for using a rational approach, which is described as

the parent teaching the child to approach the emotion in a way to figure it out or analyze

it in order to discover a rational way to resolve the emotion.

P: I guess that it’s okay to feel angry but that she should really think about

what she’s getting angry about. And you know. Ask herself whether it is

really worth it. You know. Because there’s always, you know sometimes

she’ll just get frustrated and angry and I’ll tell her you know, (child’s name)

there’s more than one way to skin a cat. You know. Or something to that

effect. Or there’s other ways of working around problem instead of hitting

them head on and getting angry about them. And I’m just trying to teach her

that it’s okay to get angry about certain things. But uh, it’s not always the best

71
 way of going about it. And you know. Working through a problem is better

than just getting angry sometimes.

All fathers in the present study specifically stated that they want their children to

talk with them about both sadness and anger.

I: OK. If you could sum it up, what are you trying to teach (daughter’s name)

about the world of feelings?

P: That it’s not bad to have any feeling. And, it’s alright to even express your

feelings. As her dad I hope she can tell me whenever she’s having any of

these feelings.

Similarly, another father described how he felt that sadness is a valuable

experience. Sadness was described as important for development.

P: But I, I think there’s probably still a lesson to be learned. And I'm really

having a hard time describing exactly what that is. But I think there’s

probably something that you know, makes your heart good. Makes you a

better person when under certain circumstances you experience sadness. And

it’s going to be hard for me to pinpoint it if you want me to get more direct

than that. But I really feel that there is something there. Something to be said

through that experience.

72
 Some parents, such as this father, specifically stated they do not use distraction

techniques with their children. Under the child acceptance dimension, this refers to

Meta-Emotion Interview item E11: ―Parent ever distracts from emotion.‖

P: Mm. We just talk. Talk about it. We’ll discuss it. I ask her questions and

I’ll let her tell me what’s on her mind. And I don’t say ―Let’s go get ice

cream‖ or whatever.

Emotion-related parenting styles can differ greatly from parent to parent, even

when the sample is homogenous in terms of demographics. Examining specific

comments from the Meta-Emotion Interview was helpful in further exploring how one’s

meta-emotion philosophy shapes emotion-related parenting styles.

73
 CHAPTER V

DISCUSSION

The aim of the present study was to evaluate the construct validity of the

Emotion-Related Parenting Styles Self-Test – Likert (ERPSST-L) and the Emotion-

Related Parenting Styles (ERPS) in conjunction with the Meta-Emotion Interview.

Findings partially supported the construct validity of the ERPSST-L and ERPS in that

correlation coefficients, when statistically significant, presented in the hypothesized

direction.

However, unanticipated nonsignificant correlations existed in the study results

such as nonsignificant differences in mothers’ emotion socialization of boys and girls,

and this may have been a reflection of a small, homogenous sample. Lack of construct

validity, according to Mitchell (1985), may indicate contamination (variance in the

measure that is not present in the construct) and/or deficiency (variance in the construct is

not captured by the measure). Threats to construct validity (Cherulnik, 2001, p. 67) were

considered and examined in the present study. If threats to construct validity are not

addressed, construct validity may not be found. A lack of construct validity might also

indicate that there is a problem with the theory, measurement strategy, item content, or

the construct might not be specified very well (Westen & Rosenthal, 2005).

Construct Validity of the Emotion-Related Parenting Styles Self-Test – Likert

(ERPSST-L)

It was expected that scores on the long form would correlate with scores on the

Meta-Emotion Interview in previously specified directions (refer to Table 2). This

hypothesis was partially supported. Specifically, a positive correlation was anticipated

74
between scores on the emotion coaching subscale and ratings on all Meta-Emotion

Interview child dimensions (i.e., awareness, acceptance, regulation, and coaching).

When correlation coefficients were statistically significant, they also were in the

anticipated direction. A notable positive correlation was present between scores on

emotion coaching and ratings on the child’s ability to regulate anger. The significance of

emotion socialization pertaining to anger is later discussed. Next, it was expected that the

laissez-faire scale would positively correlate with the child awareness and child

acceptance dimensions and would negatively correlate with the child regulation and

coaching dimensions. This expectation was not confirmed, as correlations between this

scale and the Meta-Emotion Interview child dimensions were very small or had a zero

correlation. This may be due to the fact that the laissez-faire scale had an internal

consistency reliability that was lower (α = .59) relative to the other subscales. Thus, the

results linked to the construct validity for the laissez-faire scale should be interpreted

with caution. Scores on the dismissing scale were expected to negatively correlate with

all of the Meta-Emotion Interview child dimensions. When results were statistically

significant, correlations were moderate and negative, supporting the hypothesis. In

particular, high scores on the child acceptance dimensions and coaching dimensions were

associated with lower scores on the dismissing subscale. Scores on the disapproving

scale were expected to negatively correlate with Meta-Emotion Interview child

dimensions. The disapproving subscale negatively correlated with the acceptance and

regulation of child anger. Overall, when statistically significant relations were found

between the ERPSST-L subscales and the Meta-Emotion Interview dimensions, they also

were in the hypothesized direction.

75
Construct Validity of the Emotion-Related Parenting Styles (ERPS)

Another goal of the present study was to test the construct validity of the short

form (ERPS) in conjunction with scores on the Meta-Emotion Interview child

dimensions. Results partially supported the hypotheses (refer to Table 3). It was

expected that scores on the emotion coaching scale would positively correlate with all

Meta-Emotion Interview child dimensions. Emotion coaching correlated positively with

child regulation of anger and the regulation of combined sadness and anger. Next, it was

predicted that parental rejection of negative emotion would correlate negatively with all

Meta-Emotion Interview child dimensions. In the present study, mothers who were rated

as high in parental rejection of negative emotion also were rated as being low in the

coaching of child anger. As was the case for the emotion coaching subscale, it was

expected that parental acceptance of negative emotion would correlate positively to all

Meta-Emotion Interview child dimensions. Mothers who scored high in parental

acceptance of negative emotion tended to score high in awareness and regulation of child

anger. Finally, it was expected that mothers who feel uncertain or ineffective in emotion

socialization also would be rated as having low scores on all Meta-Emotion Interview

child dimensions. Findings suggest that mothers who scored highest in uncertainty and

ineffectiveness had children who were rated as having difficulty regulating anger. This

may exemplify a situation in which a mother of a child who has great difficulty

regulating his or her anger may feel inadequate or unsuccessful in helping the child deal

with anger. The finding for the uncertainty/ineffectiveness scale is in contrast to the

finding that scores on the ERPS emotion coaching and parental acceptance scales

positively correlated with scores on child regulation of anger. Though not tested directly,

76
it can be suggested that a child’s ability to regulate anger may have a bidirectional effect

on a parent’s emotion-related parenting style. Otherwise, the relation between the

uncertainty/ineffectiveness parenting style and dimensions on the Meta-Emotion

Interview were not statistically significant. This may have been related to the fact that

this subscale had a relatively low internal consistency reliability of .67. Additionally, it is

important to note that the uncertainty/ineffectiveness scale was designed in a study that

used a sample of parents of children with developmental disabilities.

The relation between parent dimensions and child dimensions of the Meta-

Emotion Interview also was explored. The expectation that Meta-Emotion Interview

child dimensions would positively correlate with Meta-Emotion Interview parent

dimensions was only partially supported. The strongest findings were found between the

statistically significant, positive relation between child awareness dimensions and parent

awareness dimensions. The association between the parent and child dimensions of the

Meta-Emotion Interview have been explored in other studies. For example, Hunter and

colleagues (2011) administered the Meta-Emotion Interview to 148 mothers and 106

fathers. They found that both mothers’ and fathers’ scores on the parent dimensions of

the Meta-Emotion Interview positively related to scores on the child dimensions. They

also found that mothers’ scores on the parent dimensions positively correlated with

fathers’ scores on the parent dimensions. In summary, though the relation between

parent dimensions and child dimensions of the MEI was only partially supported in the

small sample used in the present study, it has been clearly supported in other research

using a larger sample (Hunter et al., 2011).

77
Limitations and Directions for Future Research

The first limitation of the present study is the small sample size (N = 33 mothers).

Small sample size limits statistical power, which is why trends often were found when

relations were not always statistically significant. Statistical power also was reduced

when multiple control variables were used. A limitation of the present study is the

exclusion of fathers from analyses. Fathers were not included in the main analyses of the

present study because there were so few. The current study focused on maternal meta-

emotion philosophy and should not be interpreted to be representative of paternal

emotion socialization practices. Father-child relationships need to be explored in terms

of emotion socialization processes. Within intact families, it may be useful to explore

how spouse pairs and their children contribute to the many multi-directional processes at

play (Eisenberg, 1996).

Second, the sample was homogenous in terms of ethnicity and level of education.

External validity also may be limited because the sample in the present study was quite

homogenous in terms of emotion-related parenting styles. Approximately 81.8% (n = 27)

of the mothers were classified as predominately emotion coaching by the ERPSST-L, and

78.8% (n = 26) of the mothers were classified as predominately emotion coaching by the

ERPS. No mother was classified as predominately dismissing, disapproving, rejecting of

negative emotions, or uncertain/ineffective. The group, therefore, was not diverse and

was restrictive in terms of emotion-related parenting styles. When a range of scores is

restricted, correlations may show as weak or nonexistent. Contrary to expectations based

on past research, the results from an ANOVA indicated that Meta-Emotion Interview

scores did not differ based on either an ERPSST-L or ERPS emotion coaching or non-

78
emotion coaching status. Preliminary findings in the present study indicate that in order

to better understand the emotion socialization process, maternal experience is a variable

worthy of exploration in future studies. Future studies should incorporate a more diverse

sample and also should consider other variables, such as maternal experience.

Potential threats to hypothesis validity, as described by Wampold, Davis, and

Good (1990), were evaluated in the present study. The first threat to address is

inconsequential research hypotheses. In this study, results have the potential to answer an

important question in the theory of meta-emotion: Researchers need to know if the

ERPSST-L and ERPS hold construct validity in conjunction with the original Meta-

Emotion Interview. The second threat, ambiguous research hypotheses, also was

considered. The hypotheses in the present study were stated with specific directions and

predictions in order to reduce ambiguity. The third threat is the noncongruence of

research hypotheses and statistical tests. In this study, both the hypotheses and statistical

tests accentuate the direction and strength of the relations as assessed through correlation.

The final threat is diffuse statistical hypotheses and tests, which is when too many

analyses are conducted per hypothesis. In the present study, an effort was made to avoid

using multiple statistical tests for any given hypothesis.

The measures used in the present study also pose limitations. One limitation of

the Meta-Emotion Interview is that the replacement of ―don’t know‖ scores with

dimension mean scores may be an inaccurate representation of the mother’s meta-

emotion philosophy. For example, on the awareness of child’s emotion dimension (9

items with a possible score range from 1 to 4), a mother received a score of 4 on two

items but was coded ―don’t know‖ for the remaining seven items. After mean

79
substitution, that mother received a perfect score of 36 on that dimension, which is likely

an overestimate. When administering the Meta-Emotion Interview, is that it is the

responsibility of the interviewer to ensure that the parent continues to discuss only the

target child. However, on the questionnaire, it is possible for a parent to deviate from this

expectation, as shown in the comments from parents below.

P: Now if we’re talking about my other son it’s different. It’s a whole

different relationship. Which is why on the questionnaire I think I may have

mismatched some questions.

Another parent discussed a limitation of both the questionnaire and the possible

inconsistency and/or situationally-dependent emotion-related parenting style:

P: ….. Like I said with the, with the questionnaire. I found that it was kind of

difficult to answer because there are different situations for when he’s feeling

angry. And then there’s different situations for when he’s feeling sadness.

But, but I mean for different reasons for different…

…..So it’s kind of hard to answer like in a questionnaire. The different

situations when it’s happening.

Gathering the full picture of emotion socialization in families may be best done

by using multiple methodologies. The interview provides parents with an opportunity to

describe and clarify their meta-emotion philosophies by using a variety of examples.

80
Questionnaires, such as the ERPSTT-L and ERPS, are helpful in quickly getting a sense

of a parent’s meta-emotion philosophy.

The exploration of meta-emotion philosophy as developed by Gottman, Katz, and

Hooven (2006) has expanded to areas beyond parent-child relationships. Over the past

decade, there have been some promising developments in the area of meta-emotion. The

construct of meta-emotion also has been explored in marital relationships (e.g., Schwab,

2001; Yoshimoto, 2005) and has been used to enhance an understanding of relational

aggression and emotional regulation (Bowie, 2010). Currently, meta-emotion is being

explored in children and adolescents. For instance, Taylor and Carrère (2002) established

a meta-emotion interview and coding system for children ages 7 to 8, known as the

Family Health Project Child Meta-Emotion Interview. For older children, Windecker-

Nelson and Katz (2004) published the Child–Adolescent Meta-Emotion Coding System.

This measure has been used to investigate emotional competence and risky behaviour of

adolescents (Hessler & Katz, 2010), to assess emotion regulation and physiological

responses during peer provocation (Hessler & Katz, 2007), and to investigate the emotion

competence of children who have been exposed to domestic violence (Katz, Hessler, &

Annest, 2007). In a study by Hunter and colleagues (2011), 75 depressed and 77 healthy

adolescents completed the Child and Adolescent Meta-Emotion Interview (Katz &

Windecker-Nelson, 2004), while mothers and fathers completed the Meta-Emotion

Interview. They found that dimensions of the Child and Adolescent Meta-Emotion

Interview were found to significantly correlate with parent dimensions of the Meta-

Emotion Interview. Thus, it remains appropriate to examine meta-emotion philosophy in

adolescence. With both a measure for child meta-emotion and adolescent meta-emotion,

81
it may be possible to follow a child’s meta-emotion longitudinally. Perhaps meta-

emotion philosophy can later be investigated in other family compositions (e.g.,

grandparent-child families and stepparent-stepchild relationships), as well as in child-care

settings (e.g., socialization practices of Early Childhood Educators).

Clearly, meta-emotion philosophy has expanded beyond parent-child dyads. It

would be optimal to explore meta-emotion philosophy using multiple informants (e.g.,

teachers) via multiple methods (i.e., interview, self-report questionnaires, and

observations). Using solely a self-report questionnaire poses a number of problems. The

first is social desirability, which is a research concern particularly for face valid

questionnaires. In addition to social desirability, another limitation of using a self-report

questionnaire is that results may not necessarily inform what occurs in the home.

Respondents may describe emotion coaching techniques, for example, that they would

ideally use, not ones that are typically implemented. It also would be useful to examine

the ERPSST-L and ERPS in more longitudinal studies, in order to examine the stability

of emotion-related parenting styles.

It is recommended that when analyzing data from the Meta-Emotion Interview,

the dimension scores should be examined separately for each emotion as well as a

combined scale. As previously discussed, the most statistically significant findings in the

present study involved dimensions related to anger. Does anger, as compared to sadness,

play a special role in parental meta-emotion philosophy? Are there other possible

explanations for this relation? Displays of anger in children tend to be met with more

negative consequences from parents than are displays of sadness (Shipman et al., 2007).

Anger may draw for more of the dismissing and disapproving emotion-related parenting

82
styles. However, when mothers coach adolescents through anger, the adolescents tend to

have better anger regulation and less externalizing behaviour (Shortt, Stoolmiller, Smith-

Shine, Eddy, & Sheeber, 2010). Thus, the context and nature of the emotion may inform

the emotion socialization process. In a study by Shipman and colleagues (2007), it was

found that mothers who physically maltreat their children tended to use more invalidation

towards child anger as compared to child sadness or fear. Emotion type may play a role

in the context for emotion socialization. As compared to expressions of sadness or fear,

these expressions of anger in children may more likely be met with invalidating parental

behaviours such as minimization (Shipman et al., 2007). Thus, when parental response to

child anger is evaluated, the dismissing and disapproving responses tend to be utilized to

a greater degree as compared to parental response to child sadness or fear. Another

consideration is that when working with clinical samples, depressed adolescents may

display more intense anger than healthy adolescents (Sheeber, Allen, Leve, Davis, Shortt,

& Katz, 2009). In a case where parents of adolescents with intense anger are tested, it

would be important to remember that one’s predominant emotion-related parenting styles

may vary for different emotions. For example, a parent of a child with intense anger may

be highly disapproving of anger but may also score high in emotion coaching of sadness.

As discussed earlier, more questions on the ERPSST-L and ERPS targeted anger as

compared to sadness. The imbalance of questions for sadness and anger on the ERPSST-

L and ERPS may be one explanation for the differential findings; however, past research

may be informative as well. As described earlier, anger may draw for more of the

dismissing and disapproving emotion-related parenting styles because mothers tend to

83
respond to expressions of anger with more negative consequences than they would for

sadness (Shipman et al., 2007).

Studying meta-emotion philosophy is not only rich with information for

researchers, but it can also be a valuable experience for parents. In the Meta-Emotion

Interview, many parents commented on their experience in participating in the present

study. Some parents expressed that it is important to socialize positive emotions as well.

P: Cause I don’t think—uh in part of the question here it said something

about reading stories about emotions. I don’t know if I really ever read

anything in particular dealing with a particular problem. But this study has

open my eyes to that. Well maybe when there is a problem we’ll get a book.

And read it together. Because I read to her everyday but not just stories that

she likes like fun and happiness. You know?

I: Yes. It sounds like you have unconsciously addressed some of these issues.

The parent in the previous excerpt indicated that emotion socialization through

narratives tends to concern positive emotions. It is important to remember that meta-

emotion philosophy is not restricted to unpleasant emotions, and it is a limitation that

only sadness and anger were explored in the present study. Both Gottman and colleagues

(1996) and Cowan (1996) proposed that it would be valuable to use the Meta-Emotion

Interview to assess positive emotions. Recently, the interview has been adapted to

evaluate other emotions such as pride, love, and affection. This measure, known as the

Parenting Meta-Emotion Interview: A Modification of the Original Meta-Emotion

84
Interview, was developed by Katz and Carrère (as cited in Doohan, Carrère, & Taylor,

2004). As previously described, parents may be high in acceptance for some emotions

but not others. It is, therefore, important to apply measures of emotion-related parenting

styles to positive emotions as well.

One part of the present study involved examining the relation between scores on

Meta-Emotion Interview parent dimensions and scores on child dimensions. The purpose

of this was to inform, but not necessarily to confirm, factors that may be involved in the

bidirectional exchange of emotion socialization between mothers and their children. For

example, mothers high in awareness of their own sadness were also rated as being well-

aware of their child’s sadness. Thus, parental factors, such as emotional awareness, can

inform the way they perceive and accordingly interact with their children. Katz, Maliken,

and Stettler (2012) urged researchers to explore bidirectional relations between parent

characteristics and child characteristics.

A major strength of the present study was that it fulfilled the need to test the

construct validity of the ERPSST-L and ERPS using the original meta-emotion interview

scores. Testing the construct validity of measures for meta-emotion philosophy is just the

beginning. With well-validated measures of meta-emotion, researchers can better explore

emotion socialization processes that contribute to emotional development in children. In

summary, construct validity for the ERPSST-L and ERPS was partially supported in the

present study. Statistically significant correlations were presented in the hypothesized

direction. The further validation of time-efficient and user-friendly measures, such as the

ERPSST-L and ERPS, may encourage more research concerning the construct of meta-

emotion philosophy.

85
 REFERENCES

Baker, J. K., Fenning, R. M. & Crnic, K. A. (2010). Emotion socialization by

mothers and fathers: Coherence among behaviors and associations with parent

attitudes and children's social competence. Social Development, 20(2), 412–430.

doi:10.1111/j.1467-9507. 2010.00585.x

Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology

Monograph, 4, 1-103.

Bowie, B. H. (2010). Understanding the gender differences in pathways to social

deviancy: Relational aggression and emotion regulation. Archives of Psychiatric

Nursing, 24(1), 27–37. doi:10.1016/j.apnu.2009.04.007

Cherulnik, P. D. (2001). Methods for behavioural research: A systematic approach.

California: Sage (pp. 15–236).

Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in objective scale

development. Psychological Assessment, 7(3), 309–319.

Cowan, P. A. (1996). Meta-thoughts on the role of meta-emotion in children’s

development: Comment on Gottman et al. (1996). Journal of Family Psychology,

10(3), 277–283. doi:10.1037/0893-3200.10.3.277

Crano, W. D., & Brewer, M. B. (2002). Principles and methods of social research:

Second edition. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. (pp. 33–113).

Cronbach, L. J, & Meehl, P. E. (1955). Construct validity in psychological tests.

Psychological Bulletin, 52, 281–302.

Cunningham, J. N., Kliewer, W., & Garner, P. W. (2009). Emotion socialization, child

86
 emotion understanding and regulation, and adjustment in urban African American

families: Differential associations across child gender. Development and

Psychopathology, 21, 261–283. doi:10.1017/50954579409000157

DeOliveira, C. A., Moran, G., & Pederson, D. R. (2005). Understanding the link between

maternal adult attachment classifications and thoughts and feelings about

emotions. Attachment & Human Development, 7(2), 153–170.

doi:10.1080/14616730500135032

Doohan, E., Carrère, S., & Taylor, M. (2004). The meta-emotion interviews and coding

systems. In V. Manusov (Ed.), Nonverbal communication: A sourcebook of

methods. Hillsdale, NJ: Lawrence Erlbaum Associates.

Dunsmore, J. C., Her, P., Halberstadt, A. G., & Perez-Rivera, M. B. (2009). Parents’

beliefs about emotions and children’s recognition of parents’ emotions. Journal of

Nonverbal Behavior, 33, 121–140. doi:10.1007/s10919-008-0066-6

Eisenberg, N. (1996). Meta-emotion and socialization of emotion in the family–A topic

whose time has come: Comment on Gottman et al. (1996). Journal of Family

Psychology, 10(3), 269–276. doi:10.1037/0893-3200.10.3.269

Eisenberg, N., Cumberland, A., & Spinrad, T. L. (1998). Parental socialization of

emotion. Psychological Inquiry, 9(4), 241–273. doi:10.1207/s15327965pli0904_1

Eisenberg, N., Spinrad, T. L., & Cumberland, A. (1998). The socialization of emotion:

Reply to commentaries. Psychological Inquiry, 9(4), 317–333.

doi:10.1207/s15327965pli0904_17

Fabes, R. A., Eisenberg, N., & Bernzweig, J. (1990). The Coping with Children’s

87
 Negative Emotions Scale: Description and scoring. Unpublished scale,

Department of Family Resources and Human Development, Arizona State

University, Tempe.

Fabes, R. A., Eisenberg, N., Karbon, M., Bernzweig, J., Speer, A. L., & Carlo, G. (1994).

Socialization of children’s vicarious emotional responding and prosocial

behavior: Relations with mothers’ perceptions of children’s emotional reactivity.

Developmental Psychology, 30(1), 44–55. doi:10.1037/0012-1649.30.1.44

Field, A. P. (2009). Discovering Statistics Using SPSS (3rd Ed.) (pp. 113–359). London:

Sage.

Ginott, H. G. (1965). Between parent and child. New York: Macmillan.

Gottman, J., & DeClaire, J. (1997). Raising an emotionally intelligent child: The heart of

parenting. New York: Simon & Schuster.

Gottman, J. M., Katz, L. F., & Hooven, C. (1996). Parental meta-emotion philosophy and

the emotional life of families: Theoretical models and preliminary data. Journal of

Family Psychology, 10(3), 243–268. doi:10.1037/0893-3200.10.3.243

Gottman, J. M., Katz, L. F., & Hooven, C. (1997). Meta-emotion: How families

communicate emotionally. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

Hakim-Larson, J., Dunham, K., Vellet, S., Murdaca, L., & Levenbach, J. (1999). Parental

affect and coping. Canadian Journal of Behavioural Sciences, 31(1), 5–18.

doi:10.1037/h0087069

Hakim-Larson, J., Parker, A., Lee, C., Goodwin, J., & Voelker, S. (2006). Measuring

88
 parental meta-emotion: Psychometric properties of the Emotion-Related Parenting

Styles Self-Test. Early Education and Development, 17(2), 229–251.

doi:10.1207/s15566935eed1702_2

Halberstadt, A. G., Cassidy, J., Stifter, C. A., Parke, R. D., & Fox, N. A. (1995). Self-

expressiveness within the family context: Psychometric support for a new

measure. Psychological Assessment, 7, 93–103.

Haynes, S. N. (2001). Clinical applications of analogue behavioral observation:

Dimensions of psychometric evaluation. In A. Kazdin, Methodological issues and

strategies in clinical research: Third edition (pp. 235–264). Washington, DC:

American Psychological Association.

Hessler, D. M., & Katz, L. F. (2007). Children’s emotion regulation: Self-report and

physiological response to peer provocation. Developmental Psychology, 43(1),

27–38. doi:10.1037/0012-1649.43.1.27

Hessler, D. M., & Katz, L. F. (2010). Brief report: Associations between emotional

competence and adolescent risky behaviour. Journal of Adolescence, 33(1), 241–

246. doi:10.1016/j.adolescence.2009.04.007

Hunter, E. C., Katz, L. F., Shortt, J. W., Davis, B., Leve, C., Allen, N. B., & Sheeber, L.

B. (2011). How do I feel about feelings? Emotion socialization in families of

depressed and healthy adolescents. Journal of Youth and Adolescence, 40, 428–

441. doi:10.1007/s10964-010-9545-2

Jäger, C., & Bartsch, A. (2006). Meta-emotions. Grazer Philosophische Studien. 73, 179–

204.

Katz, L. F., & Gottman, J. M. (1986). The Meta-Emotion Interview. Unpublished

89
 manuscript, University of Washington.

Katz, L. F., Gottman, J. M., & Hooven, C. (1996). Meta-emotion philosophy and family

functioning: Reply to Cowan (1996) and Eisenberg (1996). Journal of Family

Psychology, 10(3), 284–291. doi:10.1037/0893-3200.10.3.284

Katz, L. F., Hessler, D. M., & Annest, A. (2007). Domestic violence, emotional

competence, and child adjustment. Social Development, 16(3), 513–538.

doi:10.1111/j.1467-9507. 2007.00401.x

Katz, L. F., & Hunter, E. C. (2007). Maternal meta-emotion philosophy and adolescent

depressive symptomatology. Social Development, 16(2), 343–360.

doi:10.1111/j.1467-9507.2007.00388.x

Katz, L. F., Hunter, E., & Klowden, A. (2008). Intimate partner violence and children’s

reaction to peer provocation: The moderating role of emotion coaching. Journal

of Family Psychology, 22(4), 614–621. doi:10.1037/a0012793

Katz, L. F., Maliken, A. C., & Stettler, N. M. (2012). Parental meta-emotion philosophy:

A review of research and theoretical framework. Child Development Perspectives,

0(0), 1–6. doi:10.1111/j.1750-8606.2012.00244.x

Katz, L. F., Mittman, A., & Embry, L. (n.d.). The Meta Emotion Coding System Coding

Training Manual. Unpublished manuscript, University of Washington.

Katz, L. F., & Windecker-Nelson, B. (2004). Parental meta-emotion philosophy in

families with conduct-problem children: Links with peer relations. Journal of

Abnormal Child Psychology, 32(4), 385–398.

doi:10.1023/B:JACP.0000030292.36168.30

Katz, L. F., & Windecker-Nelson, B. (2006). Domestic violence, emotion coaching, and

90
 child adjustment. Journal of Family Psychology, 20(1), 56–67.

doi:10.1037/0893-3200.20.1.56

Lagacé-Séguin, D. G., & Coplan, R. J. (2005). Maternal emotional styles and child social

adjustment: Assessment, correlates, outcomes, and goodness of fit in early

childhood. Social Development, 14(4), 613–636.

doi:10.1111/j.1467-9507. 2005.00320.x

Lagacé-Séguin, D. G., & Gionet, A. (2009). Parental meta-emotion and temperament

predict coping skills in early adolescence. International Journal of Adolescence

and Youth, 14¸ 367–382. doi:10.1080/02673843. 2009.9748015

Lee, C. H. (1999). The Parenting Styles Self-Test: Reliability and construct validity.

Unpublished Master’s Thesis, University of Windsor, Windsor, Ontario, Canada.

Lee, C. H., Hakim-Larson, J., & Voelker, S. (2000, June). The Parenting Styles Self-Test:

Psychometric properties. Poster session presented at the annual meeting of the

Canadian Psychological Association, Ottawa, Ontario.

McGraw, K. O., & Wong, S. P. (1996). Forming inferences about some intraclass

correlation coefficients. Psychological Methods, 1(1), 30–46. doi: 10.1037/1082-

989X.1.1.30

Mills, S. L., Freeman, W. S., Clara, I. P., Elgar, F. J., Walling, B. R., & Mak, L. (2007).

Parent proneness to shame and the use of psychological control. Journal of Child

and Family Studies, 16, 359–374. doi:10.1007/s10826-006-9091-4

Mitchell, T. R. (1985). An evaluation of the validity of correlational research conducted

in organizations. Academy of Management Review, 10(2), 192–205.

91
Paterson, A. D., Babb, K. A., Camodeca, A., Goodwin, J., Hakim-Larson, J., Voelker, S.,

& Gragg, M. (2012). Emotion-Related Parenting Styles (ERPS): A short form for

measuring parental meta-emotion philosophy. Early Education and Development,

23, 583-602.

Saarni, C. (1985). Indirect processes in affect socialization. In M. Lewis & C. Saarni

(Eds), Handbook of emotions (pp. 187–209). New York: Plenum.

Saarni, C. (1999). The Development of Emotional Competence. New York: Guilford.

Schwab, J. H. (2001). Meta-emotion and marital satisfaction. Dissertation Abstracts

International, 62(10B), 4804.

Sheeber, L. B., Allen, N. B., Leve, C., Davis, B., Shortt, J. W., & Katz, L. F. (2009).

Dynamics in affective experience and behaviour in depressed adolescents.

Journal of Child Psychology and Psychiatry, 50(11), 1419–1427.

doi:10.1111/j.1469-7610.2009.02148.x

Shine, S., & Wampler, K. S. (1997). The role of emotion in the family: Building a theory.

Contemporary Psychology, 43(6), 421–422. doi:10.1037/001685

Shipman, K. L., Schneider, R., Fitzgerald, M. M., Sims, C., Swisher, L., & Edwards, A.

(2007). Maternal emotion socialization in maltreating and non-maltreating

families: Implications for children’s emotion regulation. Social Development,

162(2), 268–285. doi:10.1111/j.1467-9507.2007.00384.x

Shortt, J. W., Stoolmiller, M., Smith-Shine, J. N., Eddy, J. M., & Sheeber, L. (2010).

Maternal emotion coaching, adolescent anger regulation, and siblings’

externalizing symptoms. Journal of Child Psychology and Psychiatry, 51(7), 799–

808. doi:10.1111/j.1469-7610.2009.02207.x

92
Smith, G. T. (2005). On construct validity: Issues of method and measurement.

Psychological Assessment, 17(4), 396–408. doi:10.1037/1040-3590.17.4.396

Tabachnick, B. G., & Fidell, L. S. (2012). Using Multivariate Statistics 6th ed. Upper

Saddle River, NJ: Pearson.

Taylor, M.G., & Carrère S. (2002). The child meta-emotion interview and coding system

for 7–8 year-olds. Unpublished manuscript, University of Washington at Seattle.

Walling, B. R., Mills, R. S. L., & Freeman, W. S. (2007). Parenting cognitions associated

with the use of psychological control. Journal of Child and Family Studies, 16,

642–659. doi:10.1007/s10826-006-9113-2

Wampold, B. E., Davis, B., & Good III, R. H. (1990). Hypothesis validity in clinical

research. In A. Kazdin, Methodological issues and strategies in clinical research:

Third edition (pp. 389-406). Washington, DC: American Psychological

Association.

Westen, D., & Rosenthal, R. (2005). Improving construct validity: Cronbach, Meehl, and

Neurath’s Ship. Psychological Assessment, 17(4), 400–412. doi:10.1037/1040-

3590.17.4.409

Windecker-Nelson, B., & Katz, L.F. (2004). The Child/Adolescent Meta-Emotion

Coding System. Unpublished manuscript, University of Washington.

Yap, M. B. H., Allen, N. B., Leve, C., & Katz, L. F. (2008). Maternal meta-emotion

philosophy and socialization of adolescent affect: The moderating role of

adolescent temperament. Journal of Family Psychology, 22(5), 688–700.

doi:10.1037/a0013104

93
Yoshimoto, D. K. (2005). Marital meta-emotion: Emotion coaching and dyadic

interaction. Dissertation Abstracts International: Section B: The Sciences and

Engineering, 66(6-B), 3348.

94
 APPENDIX A

Inter-Rater Reliability for Coding of Meta-Emotion Interviews (MEIs) for Coding Training, as Measured by Intra-Class Correlation
Coefficients

MEI Dimension Gottman Lab Rater 2 Rater 3

(15 training MEIs) (13 training MEIs) (11 training MEIs)
Parent Awareness
Sadness .94*** .73* .84**
Anger .93*** .41 .93***
Parent Acceptance
Rater 1 Sadness .91*** .71* .67*
(15 training Anger .94*** .69* .79**
MEIs) Parent Regulation
Sadness .96*** .97*** .86**
Anger .97*** .91*** .82**
Child Awareness
Sadness .68* .75* .74*
Anger .89*** .42 .46
Child Acceptance
Sadness .89*** † .90***
Anger .83*** .73* .88***
Child Regulation
Sadness .97*** .42 †
Anger .85*** .76** .69*
Child Coaching
Sadness .97*** .54 †
Anger .90*** .74* .34
Note. † = Intraclass Correlation Coefficient could not be computed because the scale had zero variance items.
* p < .05, **p < .01, ***p < .001.

95
 APPENDIX B

Inter-Rater Reliability for Coding of Meta-Emotion Interviews (MEIs) for Present Study,
as Measured by Intra-Class Correlation Coefficients

MEI Dimension Rater 2 Rater 3

(5 MEIs) (5 MEIs)
Parent Awareness
Sadness .77 .13
Anger .93* .92*
Parent Acceptance
Rater 1 Sadness .72 .78
Anger .75 .86
Parent Regulation
Sadness .85* †
Anger .91* .54
Child Awareness
Sadness .91* .94*
Anger .73 .96*
Child Acceptance
Sadness .94** .92*
Anger .67 .91*
Child Regulation
Sadness 1.00*** .87
Anger .97** †
Child Coaching
Sadness .30 †
Anger .97** .86
Note. † = Intraclass Correlation Coefficient could not be computed because the scale had
zero variance items.
* p < .05, **p < .01, ***p < .001.

96
 APPENDIX C

Comparison of Scores on Meta-Emotion Interview Dimensions Between High And Low

Emotion Coaching Groups, Based on Emotion-Related Parenting Styles Self-Test –
Likert (ERPSST-L) Scores

High in Emotion Low in Emotion t-test Result

Coaching Coaching (Two-tailed)
(n = 6) (n = 4)
MEI Dimension M(SD) M(SD)

Parent Awareness
Sadness 43.11(3.63) 41.46(4.74) t(8) = -.63, p = .548
Anger 44.61(4.19) 42.75(3.59) t(8) = -.74, p = .490
Combined 87.72(7.71) 84.21(7.65) t(8) = -.71, p = .499
Parent Acceptance
Sadness 55.76(6.90) 56.76(6.19) t(8) = .23, p = .822
Anger 56.17(7.17) 56.12(5.46) t(8) = -.01, p = .991
Combined 111.93(11.70) 112.88(6.25) t(8) = .15, p = .887
Parent Regulation
Sadness 41.38(6.57) 41.23(2.55) t(8) = -.04, p = .968
Anger 39.30(7.16) 41.48(6.70) t(8) = .48, p = .642
Combined 80.67(10.61) 82.71(8.29) t(8) = .32, p = .756
Child Awareness
Sadness 33.79(4.13) 34.60(2.13) t(8) = .36, p = .731
Anger 34.02(2.58) 32.28(3.18) t(8) = -.96, p = .368
Combined 67.81(6.50) 66.88(5.09) t(8) = -.24, p = .817
Child Acceptance
Sadness 46.51(4.28) 46.08(5.71) t(8) = -.14, p = .894
Anger 40.64(2.18) 39.31(8.82) t(8) = -.30, p = .785
Combined 87.15(4.58) 85.39(13.76) t(8) = -.25, p = .819
Child Regulation
Sadness 32.24(4.78) 33.00(4.24) t(8) = .26, p = .804
Anger 27.87(7.19) 24.38(6.97) t(8) = -.76, p = .468
Combined 60.11(10.68) 57.38(10.70) t(8) = -.40, p = .702
Child Coaching
Sadness 42.67(1.60) 41.49(1.97) t(8) = -1.05, p = .326
Anger 35.12(4.33) 36.55(5.30) t(8) = .47, p = .651
Combined 77.79(5.43) 78.04(6.83) t(8) = .07, p = .950
Note. High emotion coaching group = cases (n = 6) with an ERPSST-L emotion coaching
score 1SD or greater above the mean (i.e., a score equal to or higher than 98.99). Low
emotion coaching group = cases (n = 4) with an ERPSST-L emotion coaching score 1SD
or greater below the mean (i.e., a score equal to or lower than 77.17).

97
 APPENDIX D

Comparison of Scores on Meta-Emotion Interview Dimensions Between High and Low

Emotion Coaching Groups, Based on Emotion-Related Parenting Styles (ERPS) Scores

High in Emotion Low in Emotion t-test Result

Coaching Coaching (Two-tailed)
(n = 8) (n = 4)
MEI Dimension M(SD) M(SD)

Parent Awareness
Sadness 43.92(3.53) 41.75(5.06) t(8) = -.88, p = .402
Anger 44.93(3.59) 42.36(3.15) t(10) = -1.21, p = .254
Combined 88.86(6.91) 84.11(7.59) t(10) = -1.09, p = .302
Parent Acceptance
Sadness 57.74(6.89) 53.62(4.49) t(10) = -1.07, p = .308
Anger 56.87(6.45) 53.89(5.86) t(10) = -.78, p = .456
Combined 114.61(11.26) 107.51(3.49) t(10) = -1.21, p = .250
Parent Regulation
Sadness 40.97(5.62) 38.46(1.77) t(10) = -.86, p = .412
Anger 40.11(6.89) 40.23(5.89) t(10) = .031, p = .976
Combined 81.08(9.59) 78.69(6.37) t(10) = -.45, p = .665
Child Awareness
Sadness 34.34(3.64) 34.03(2.32) t(10) = -.15, p = .883
Anger 34.39(2.30) 33.13(2.95) t(10) = -.46, p = .430
Combined 68.73(5.76) 67.16(5.00) t(10) = -.46, p = .653
Child Acceptance
Sadness 46.96(4.21) 46.34(5.91) t(10) = -.21, p = .837
Anger 41.67(4.69) 38.63(8.50) t(10) = -.82, p = .434
Combined 88.63(7.53) 84.97(13.54) t(10) = -.61, p = .553
Child Regulation
Sadness 33.18(4.40) 29.63(2.56) t(10) = -1.47, p = .171
Anger 28.64(6.80) 23.52(5.64) t(10) = -1.29, p = .225
Combined 61.82(9.94) 53.14(5.94) t(10) = -1.59, p = .144
Child Coaching
Sadness 43.00(1.49) 42.27(2.24) t(10) = -.68, p = .512
Anger 36.60(4.58) 38.94(4.46) t(10) = .84, p = .420
Combined 79.60(5.69) 81.22(6.35) t(10) = .45, p = .665
Note. High emotion coaching group = cases (n = 6) with an ERPS emotion coaching
score 1SD or greater above the mean (i.e., a score equal to or higher than 23.52).
Low emotion coaching group = cases (n = 4) with an ERPS emotion coaching score 1SD
or greater below the mean (i.e., a score equal to or lower than 18.52).

98
 VITA AUCTORIS

NAME Shawna Alysia Scott

PLACE OF BIRTH Windsor, Ontario

YEAR OF BIRTH 1986

EDUCATION Cardinal Carter Secondary School, Leamington, Ontario

2000-2004

St. Clair College of Applied Arts and Technology, Windsor

ON

2005 – 2007, Diploma of Early Childhood Education

University of Windsor, Windsor, ON

2007 – 2010, B.A. (Hons.)

University of Windsor, Windsor, ON

2010 – 2012, M.A.

99

View publication stats