PERSONNEL PSYCHOLOGY

1991.44

CULTURAL RELATIVITY IN ACTION: A COMPARISON

OF SELF-RATINGS MADE BY CHINESE AND U.S.
WORKERS

JIING-LIH FARH
Department of Management
Louisiana State University
GREGORY H. DOBBINS
Department of Management
University of Tennessee
BOR-SHIUAN CHENG
Fu-Hsin-Kong University

Using a sample of 982 leader-subordinate dyads drawn from nine dif-

ferent organizations in the Republic of China (i.e., Taiwan), self-ratings
of performance were compared with supervisory ratings of subordinate
performance. Results indicated that Chinese employees rated their job
performance less favorably than did their supervisors (i.e., they exhib-
ited a modesty bias). This modesty bias occurred relatively uniformly
across gender, various educational levels, and age groups. These re-
sults are contrary to the typically reported U.S. finding that self-ratings
of performance are more lenient than are supervisory ratings. A fur-
ther comparison of the means of supervisory and self-ratings between
this study and previous U.S. research revealed that the modesty bias
appeared to be produced by the lower self-ratings made by Chinese
workers as compared to their U.S. counterparts. The findings suggest
that culture plays a critical role in shaping workers’ perceptions of their
own work performance. Results of this study are discussed in terms of
their implications for future research and practice in international hu-
man resource management.

Self-ratings have the potential to increase the effectiveness of ap-

praisal systems, resulting in higher levels of appraisal satisfaction (Farh,
Werbel, & Bedeian, 1988) and perceptions of procedural justice and fair-
ness (Farh et al., 1988; Folger & Greenberg, 1985). However, research
indicates that the correlation between self- and supervisor ratings is par-
ticularly weak (Harris & Schaubroeck, 1988). In addition, subordinates

We would like to express appreciation to Philip M. Podsakoff for making his data
available for comparative analysis. We would also like to thank Michae! Campion, K.K.
Hwang, Kuo-Shu Yang, and three anonymous reviewers for their valuable comments on
an earlier version of this paper.
Correspondence and requests for reprints should be addressed to Jiing-Lih Farh, De-
partment of Management, Louisiana State University, Baton Rouge, LA 70803.

COPYRIGHT 01991 PERSONNEL PSYCHOLOGY, INC

129
130 PERSONNELPSYCHOLOGY

tend to evaluate themselves much more favorably than do their super-

visors and this effect occurs across different types of employees (e.g.,
clerical, managerial, blue collar), different types of rating scales (Thorn-
ton, 1980), and different rating purposes (e.g., Farh & Werbel, 1986).
Harris and Schaubroeck (1988) conducted a meta-analysis on 36 sets
of independent samples of self- and supervisor ratings and found that
self-ratings were over one-half standard deviation higher than supervi-
sor ratings. The finding that self-ratings are substantially higher than
supervisor ratings (i.e., the leniency bias) has led some researchers to
express reservations about using self-appraisals as a source of informa-
tion in performance evaluation (cf. Thornton, 1980).
The belief that self-ratings are lenient has almost become a “received
doctrine” in the field of human resource management. Barrett (1972)
has described received doctrines as ideas and assumptions that have be-
come so ingrained that they are no longer questioned. If the assump-
tions are correct, then received doctrines are valuable in that they allow
research to proceed and clear prescriptions to be formed. The danger, of
course, is that the assumptions are not correct, leading to the dual prob-
lem of misleading research and misguiding practice. The questioning
of the hypothesis of situational specificity of test validity (e.g., Schmidt
& Hunter, 198l), long unchallenged by academicians and supported by
personnel practice, is an apt illustration of the potential danger of a
falsely received doctrine.
The present study was conducted to determine whether the received
doctrine of leniency in self-ratings is supported outside the bounds of
Western culture. Such an investigation is critical since the world econ-
omy has become more global in the past 10 years and many companies
are multinational with extensive overseas operations (Daniels & Rade-
baug, 1989). Moreover, the U.S. Department of Labor (1988) predicts
that individualsof nonwestern descent will make up a substantial amount
of the U.S. labor force by the year 2000. Thus, it is important that re-
searchers determine the extent to which the leniency bias in self-ratings,
which is prevalent in the U.S., generalizes to other cultures. The present
study focuses on workers in the Republic of China (i.e., Taiwan) because
they are a part of the rapidly expanding Pacific Basin and live in a collec-
tivist culture that is dramatically different from the individualisticculture
of the U.S. (Hofstede, 1984).

Self-Ratings of Chinese Employees

The leniency effects revealed with U.S. samples are consistent with
the notion that individuals are motivated to view themselves as posi-
tively as possible, even if this means systematically inflating self-ratings
 JIING-LIH FARH ET AL,. 131

(e.g., Farh & Dobbins, 1989; Greenwald, 1980). This notion is deeply
rooted in the tradition of Western individualism that stresses individual
achievement, self-sufficiency,and self-respect. While exaggerated self-
perceptions may be accommodated and even reinforced in individualis-
tic cultures, inflated self-perceptions are inconsistent with collectivistic
cultures that encourage interpersonal harmony, interdependence, soli-
darity, and group cohesion. In a collectivisticculture, individual achieve-
ment is often deemphasized and sometimes even suppressed in the in-
terest of interpersonal harmony and group cohesion.
Two studies have demonstrated that workers in the Republic of China
are less individualisticthan are U.S. workers. Hofstede (1984) assessed
individualism/collectivism in over 17,000 individuals from 66 countries.
He found that employees in the Republic of China were some of the
most collectivistic and workers from the U.S. were the most individual-
istic. Similarly, Meindl, Hunt, and Lee (1989) administered Hui’s (1984)
individualism-collectivism measure to a total of 742 individuals from the
People’s Republic of China, the United States, the Republic of China,
the Republic of Korea, and Hong Kong. They found that individuals
in Hong Kong, the Republic of Korea, and the Republic of China were
all more collectivistic than were individuals in the U.S. These findings
strongly demonstrate that culture is more individualisticin the U.S. than
in the Republic of China.
In collectivistic cultures, individuals are discouraged from boasting
about individual accomplishments. Yang and Chiu (1987) observed that
a person of good character in Chinese culture is expected to be less oc-
cupied with self than with the group (Le., selfless), exercise modesty in
describing one’s achievements, and constantly strive to maintain self-
control. Shenkar and Ronen (1987) found that Chinese managers view
personal recognition as less important than other work goals, such as au-
tonomy, training, and good working relationships with coworkers. This
finding is consistent with Chinese tradition, which is opposed to individ-
ual glorification and views individual aggrandizement as a threat to the
collective good of society (e.g., Pye, 1982).
The deeply held collectivistic values of Chinese culture are thought
to exert strong influence on the self-report behavior of Chinese people
(Yang & Chiu, 1987). In the context of self-evaluation of performance,
Chinese are expected to be modest in self-ratings of performance and
refrain from exaggerating their achievements. While direct research on
this hypothesis is lacking, cross-cultural research has provided some in-
direct support for it. For example, Chang (1985) compared the attitudes
of Chinese and American managers working in U.S. subsidiaries in the
Republic of China, He found that Chinese managers reported lower per-
sonal satisfaction from doing theirjobs well than did U.S. managers. This
132 PERSONNEL PSYCHOLOGY

finding is particularly intriguing because the two groups of managers had

a similar level of overall satisfaction with their jobs. In interpreting these
results, Chang stated:

In American society, with its heavy emphasis on the individual, one

tends to attributesuccess to one’s own talent and effort. In the Chinese tra-
dition, however, individuals are expected to give credit not only to them-
selves but also to their family, colleagues, or even the whole society for
“personal”success. To the Chinese managers, “personal satisfaction”may
have smacked of selfishness or an unbecoming lack of modesty (p. 148).

Purpose of the Present Study

The above findings have important implications for self-rating re-

search. As noted earlier, leniency in self-ratings has become a received
doctrine within human resource management. However, the above re-
search suggeststhat Chinese employees may not inflate their self-ratings,
but instead may be “modest.” Furthermore, modesty should be most
pronounced on ratings of actual job performance because ratings on
these dimensions represent individual achievement and inflation of them
would be perceived as boasting or self-glorification. Thus, substantial
modesty would be expected on dimensions such as quantity or quality
of work. Less modesty would be expected on dimensions that are less
evaluative and less directly related to job performance (e.g., organiza-
tional citizenship behavior) because inflation of these dimensions less
severely violates collectivistic norms. The first purpose of the present
study is to determine whether Chinese employees evaluate themselves
more favorably on performance dimensions than they are evaluated by
their supervisors.
The present study will also determine whether age, gender, and ed-
ucation affect self-ratings made by Chinese employees. A fairly large
body of research indicates that Chinese with more formal education and
more exposure to mass media tend to have “modern” or Western atti-
tudes. These attitudes are more consistent with individualism (Hwang,
1967). For example, Hwang and Yang (1972) found that Chinese col-
lege students with higher scores on a “Modern Attitude Scale” were less
likely to seek others’ opinions before making decisions about important
matters. In essence, highly educated and young Chinese may be more
self-centered than are less educated and older Chinese and, thus, more
inclined to inflate self-ratings. The second purpose of the study is to test
this prediction.
The third purpose of the present study is to directly compare the self-
ratings made by U.S. and Chinese workers. If the modesty hypothesis
 JIING-LIH FARH ET AL. 133

is correct, we would expect Chinese self-raters to evaluate their perfor-

mance lower than U.S. workers who evaluate their own performance on
similar rating scales under similar rating conditions.
Two studies were conducted to examine these issues. Study I used
a questionnaire constructed in Chinese and Study I1 used a U.S. instru-
ment that was translated into Chinese. Study I1 was necessary in order
to insure that the findings of Study I were not unique to the instrument
used in Study I.

Method

Study I

Sample. The data for Study I were collected in the context of a large-
scale leadership study (see Cheng, 1985, for a detailed description of the
study). Over 900 leader-subordinate dyads were drawn from eight differ-
ent organizations in the Republic of China (Le., Taiwan), including two
divisions of a large government-owned oil company, a hospital, an auto-
mobile manufacturer, a nonprofit hotel, several advertising agencies, a
fast-food chain, and a shoe manufacturer. Usable questionnaires were
obtained from 793 leader-subordinate dyads, resulting in an average re-
sponse rate of 85%. With the exception of the hospital sample, which
consisted of nurses and was predominantly female (99%), 63% of the
remaining sample was male, 49% had a college or vocational education,
and the average age was 32 years. The nonhospital samples primarily
consisted of clerical, managerial, engineering, and sales employees; blue
collar workers were not represented.
Procedure. The survey in which the rating scale (described below) was
imbedded was completed on company time by subordinates and then by
their supervisors in separate group sessions. In each session, respon-
dents were informed by the researchers that the study was conducted
for research purposes only and complete confidentialitywas guaranteed.
After respondents completed the survey, they returned them directly to
the researchers. By using such a procedure, the researchers insured that
both supervisors and subordinates would not be able to see each other’s
ratings.
Measures. Supervisory ratings of subordinate work behaviors were
made on four dimensions: (a) Understanding of Work Duties, (b) Work
Skill, (c) Desire to Work, and (d) Job Performance. Each of the four
dimensions was assessed with three 5-point Likert rating items. Because
these items were originally written in Chinese, an English translation of
the rating items is presented in the Appendix. As can be seen, two of
the dimensions (Job Performance and Desire to Work) asked questions
134 PERSONNEL PSYCHOLOGY

about actual job behavior, while the other two dimensions (Understand-
ing of Work Duties and Work Skill) assessed prerequisites of job perfor-
mance. We expected more modesty on the two performance dimensions
than on the two nonperformance dimensions. The Cronbach alphas for
these dimensions were .80, .82, .82, and .89, respectively. Self-ratings of
work behavior were obtained by asking subordinates to rate themselves
on identical rating scales. The Cronbach alphas for self-ratings on the
four dimensions were .75, .77, .72, and .83, respectively.

Study II

Sample. The sample for Study I1 consisted of employees drawn from

three major divisions of the Ministry of Communications in the Repub-
lic of China (i.e., Taiwan). Questionnaires were distributed to a total of
250 subordinates and their supervisors. Respondents were drawn from
all occupational groups in the organization, including managers, profes-
sional and technical workers, engineers, clerical, and blue-collar work-
ers. Usable questionnaires were obtained from 188 leader-subordinate
dyads, for a response rate of about 75%.
Procedure. The procedure in which the data were collected was very
similar to that of the first study. The only difference was that about 30%
of the surveys were returned directly to the researchers through mail or
via coordinators. The other 70% were collected by the researchers in
small groups using procedures identical to those in Study I.
Measures. Supervisors were asked to rate the performance and the
citizenship behaviors of each subordinate. Subordinate performance was
rated by four items: (a) This individual is one of the best employees
that we have working for us; (b) This individual always completes hisher
work on time; (c) The quality of this individual’s work is excellent; and
(d) This individual is one of the most productive employees. Each item
was rated on a 7-point Likert scale anchored by 1 = strongly disagree, 4
= neither agree nor disagree, and 7 = strongly agree. Subordinate or-
ganizational citizenship behaviors were assessed with the 16-item scale
developed by Smith, Organ, and Near (1983). Consistent with Williams,
Podsakoff, and Huber (1986), the citizenship behavior scale was broken
down into three dimensions: (a) Altruism, (b) Attendance, and (c) Im-
personal Conscientiousness. The Cronbach alphas for these dimensions
were .90, .81, and .73, respectively.
Self-ratings of subordinate behavior were obtained by having sub-
ordinates rate their own performance and citizenship behaviors using
identical rating scales. Coefficient alphas for the three citizenship di-
mensions (Altruism, Attendance, and Impersonal Conscientiousness)
were .82, .74., and .79, respectively. All responses were confidential and
 JIING-LIH FARH ET AL. 135

used only for research purposes. Self-raters were unaware that their re-
sponses would be compared to the ratings made by their supervisors.
The research instrument was originally written in English. The trans-
lation of the original English language version of the questionnaire into
Chinese was accomplished through a multi-stage, translation-back trans-
lation procedure. This procedure was similar to that described by Brislin
(1980). In the first stage, a paid translator translated the English version
of the questionnaire into Chinese. The professionally translated version
was then translated back into English by the first author, and then back
into Chinese and back into English once again. This process was used to
help insure an accurate prose translation (Werner & Campbell, 1970).

Results

Study I

Table 1 presents the means and standard deviations of self- and su-
pervisor ratings. A 2 x 8 (Rater x Organization) Analysis of Variance was
conducted on each of the four rating measures with Rater (supervisor
vs. subordinate) treated as a within-subject variable and Organization
treated as a between-subject variable. The results of these analyses are
presented in Table 2. As can be seen, Job Performance and Desire to
Work were affected by the Rater and Organization main effects', but not
by their interaction. Consistent with our expectations, self-raters evalu-
ated themselves lower than did supervisors on both of the dimensions
( M = 3.88 vs. A4 = 3.97 for Desire to Work; A4 = 3.57 vs. M = 3.76
for Job Performance). Omega-squared values were .068 for Job Perfor-
mance and .014 for Desire to Work. Thus, contrary to typical findings
reported with U.S. workers, Chinese workers evaluated themselves less
favorably than did their supervisors (i.e., they exhibited a modesty bias).
Understanding of Work Duties and Work Skill were affected by the
Organization main effect and the Rater x Organization interaction (see
Table 2). Thus, for these two dimensions, the data were analyzed sep-
arately for each organization- Paired t-tests indicated that self- and su-
pervisor raters differed in ratings of Work Skill in Sample 2 [t(272) =
-3.15, p < .01] and Sample 3 [t(110) = -2.17, p < .05]. As can be seen
in Table 1, self-ratings were lower than were supervisor ratings in both
of these samples. Self- and supervisor ratings differed in Understanding
Work Duties in Sample 1 [t(272) = 2.98, p < .01], Sample 2 [t(221) =
-3.77, p < .01], and Sample 6 [t(34) = 2.60, p < .05]. Self-ratings were

'While differences between organizations were revealed, they are not discussed because
they are tangential to the purpose of the present study.
136 PERSONNEL PSYCHOLOGY

TABLE 1
Means and Standard Deviations for Sev- and Supervisoty Ratings (Study I )

Self-ratings Sup. ratings Mean

Rating dimensions M SD M SD difference
Sample 1
(Hospital, N=274)
Work duties 4.25 .57 4.12 .55 .13
Work skill 3.93 .51 3.93 .54 .oo
Desire to work 3.82 .56 3.82 .60 -.01
Job performance 3.43 .47 3.62 .61 -.18
Sample 2
(oil company-division, N =223)
Work duties 4.32 .65 4.52 .54 - .20
Work skill 4.02 .63 4.26 .61 - .25
Desire to work 3.92 .67 4.10 .61 -.18
Job performance 3.70 .61 3.94 .70 - .25

Sample 3
(Oil company-headquarters, N = 112)
Work duties 4.39 .56 4.48 .48 - .09
Work skill 4.11 .61 4.26 .54 -.15
Desire to work 4.02 .63 4.23 .56 -.21
Job performance 3.81 .52 3.97 .56 -.16
Sample 4
(Automobile manufacturer, N=41)
Work duties 3.98 .75 4.07 .58 -.09
Work skill 3.81 .57 3.93 .56 -.12
Desire to work 3.75 .64 3.78 .59 -.03
Job performance 3.37 .56 3.50 .58 -.14
Sample 5
(Nonprofit hotel, N=39)
Work duties 4.14 .53 4.33 SO -.20
Work skill 3.77 .43 3.77 .43 .oo
Desire to work 3.87 .61 3.09 .60 -.21
Job performance 3.37 SO 3.63 .55 -.27
Sample 6
(Shoe manufacturer, N=36)
Work duties 4.08 .73 3.77 .39 .32
Work skill 3.49 .77 3.64 .45 -.15
Desire to work 3.53 .68 3.47 .48 .06
Job performance 3.56 .47 3.49 .5 1 .07
Sample 7
(Fast-food chain, N=36)
Work duties 4.26 .65 4.19 .53 .07
Work skill 4.01 .59 3.99 .56 .02
Desire to work 4.03 .55 4.06 .57 -.03
Job performance 3.80 .64 4.00 .54 -.20
-
Sam-
(Advertising agencies, 1V=33)
Work duties 4.03 .63 4.20 .55 -.17
Work skill 3.61 .68 3.83 .65 -.22
Desire to work 3.93 .58 4.03 .56 -.lo
Job performance 3.29 .58 3.54 .61 -.24
 JIING-LIH FARH ET AL. 137

TABLE 2
Results ofANOVAs on Ratings (Study I )
Dependent measure DF MS F

Understanding of work duties

Within subjects
Rater 1 .I7 .64
Rater X Organization 7 1.48 5.51**
Error within 785 .27
Between subjects
Organization 7 5.01 12.94**
Error between 785 .39
Work skill
Within subjects
Rater 1 2.47 9.82**
Rater x Organization 7 .65 2.59;
Error within 785 .25
Between subjects
Organization 7 6.13 15.35**
Error between 785 .40
Desire to work
Within subjects
Rater 1 1.68 5.35;
Rater X Organization 7 0.50 1.59
Error within 785 0.31
Between subjects
Organization 7 5.05 12.36**
Error between 785 .41
Job performance
Within subjects
Rater 1 6.13 25.56**
Rater x Organization 7 .25 1.06
Error within 785 .24
Between subjects
Organization 7 6.94 16.05* *
Error between 785 .43
~~~ ~

*p<.05; **p<.Ol

higher than were supervisor ratings in two of these three samples (see
Table 1).
In summary, the findings of Study I provide strong evidence that
Chinese self-raters evaluate their job performance lower than do their
supervisors. The modesty effect was revealed on both of the measures of
actual job performance (Desire to Work and Job Performance). On the
dimensions that were prerequisites for job performance (Understanding
of Work Duties and Work Skill), the pattern of results was less clear,
although some modesty bias was revealed in ratings of Work Skill.
138 PERSONNEL PSYCHOLOGY

Study 11

Table 3 presents the means, standard deviations, and results of paired

t-tests for performance and citizenship behavior ratings. Results indi-
cated that self-ratings of performance were significantly lower than su-
pervisory ratings of subordinate performance on each of the four job
performance items. Omega-squared values for these effects ranged from
.054 to .299.
Paired t-tests were also conducted on the citizenship dimensions. As
can be seen in Table 3, self-ratings were lower than were supervisory rat-
ings for the Altruism and Attendance dimensions, although both were
nonsignificant. However, for the Impersonal Conscientiousness dimen-
sion, self-ratings were found to be significantly higher (i.e., more le-
nient) than supervisory ratings. Thus, while modesty was revealed in
self-ratings of performance in Study 11, leniency was detected in ratings
of Impersonal Conscientiousness.

Effects of Education, Age, and Gender

Hierarchical regression analyses were conducted to determine

whether self-ratings could be predicted by subordinate age, subordinate
education, or subordinate gender, after controlling for supervisor rat-
ings. Self-ratingswere first regressed onto supervisor ratings. In the next
step, education, age, and genderwere added to the equation. The hierar-
chical regression procedures indicated that the addition of demographic
variables added significantly to the prediction of self-ratings above and
beyond supervisor ratings in Study I (Incremental R2 = .045, p < .Ol).
As can be seen in Table 4, this increased prediction occurred because age
was a significant predictor of self-ratings. Specifically, and contrary to
our expectations, younger workers evaluated themselves as less effective
than older workers. In Study 11, the addition of demographic variables
did not add significantly to the prediction of self-ratings.

Comparison of Pe~ormanceRatings Between US.and Chinese Samples

One potential explanation for the divergent finding between previous

U.S. studies and this study is that Chinese supervisors might have rated
their subordinates more leniently than did American supervisors. To ex-
amine this possibility, we compared the supervisory ratings coltected in
Study I1 for citizenship behavior with ratings obtained from US. super-
visors using the same rating scales. In both settings, ratings were made
for research purposes. The results are shown in Table 5. As can be seen,
 JIING-LIH FARH ET AL. 139

TABLE 3
Results of Paired t-tests, Means, and Standard Deviations
of Self- and Supervisor Ratings (Study ZZ, N =188)
Self-ratings Sup. ratings Mean t
Rating dimensions M SD A4 SD difference values
Performance
One of the best employees 4.22 1.39 5.47 1.32 -1.26 -9.01**
Complete work on time 5.51 1.03 5.85 1.05 -.34 -3.44**
Work of excellent quality 5.00 1.06 5.61 1.11 -.61 -5.86**
Most productive employee 4.71 1.20 5.25 1.34 -.54 -4.14**
Citizenship behavior
Altruism (6 items) 30.73 5.04 31.36 6.07 -.64 -1.13
Attendance (4 items) 22.64 3.54 23.17 3.71 -.52 -1.54
Impersonal conscientiousness 27.20 4.68 25.76 4.91 1.44 3.17**
(5 items)
*p<.05; **p<.O1

TABLE 4
Results of Multiple Regression Analyses Using Supervisor Ratings,
Subordinate Education, Subordinate Gender; and Subordinate Age
as Predictors of Self-Ratings of Performanceab

Supervisor
ratings Education Gender Age RZ
Study 1 .316** -.011 .011 .216** .165**
(N=792) (.346**) (-,036) (-.124**) (.261**)
Study 2 .096 .079 - .046 .082 .026
(N=171) (.120) (.075) (-.055) (.071)
* p < . 0 5 (two-tailed test): **p<.O1 (two-tailed test)
Study I, only the Job Performance scale was used. In Study 11, the four items were
summed to form a single measure of job performance.
bStandardized regression coefficients are shown in the table. Simple correlations are
shown below the regression coefficients in parentheses.

the mean ratings by Chinese supervisors on the three dimensions of cit-

izenship behavior were very similar to the mean ratings made by U.S.
supervisors and were not statistically different. Table 5 also presents a
comparison of performance ratings between Chinese and U.S. supervi-
sors on three of the four performance rating items (the fourth perfor-
mance measure was not used with the U.S. sample). Here again, ratings
were made for research purposes. The results indicated that in com-
parison with U.S. supervisors, Chinese supervisors rated their subordi-
nates significantlyhigher on completing work on time, significantlylower
140 PERSONNEL PSYCHOLOGY

TABLE 5
Comparison between Chinese and American Supervisors on
Ratings of Subordinate Pelformance Using Identical Rating Scalesa

Chinese (N=1881 American (N=453) Mean t

Citizenship behavior M SD M SD difference values
Altruism 31.36 6.07 31.23 7.11 .13 .23
Attendance 23.17 3.71 22.54 4.82 .63 1.78
Impersonal conscien. 25.76 4.91 25.39 6.35 .37 .79
American
Chinese (N=188) Sample 2 (N=932) Mean t
Performance M SD M SD difference values
Complete work on time 5.85 1.05 5.11 1.51 .74 8.12**
Work excellent quality 5.61 1.11 5.59 1.30 .02 .22
Most productive empl. 5.25 1.34 5.49 1.46 -.24 -2.20*
*p<.05 (two-tailed test); **p<.O1 (two-tailed test)
asample consisted of employees of nursing home facilities as reported by Williams,
Podsakoff, and Huber (1986). Sample 2 consisted of employees of an oil company; this data
set was provided to the authors through the courtesy of Philip M. Podsakoff. A description
of Sample 2 characteristics may be found in Podsakoff, MacKenzie, Moorman, and Fetter
(1990).

on most productive employee, and about the same on quality of perfor-

mance. These findings suggest that Chinese supervisors did not evaluate
behavior more leniently than did their U.S. counterparts.
We were also able to identify two studies that had U.S. supervisors
make performance ratings on 5-point scales similar to those used in
Study I. Ferris, Yates, Gilmore, and Rowland (1985) found a mean rating
of 3.93 and Dreher (1981) obtained a mean rating of 3.5. These values
are very similar to the performance ratings given by Chinese supervisors
in Study I (A4= 3.76), once again suggesting that Chinese supervisors’
ratings were not higher than were the ratings made by U.S. supervisors.
We were also able to identify three U.S. studies which asked self-
raters to make evaluations on scales similar to those used in Study I or
Study 11. Once again, only self-ratings made for research purposes were
considered. Steel and Ovalle (1984) asked three samples of employees
to make self-ratings on the dimensions of Quality of Work and Quan-
tity of Work using 7-point scales similar to those used in Study 11. They
found mean self-ratings of 5.87, 6.42, 5.18, 5.17, 5.00, and 5.17. Simi-
larly, Holzbach (1978) found that the average self-rating of overall ef-
fectiveness on a 7-point scale was 6.25. These values are substantially
higher than the self-ratings made by Chinese employees in Study I1 (A4
= 4.86). Shore and Thornton (1986) asked employees to evaluate their
performance on a 5-point graphic rating scale and found a mean rating
of 3.93 for Quality of Work and 3.67 for Quantity of Work. Once again,
these values tend to be larger than the self-ratings of performance made
 JIING-LIH FARH ET AL. 141

by Chinese employees in Study I (A4= 3.57). These combined findings

suggest that the modesty revealed in the present study was probably due
to the low ratings given by self-raters, not by higher supervisor ratings.

Discussion

The findings provide strong evidence that Chinese employees rate

their own job performance lower than their supervisors evaluate their
performance. This modesty bias was found on both measures of job
performance in Study I and on all four of the job performance mea-
sures in Study 11. These findings are clearly inconsistent with the re-
ceived doctrine that self-ratings will be higher than supervisor ratings
and support the collectivistic culture hypothesis. The pattern of findings
demonstrates the importance of considering culture when proposing ma-
jor principles of human resource management.
The findings also support the hypothesis that modesty bias will be
stronger on ratings of actual job performance than on nonperformance
dimensions. For example, in Study I, modesty bias was found robustly
on both measures of job performance, but not on the nonperformance
measures. Similarly, in Study 11, while the modesty bias occurred on all
core performance rating items, such bias was not observed on any of
the citizenship behavior dimensions. In fact, leniency bias was found on
Impersonal Conscientiousness. A careful analysis of the citizenship be-
havior scale indicates that there are three negatively worded items on the
15-item scale and all of these items pertain to Impersonal Conscientious-
ness. A n item-by-item analysis showed that it is precisely on these three
items that self-ratings were found to be more lenient than supervisor
ratings. This finding is consistent with Yang and Chiu’s (1987) sugges-
tion that denying one’s possession of shortcomings (Le., disagreeing with
negatively worded items) may not invoke such a strong concern for im-
modest behavior. In other words, negatively worded items may be able
to bring out a more “truthful” answer from Chinese subjects-ne that
is less prone to modesty self-presentation concerns. Because item con-
tent and item wording are confounded in the citizenship behavior scale,
future research should explore this intriguing hypothesis with a more ap-
propriate research design.
Contrary to our expectation that younger and better educated Chi-
nese workers would be more likely to inflate self-ratings than would,older
and less educated Chinese workers, our results showed that education
was unrelated to self-ratings. Moreover, age was found to be positively
related with self-ratings in Study I, but not in Study 11.While these com-
bined results suggest that demographic variables play a relatively minor
142 PERSONNEL PSYCHOLOGY

role in determining self-ratings, future research needs to more carefully

examine the age variable due to its inconsistent effects across studies.
A large body of research has indicated that self-ratings are not highly
correlated with supervisor ratings. For example, Harris and Schaubroeck
(1988) conducted a meta-analysison 36 independent self-supervisorcor-
relations and found that the mean correlation was .29 when ratings were
made on global performance appraisal scales similar to the ones used
in the current study. Leniency is frequently cited as one cause for this
low interrater agreement (e.g., Thornton, 1980). Because Chinese self-
ratings were less lenient than U.S. self-ratings, one intriguing hypothesis
is that Chinese self-ratings should correlate more highly with supervisor
ratings than has typically been found with U.S. samples.
The meta-analysis procedures of Hunter, Schmidt, and Jackson
(1982) were employed to determine the average correlation between
Chinese self- and supervisor ratings for the nine samples in Studies I
and 11. This procedure revealed an average correlation of .25, which is
slightly lower than the .29 average correlation reported by Harris and
Schaubroeck (1988). These findings indicate that self-ratings have low
convergence with supervisor ratings in both the U.S. and China. Such
findings are not very surprising because leniency and accuracy are con-
ceptually distinct. While suppressing leniency changes the mean rating,
it does not necessarily increase the accuracy of evaluations. These find-
ings are consistent with other research that has shown the independence
of leniency and accuracy in supervisor ratings (e.g., Murphy & Balzer,
1989).

Implications and Future Research

Several implications of the findings should be considered. First, it

is clear that the use of self-ratings by multinational firms may be biased
against Chinese employees. Such employees may evaluate themselves
as less effective than equally performing US.employees and, thus, may
be unfairly discriminated against in any administrative decisions which
are based upon self-ratings. Second, and probably more important, the
findings suggest that Chinese employees may be more reluctant to en-
gage in impression management behaviors which promote their own ef-
fectiveness (e.g., making sure that superiors are aware of their accom-
plishments). Failure to engage in these behaviors to the same degree as
U.S. workers could result in Chinese employees being evaluated as less
effectiveby supervisors than are U.S. employees. Obviously, the findings
of the present study do not directly test this proposition, but it should be
examined in future research.
 JIING-LIH FARH ET AL. 143

Future research should also examine the effects of individual differ-

ences within a culture on modesty bias. While the present study found
that modesty bias occurred rather uniformly across gender, educational
levels, and age groups, these results should not be interpreted as indi-
cating that individual differences are not important. The demographic
variables examined in the present study were, at best, surrogate measures
for an individual’s individualism-collectivism. Future research should di-
rectly assess individualism-collectivism using a scale similar to the one
developed by Wagner and Moch (1986). We would expect that Chinese
employees would score lower (more collectivistic)on this measure than
would their U.S. counterparts and that there would be a significant rela-
tionship between scores on the individualism-collectivism measure and
modesty bias within cultures.

Limitations

Several limitations of the present study should be noted. First, as is

often true in field studies, the groups here represent convenience sam-
ples, which inevitably raises questions about external validity. Such con-
cern is less troublesome given the exploratory nature of the study and
the fact that the samples were drawn from nine different organizations.
Given the size of the sample and its diversity, generalization of the find-
ings to other samples in the Republic of China should not be problem-
atic.
A second issue of external validity is the extent to which the findings
generalize to employees of Chinese descent outside of the Republic of
China. Because Chinese culture is opposed to individual glorification,
the findings may generalize to Chinese employees in other countries.
Consistent with this statement, Shenkar and Ronen (1987) found that
individual recognition was rated substantially less important than other
work goals by Chinese employees in the People’s Republic of China,
Hong Kong, the Republic of China (i.e,, Taiwan), and Singapore. We
realize, however, that generalization is ultimately an empirical question
and, thus, encourage replication of the findings, particularly with Chi-
nese employees in Western organizations.
A third limitation is that all ratings were made for research pur-
poses. Research indicates that both self- and supervisor ratings made
by U.S. employees are affected by the purpose of evaluation (e.g., Dob-
bins, Cardy, & Trwillo, 1988; Farh & Werbel, 1986; Zedeck & Cascio,
1982). Thus, future research needs to determine whether the modesty
effect revealed in the present study would also occur when self- and su-
pervisor ratings are made for administrative decisions.
144 PERSONNEL PSYCHOLOGY

A fourth limitation is that alternative processes could explain the

modesty findings. One possibility is that Chinese employees have higher
standards than their U.S. counterparts, thus resulting in lower self-
ratings. While such an explanation is possible, it seems unlikely be-
cause Chinese self-raters did not evaluate themselves lower on nonper-
formance dimensions (e.g., citizenship behavior). A second explanation
is that the modesty bias occurs due to self-presentation. In other words,
Chinese workers may evaluate their performance as less effective than
they actually perceive it in compliance with social norms. While the
anonymity of the questionnaire should have eliminated some of the self-
presentation motive, future research is needed to clarify the processes
which underlie modesty in Chinese self-ratings.

REFERENCES

Barrett GV. (1972). Research models of the future for industrial and organizational psy-
chology. PERSONNEL PSYCHOLOGY, 25,l-17.
Brislin RW. (1980). Translation and content analysis of oral and written material. In Trian-
dis HC, Berry JW (Eds.), Handbook of cross-culturalpsychology:Vo12. Methodology
(pp. 349-444). Boston: Allyn & Bacon.
Chang SKC. (1985). American and Chinese managers in U.S. companies in Taiwan: A
comparison. California Management Review, 27, 144-156.
Cheng BS. (1985). Task-oriented leadership behavior and subordinate performance: A sup-
plementary model and its validation. Unpublished Doctoral Dissertation, National
Taiwan University, Taiwan (in Chinese).
Daniels JD, Radebaug LR. (1989). International business: Environments and operations.
Reading, M A Addison-Wesley Publishing.
Dobbins GH, Cardy RL, Truxillo D T (1988). The effects of purpose of appraisal and
individual differences in stereotypes of women on sex differences in performance
ratings: A laboratory and field study. Journal ofApplied Psychology, 73,551-558.
Dreher GF. (1981). Predicting the salary satisfaction of exempt employees. PERSONNEL
PSYCHOLOGY, 34,579-590.
Farh JL, Dobbins GH. (1989). Effects of self-esteem on leniency bias in self-reports of
performance: A structural equation model analysis. PERSONNEL PSYCHOLOGY, 42,
835-849.
Farh JL, Werbel JD. (1986). Effects of purpose of the appraisal and expectation of valida-
tion on self-appraisal leniency. Journal of Applied Psychology, 71,527-529.
Farh JL, Werbel JD, Bedeian AG. (1988). A n empirical investigation of self-appraisal-
based performance evaluation. PERSONNEL PSYCHOLOGY,41,141-156.
Ferris GH, Yates VL, Gilmore DC, Rowland KM. (1985). The influence of subordinate
age on performance ratings and causal attributions. PERSONNEL PSYCHOLOGY, 38,
545-557.
Folger R, Greenberg J. (1985). Procedural justice: An interpretive analysis of personnel
systems. In Rowland KR, Ferris G R (Eds.), Research in Personnel and Human
Resources Management (Vol3, pp. 141- 183). Greenwich, CT: JAI Press.
Greenwald AG. (1980). The totalitarian ego: Fabrication and revision of personal history.
American Psychologist, 35,603-618.
Harris MM, Schaubroeck J. (1988). A meta-analysis of self-supervisor, self-peer, and peer-
supervisor ratings. PERSONNEL PSYCHOLOGY, 41,43-62.
 JIING-LIH FARH ET AL. 145

Hofstede G . (1984). Culture’s consequences: International differences in work-related values.

Beverly Hills: Sage Publications.
Holzbach RL. (1978). Rater bias in performance ratings: Superior, self, and peer ratings.
Journal ofApplied Psychology, 63,579-588.
Hui CH. (1984). Development and validation of an individualism-collectivism scale. (Tech-
nical Report, ONR-31). Alexandria, VA. Office of Navy Research.
Hunter JE, Schmidt FL, Jackson GB. (1982). Meta-analysis: Cumulating research findings
across studies. Beverly Hills: Sage Publications.
Hwang CH. (1967). A study of the personal preference of Chinese university students by
Edwards Personal Preference Schedule. Psychology and Education, I , 52-67.
Hwang K, Yang K. (1972). Studies on individual modernity and social orientation. Bulletin
of the Institute of Ethnology, Academia Sinica, 32,245-278 (in Chinese).
Meindl JR, Hunt RG, Lee W. (1989). Individualism-collectivism and work values: Data
from the United States, China, Taiwan, Korea, and Hong Kong. In Ferris G , Row-
land K (Eds.), Research in Personnel and Human Resources Management (Supple-
ment 1, pp. 59-77). Greenwich, CT:JAI Press.
Murphy KR, Balzer WK. (1989). Rater errors and rating accuracy. Journal of Applied
Psychology, 71,39-44.
Podsakoff PM, MacKenzie SB, Moorman RH, Fetter R. (1990). The impact of transfor-
mational leader behaviors on employee trust, satisfaction, and organizational citi-
zenship behaviors. Leadership Quarterly, 1, 107-142.
Pye L. (1982). Chinese commercial negotiating style. Cambridge, MA: Oelgeschlager,
Gunn, & Hain.
Schmidt FL, Hunter JE. (1981). Employment testing: Old theories and new research.
American Psychologist, 36,1128-1 137.
Shenkar 0, Ronen S. (1987). Structure and importance of work goals among managers in
the People’s Republic of China. Academy of Management Journal, 30,564-576.
Shore LF, Thornton GC. (1986). Effects of gender on self- and supervisory ratings.
Academy of Management Journal, 29,115-129.
Smith.CA, Organ DW, Near JP. (1983). Organizational citizenship behavior: Its nature
and antecedents. Journal ofApplied Psychology, 68,655-663.
Steel RF‘, Ovalle NK. (1984). Self-appraisal based upon supervisory feedback. PERSONNEL
PSYCHOLOGY, 37,667-685.
Thornton GC. (1980). Psychometric properties of self-appraisals of job performance.
PERSONNEL PSYCHOLOGY, 33,263-271.
U.S. Department of Labor. (1988). Opportunity 2000. Washington, DC: Government
Documents.
Wagner JA, Moch MK. (1986). Individualism-collectivism: Concept and measure. Group
and Organization Studies, 11,280-304.
Werner 0, Campbell D. (1970). Translating, working through interpreters, and the prob-
lem of decentering. In Naroll R, Cohen C (Eds.), A handbook of methods of cultural
anthropology (pp. 398-420). New York: Natural History Press.
Williams LJ, Podsakoff RM, Huber V. (1986, August). Determinants of organizational cit-
izenship behaviors: A structural equation analysis with cross-validation. Paper pre-
sented at the Annual Meeting of the Academy of Management, Chicago, IL.
Yang CF, Chiu CY. (1987). The dilemmas experienced by Chinese subjects: A critical
review of their influence on the use of Western originated rating scales. Acta Psy-
chologica Taiwanica, 29,113-132 (in Chinese).
Zedeck S, Cascio WF. (1982). Performance appraisal decisions as a function of rater
training and purpose of appraisal. Journal ofApplied Psychology, 67,752-758.
146 PERSONNEL PSYCHOLOGY

Appendix

English Translation of Rating Scale Used in Study I

Understandingof Work Duties

a. Does he understand his work contents? In other words, does he

understand, on a daily basis, what he needs to carry out on his job
and what equipment and tools he needs to use? (1 =Very Inadequate,
5=Very Clear Understanding)
b. Does he understand his work objectives? In other words, how well
does he understand the assigned work goals and the work require-
ments? (1 =Very Inadequate, 5 =Very Clear Understanding)
c. Does he understand his job responsibilities? In other words, every
task has its responsibilities and requirements. If not properly com-
pleted, it may incur some costs and losses. How well does he un-
derstand his responsibilities? (1=Very Inadequate, 5=Very Clear
Understanding)

Work Skill

a. Does he have enough required work knowledge? In other words,

does he have sufficient technical know-how to carry out his job pro-
ficiently? (1 =Very Inadequate, 5 =Very Clear Understanding)
b. Does he understand the work methods? In other words, does he un-
derstand the steps, procedures, and methods required for carrying
out his job? (1=Little Understanding, 5=Very Clear Understand-
ing)
c. Is he proficient in his work skills? In other words, is he familiar with
the skills or techniques required on the job to perform effectively?
(1=Very Unfamiliar, 5=Very Familiar)

Desire to Work

a. Is he enthusiastic about his job? In other words, will he still be

enthusiastic about his present job if enthusiasm is not required by
company rules or regulations? (1 =Very Low, 5=Very High)
b. How persistent is he on his job? In other words, does he grow tired
and tardy on his present job? (l=Very Tired, 5=Never Feels Tired)
 JIING-LIH FARH ET AL. 147

c. How is his concentration level on his job? In other words, can he

concentrate on and give his best to his job? (1=Low Concentration,
5 =Very High Concentration)

Job Performance

a. What do you think of his quality of work? In other words, are his
work outcomes perfect, free of error, and of high accuracy? (1 =Very
Low Quality, 5 =Excellent Quality)
b. What do you think of his work efficiency? In other words, what is
your assessment of his work speed or quantity of work? (1=Very
Low Efficiency, 5 =Excellent Efficiency)
c. What do you think of his work performance? In other words, is he
able to complete quality work on time? (1 =Very Poor Performance,
5 =Excellent Performance)