52 Suicide and Life-Threatening Behavior 33(1) Spring 2003

 2003 The American Association of Suicidology

Development of a Questionnaire on Attitudes

Towards Suicide (ATTS) and Its Application
in a Swedish Population
ELLINOR SALANDER RENBERG, PHD, AND LARS JACOBSSON, MD, PHD

An instrument measuring attitudes toward suicide was developed through

two postal questionnaire surveys, 1986 and 1996, approaching general population
samples of 700 and 1,000 persons, respectively, and with response rates of 75%
and 64%. A 10-factor model was obtained and comparisons between waves re-
vealed an overall more permissive and pro-preventional attitude in 1996. The
results implied complex relationships at the aggregated level between attitudes
and suicidal behavior, depending on type of suicidal behavior, attitude factor, and
subgroup. At the individual level, persons answering affirmative on own suicidal
behavior, as compared to the nonsuicidal group, held a more permissive and un-
derstanding attitude toward suicide.

Suicide and suicidal behavior occur in a social on the normative evaluation of suicide in or-
and cultural context, reflected in the signifi- der to improve our understanding of the
cant variations in rates of suicide and attempted mechanisms influencing suicide and suicidal
suicide in different countries. These differ- behaviors at an aggregated as well as an indi-
ences cannot sufficiently be explained by var- vidual level (e.g., Canetto, 1992–93; Diek-
ying living conditions (e.g., unemployment, stra, 1989; Farber, 1968; Lester, 1972; Neele-
divorce rates, alcohol consumption), nor prev- man & Lewis, 1999; Platt, 1989; Schmidtke,
alence of psychiatric disorders (Neeleman & 1997).
Lewis, 1999). Many researchers have dis- A predominant view among research-
cussed the necessity of more explicit studies ers has been that the more permissive atti-
tudes toward suicide in a specific group, the
higher the suicide rate, and that repressive at-
ELLINOR S. RENBERG and LARS JACOB- titudes toward suicide may deter vulnerable
SSON are with the Department of Clinical Sci- individuals from suicidal behaviors (e.g., Farb-
ences, Division of Psychiatry and WHO Collabo- erow, 1989; Ramsay & Bagley, 1985). Results
rating Centre, Umeå University, Umeå, Sweden. from different studies on attitudes toward
This study was supported by grants from suicide, however, demonstrate no unequivo-
the Swedish Council for Planning and Coordinat-
ing of Research and from the Swedish Council for cal empirical evidence for such an associa-
Social Research. Special thanks to Birgitta Törnk- tion. In the European Values Study (EVS;
vist, Department of Statistics, Umeå University, Halman, 1993), comparisons between coun-
for her statistical expertise. tries’ rank order of suicide rates and permis-
Address correspondence to Ellinor Salander sive attitudes toward suicide revealed no as-
Renberg, Department of Clinical Sciences, Divi-
sion of Psychiatry, WHO Collaborating Cen- sociation. Even changes over time disclosed no
tre, Umeå University, SE-901 85 Umeå, Sweden; association, and while Sweden experienced a
E-mail: ellinor.salander.renberg@psychiat.umu.se. decrease in suicide rates between 1981 and
RENBERG AND JACOBSSON 53

1990, the permissiveness had increased. Most The study constituted part of a suicide pre-
studies on the association between suicide at- vention project where special interest was di-
tempt rates and permissive attitudes imply a rected toward attitudes concerning preven-
negative relationship (e.g., Kerkhof & Nata- tion of suicide.
wat, 1989; Platt, 1989; Sale, Williams, Clark,
& Mills, 1975). More hostile and intolerant
attitudes were found in areas with high rates STUDY AREA
of attempted suicide as compared to low-rate
areas. However, at an individual level, per- The study area consisted of the county
sons with a history of suicidal behavior held of Västerbotten in the north of Sweden, with
a more permissive attitude. A general prob- a total population of about 260,000 (Statistics
lem in this matter, first expressed by Gibbs Sweden, 1996). A detailed description of the
(1968), is the difficulty in assessing the direc- area, including information on suicide and
tion of the influence (cause-effect sequence); parasuicide rates, can be found elsewhere
that is, whether attitudes influence behavior (Salander Renberg, & Jacobsson, 1999; Sa-
or if behavior induces attitude changes. lander Renberg, 1999). The county is as an
A variety of methods and designs have appropriate area for conducting epidemio-
been applied to measure attitudes toward sui- logical studies since it includes quite different
cide, resulting in limited possibilities to com- human milieus with divergent suicide rates.
pare results from different studies. The meth- The highest suicide rates were found in the
ods have varied from open, semistructured most depopulated rural areas and in the most
interviews, often combined with simple se- expanding urban areas, whereas the lowest
mantic differential scales (Osgood, Suci, & rates were found in the remaining “mixed
Tannenbaum, 1957), to large-scale surveys areas.”
using questionnaires. The attitudes measured
have also varied, from one dimensional atti-
tudes toward suicide (e.g., favorable-unfavor- CONSTRUCTION OF THE
able), as was done in the EVS, to comprehen- ATTITUDE QUESTIONNAIRE
sive studies on different attitude components
(Domino, Moore, Westlake, & Gibson, 1982), The questionnaire measuring attitudes
also taking into consideration the role of who toward suicide (ATTS) was developed and
the target person is and who the evaluator is fielded on two different occasions, with ten
(Deluty, 1988–89; Diekstra & Kerkhof, 1989; years in between; consequently, the develop-
Marks, 1988–89; Stillion, McDowell, & Sham- ment of the questionnaire and data collection
blin, 1984). In a recent study by Jenner and were conducted simultaneously. Both ques-
Niesing (2001), an interesting combination tionnaire studies were approved by the ethi-
of the semantic differentiated scale method cal committee of the Medical Faculty of Umeå
and different actors is applied; however, the University. Statistical methods used in the
instrument is limited to measuring only two development of the instrument were factor
dimensions of attitudes towards suicide (health/ analysis using varimax rotation, chi-square
illness and acceptance/rejection). tests and F-tests ( p ≤ .05) for comparisons
The aim of the present study was to between groups; and for internal consistency
construct a feasible instrument to measure Cronbachs alpha was applied.
attitudes toward suicide in large-scale surveys
in the general population. Some of the goals First Wave
were to cover multidimensional aspects of at-
titudes toward suicide, and to describe atti- With influence from the Suicide Opin-
tude changes over time, as well as the associ- ion Questionnaire (SOQ; Domino et al.,
ation between attitudes and suicidal behavior. 1982), a first version of the ATTS was devel-
54 QUESTIONNAIRE ON ATTS

oped in the middle of the 1980s, using a 5- dence) a response rate of 75% was obtained.
point Likert scale (agree completely, agree to a Representativity of the answering group was
large extent, doubtful, do not agree and agree not ascertained by comparing the population, sam-
at all). Several clinicians with a broad experi- ple, and answering group concerning sex, age,
ence in suicidal problems and a group of lay- cohabiting status, educational length, and resi-
men were asked to check the questions in the dence, and is presented in detail elsewhere
pool and give suggestions on new questions. (Salander Renberg, 2001). The responders
With the help of their opinions, 80 questions matched well with the population on vari-
were chosen for a preliminary version of the ables such as age, educational length, and res-
questionnaire, which was piloted in a group idence, and also exhibited a response pattern
of students (N = 50). After exclusion of items common in postal surveys: a higher propor-
with extremely low discriminative power tion of females (56% vs 49%) and cohabiting
or shortcomings in formulation, 62 attitude persons (73% vs 63%).
questions remained, covering the three com- It was not possible to obtain an inter-
ponent areas instrumental, affective, and cog- pretable factor model by explorative factor
nitive. Included in the questionnaire were analysis for the original 62 items. Therefore
questions on basic demographic data: sex, age a reassessment of the included items was per-
(18–30, 31–50 and 51–65), educational length formed. The distribution for the obtained
(9 and under, 10–13 and 14+ years), cohabit- answers was calculated for each item and in
ing status, region (urban, rural, and mixed), cases of extremely skewed or high kurtosis
and experience of suicide and suicidal prob- (leptokurtic) distribution the items were ex-
lems among significant others. The question- cluded from further analysis as they were as-
naire constituted part of an instrument that sumed to have a too low discriminative power;
also measured suicidal behaviors, suicide items with very low correlation with other
attempts, plans, ideation, death wishes, and items were also excluded. Items were again
life-weariness using questions developed by scrutinized concerning formulation and re-
Paykel, Myers, Lindenthal, and Tanner (1974). analyzed concerning relevance for the aims
Results from this part have been presented of the research project. At this point it was
elsewhere (Salander Renberg, 2001). decided to emphasize and focus on preven-
Postal questionnaires were distributed tive aspects instead of cognitive aspects, lead-
to a randomly selected sample of 710 persons ing to the exclusion of items primarily con-
from the general population, aged 18 to 65 cerning causes to suicide. An explorative factor
years. The first study was actually conducted analysis was performed for a remaining num-
on two different occasions, in spring 1985 ber of 22 items, using varimax rotation, and
and spring 1987, but as there was no differ- after exclusion of two items with low com-
ence between the studies concerning distri- munality (<.40)—that is, the total variance of
bution of demographic variables, they were a variable accounted for by the combination
pooled together and labeled as the 1986 study of all common factors—an interpretable
(Salander Renberg, 2001). A letter of infor- eight factor model was obtained including 20
mation, in which the study was presented and items. The eight factors accounted for 63 %
a decoding procedure guaranteed, was sent of the total variance, all with eigenvalues over
together with the questionnaire. Two follow- 1 and each over 5% of the total variance ex-
up letters were sent as reminders to those plained. Items with loadings over ±.40 were
who did not respond after 2 and 4 weeks, re- chosen in the interpretation of the factors
spectively. shown in Table 1.
From the original sample a total of 522
persons did answer the questionnaire, and Second Wave
after excluding 12 subjects from the original
population (e.g., because of language and Ten years later the research group was
learning difficulties, or recent change of resi- interested in repeating the study as well as
RENBERG AND JACOBSSON 55

TABLE 1
Obtained Factors, Explained Variance, Items (abbreviated), Factor-Loadings
and Internal Consistency for the 20-item Version, 1986
Explained Factor- Internal
Factor Variance Loading Consistency*

1 Permissiveness 13.2% .66

Right to commit suicide .71
Situations where suicide is the only solution .67
Suicide ones own business .64
Suicide acceptable to end an incurable suffering .63
2 Unpredictability 11.5% .60
Suicide happens without warning .78
Relatives have no idea about what is going on .75
3 Incomprehensibility 8.3% .40
Suiciders are mentally ill .73
Suicides among younger people particularily puzzling .63
Anybody can commit suicide −.46
4 Noncommunication 6.9% .65
Communication not serious .80
People who make threats seldom complete suicide .80
5 Right to prevent 6.5% .34
Attempts a cry for help .75
Duty to restrain a suicide act .65
6 Preventability 6.0% .53
Can always help .87
Suicide decision can’t be reversed −.69
7 Relation-caused 5.5% .45
Attempts due to interpersonal conflicts .79
Attempts due to revenge or punishment .72
8 Suicidal process (duration) 5.0% .38
Suicides considered for a long time .80
Attempts are impulsive actions −.67
Suicidal thoughts will never disappear .45
WHOLE INSTRUMENT 63% .51

*Cronbach’s alpha

finding ways to increase reliability and valid- on other referents; that is, for the respondent
ity, now with influences from recent studies him/herself or for a close relative. A total of
in the field of attitudes toward suicide (Diek- 69 questions were initially constructed and
stra & Kerkhof, 1989; Rogers & DeShon, after comparison with the results from the
1995). The first step included a theoretical Dutch study (Diekstra & Kerkhof, 1989),
definition of attitude areas with regard to the only referent levels that had been found to
different referent levels described by Diekstra be of crucial importance were kept. This sim-
and Kerkhof (1989). The original 20 questions plified the procedure considerably, since
quite well into the model and were kept in fewer combinations of different referents
their original formulation to make compari- were necessary to use, and out of 69 attitude
sons possible; however, since items often were items 40 were retained.
formulated on a people in general referent level In the second study, performed in
they were complemented with formulations spring 1996, a likewise randomly selected
56 QUESTIONNAIRE ON ATTS

group of 1,000 persons, between 18 and 65 collapsed into three categories, reflecting agree-
years of age, was approached. A similar pro- ment, undecided, and disagreement. Overall
cedure as in 1986 was applied, and a total differences between waves were found con-
response rate of 64% was obtained. When cerning 4 of 8 selected attitude questions, but
comparing the responding group with popu- only on the right to commit suicide question
lation, there were, as in 1986, differences was the attitude change valid for both gen-
concerning gender and cohabiting status. A ders. Comparisons between genders also re-
confirmatory factor analysis with the restric- vealed a pattern where some differences dis-
tions of 8 factors was conducted with the appeared (e.g., right to commit suicide) and
original 20 questions, resulting in a model some emerged (e.g., possibilities to help a
explaining 63% if the variance. The obtained suicidal person). Age-specific responses on
eight factors corresponded very well with the the different attitude questions (not shown in
original factor model (it included the same the table) revealed that for the right to com-
questions). When adding the new questions mit suicide, there were no differences be-
into the factor analysis, and after exclusion of tween age groups, and the higher level of
items with low loadings (<.40) or low com- agreement in 1996 was significant ( p ≤ .05)
munality (<.40), a ten-factor model was ob- for all age groups. Older persons (51–65), of
tained explaining 60% of the total variance both genders, agreed to a significantly higher
(Table 2). This model included 34 questions. extent on the sentence that people who com-
In this factor model, as compared to mit suicide are mentally ill, but there was a
the 20-item model from 1986, items from the general decrease in agreement on this item
original permissive factor were spread out among all age groups over time. Younger per-
into three separate factors: suicide as a right, sons (18–35), of both genders, held a stronger
tabooing, and resignation. The predictability belief than others that it is always possible to
and communication factors were combined help a suicidal person. In 1996, there was a
into one communication factor, and the in- significant higher level of agreement on this
comprehensibility factor was divided in two: item in all age groups, as compared to 1986.
incomprehensibility and the normal-common Suicidal communication was considered as
factor. The internal consistency was high more serious by the youngest age group, in
(>.70) for the two first factors, and had in- both waves.
creased for the whole instrument. However,
for some factors, the internal consistency was Attitudes Toward Suicide in Different
rather poor, especially factors 5, 8, and 9. Demographic Groups

Data from the 1996 survey disclosed in-

ATTITUDES TOWARD SUICIDE teresting patterns concerning attitudes toward
IN DIFFERENT GROUPS suicide in different groups (Table 4). F-tests
were applied to compare mean factor scores
For comparisons between the two sur- (regression technique) between the groups. The
veys, the original 20-item version from 1986 factor scores are standardized measures with
was used as the bases. The 1996 34-item model an overall mean of 0 and a standard deviation
was used to describe attitudinal differences be- of ±1.
tween subgroups. The youngest as well as noncohabiting
persons held the most permissive attitudes,
Comparisons between waves however, this was significant only for women.
For men the most permissive attitudes were
Distribution of responses on the eight found among those living in an urban area.
factors from 1986, represented by the item While no gender differences were disclosed
with the highest factor loading are presented on the right to commit suicide factor, differ-
in Table 3. The five response categories were ences were found on factors such as non-
RENBERG AND JACOBSSON 57

TABLE 2
Obtained Factors, Explained Variance, Items (abbreviated), Factor-Loadings and Internal Consistency
for the 34-item Version 1996
Explained Factor- Internal
Factor Variance Loading Consistency*

1 Suicide as a right 16.3% .86

Get help to commit suicide if severe, incurable disease–myself .82
Give help to commit suicide if severe, incurable disease–people .78
Suicide understandable if severe, incurable disease–people .78
Consider suicide if severe, incurable disease–myself .76
Suicide acceptable means to end incurable disease–people .72
Right to commit suicide–people .66
Suicide should not always be prevented .43
2 Incomprehensibility 9.9% .72
Suicides among younger people particularily puzzling .70
Suicide among the worst thing to do to relatives .70
Not understandable that people can take their lives .69
Suicide can never be justified .55
Low estimated own suicide probability .41
3 Noncommunication 6.1% .64
Communication not serious .66
People who make threats seldom complete suicide .66
Suicide happens without warning .64
Relatives have no idea about what is going on .64
Most people avoid talking about suicide .46
4 Preventability 5.5% .63
Can always help .82
Suicide can be prevented .68
Suicide decision can’t be reversed −.59
5 Tabooing 4.9% .45
Should or would rather not talk about suicide .74
Risk to evoke suicidal thoughts if asked about .60
Suicide ones own business .41
6 Normal—common 3.8% .51
Could express suicide wish without meaning it—myself .70
Everyone has considered suicide .63
Anybody can commit suicide .46
7 Suicidal process 3.7% .51
Suicides considered for a long time .86
Attempts are impulsive actions −.75
8 Relation—caused 3.5% .45
Attempts due to interpersonal conflicts .77
Attempts due to revenge and punishment .71
9 Preparedness to prevent 3.1% .38
Prepared to help a suicidal person—myself .79
Duty to restrain a suicidal act .57
10 Resignation 3.1% .54
Suicide a relief .70
Situations where suicide is the only solution .44
WHOLE INSTRUMENT 60% .60

*Cronbach’s alpha
 58

TABLE 3
Responses on Eight Single Questions, Each Representing a Factor, 1986 and 1996, in Percent for Women (W)
and Men (M), Respectively
Diff*
Gender 1986–1996
ATTITUDE QUESTION 1986 Gender 1996 Gender spec. Overall
(abbreviated) Gender Agree ? Disagree diff* Agree ? Disagree diff* diff* diff*

Right to commit suicide W 8 17 75 .00066 20 31 49 ns .00000 .00000

M 16 25 59 25 30 45 .00488
Suicide happens without warning W 25 29 46 ns 27 33 40 .03816 ns .03580
M 25 27 49 35 26 40 .03385
Suiciders are mentally ill W 27 27 46 .00611 30 29 42 ns ns ns
M 40 22 38 26 29 45 .00314
Communication not serious W 31 33 37 ns 22 48 30 .00592 .00045 .00029
M 31 41 28 31 48 21 ns
Suicide attempt cry for help W 97 3 3 .02843 94 5 1 .00756 ns ns
M 92 6 2 88 11 1 ns
Can always help W 45 45 9 ns 53 42 6 .02738 ns .00546
M 54 36 11 63 32 6 .03628
Attempts due to interpers conflicts W 26 43 31 ns 32 35 33 ns ns ns
M 24 44 32 29 44 27 ns
Suicides considered for long time W 73 23 4 ns 75 21 4 ns ns ns
M 69 25 6 73 23 4 ns

? = undecided
*chi-square test, p-value
ns = p > .05
QUESTIONNAIRE ON ATTS
RENBERG AND JACOBSSON 59

TABLE 4
Responses on the Right to Commit Suicide Factor in 1996, Gender Specific
Mean Factor Scores in Different Demographic Groups
Women Men
Gender
Factor- Diff* Factor- Diff* Diff*
Demographic Group Score p-Value Score p-Value p-Value

Total −.0118 .0198 ns

Age 18–30 .1134 .0797 ns
31–50 .0267 .0281 .0654 ns ns
51–65 −.2847 −.1056 ns
Region Urban .0920 .2837 ns
Mixed −.1602 ns −.1916 .0037 ns
Rural .1138 −.0048 ns
Education −9 −.1621 .0408 ns
10–14 .0646 ns .0377 ns ns
14+ −.0088 −.0325 ns
Cohabiting Yes −.0888 .0340 −.0331 ns ns
No .1847 .1371 ns
Suicide nearby Yes −.0777 ns −.0123 ns ns
No −.0012 .1817 ns
*F-test
ns = p > .05

communication, tabooing, normal-common ferences were found only on the incomprehen-

and resignation (suicide as a solution), where sibility factor, where non-cohabiting women
men to a higher extent agreed on the state- were more understanding than the cohabit-
ments included. As for other attitudes and ing. Finally, women reporting on experiences
background factors, age-group comparisons of a suicide nearby were less agreeing on the
showed least understanding among the oldest relation-caused factor than other women.
women, while significantly more pro-preven-
tional attitudes and less tabooing were found Association to Self-reported
among the youngest, both genders. The oldest Suicidal Behaviors
men showed, as compared to the younger, the
most resigned attitudes; that is, that suicide Gender-specific comparisons were made
could sometimes be a relief and the only so- between the group responding affirmative on
lution. Concerning regional differences, men any of the questions on suicidal behavior con-
in urban areas were the most pro-preventional cerning last year (life-weariness, death-wishes,
men, and women in mixed areas responded suicidal ideation, suicidal plans, attempted sui-
with more resignation toward suicide than did cide) and the ’“non-suicidal” group on each
women in other areas. Concerning educa- of the 10 factors (Table 5).
tion, the well educated (14 years or more), of Suicidal women were, as compared to
both genders, were more understanding, did nonsuicidal women, significantly more agree-
believe more in communication prior to sui- ing on suicide as a right, were more under-
cide, and were less tabooing. However, well- standing, considered suicidal behaviors as more
educated men showed a more resigned atti- normal-common and more relation-caused.
tude toward suicide than did men with shorter Suicidal men, as compared to the non-suicidal,
education. When comparing gender-specific to a larger extent regarded suicidal behavior
attitudes related to cohabitational status, dif- as normal and relation-caused.
60 QUESTIONNAIRE ON ATTS

TABLE 5
Mean Factor Scores on the Ten Attitude Factors Obtained in 1996, with Regard
to Self-Reported Suicidal Behavior During Last Year (Yes, No), Gender Specific and Total
Women Men TOTAL
Yes No Diff* Yes No Diff* Yes No Diff*
FACTOR n = 118 n = 229 p-Value n = 95 n = 192 p-Value n = 213 n = 421 p-Value

Suicide as a right .1434 −.0971 .048 .1215 −.0314 ns .1337 −.0703 .024
Incomprehensibility −.3372 .1353 .000 −.0757 .0879 ns .2217 −.1147 .000
Noncommunication −.2915 −.0548 ns .1123 .1815 ns −.1132 .0581 ns
Preventability −.1284 .0015 ns −.0436 .0967 ns −.0910 .0458 ns
Tabooing .0800 −.2259 ns .3063 .1602 ns .0906 −.0454 ns
Normal-common .3718 −.3309 .000 .5317 −.1259 .000 .4424 −.2312 .000
Suicidal Process .1480 .0009 ns −.0316 −.0722 ns .0687 −.0368 ns
Relation-Caused .1592 −.1384 .023 .2139 −.0391 .036 .1834 −.0933 .002
Preparedness
to Prevent .0652 −.0344 ns −.0284 .0068 ns .0238 −.0133 ns
Resignation −.1133 −.1045 ns .2113 .0859 ns .0300 −.0152 ns

*chi-square test
ns = p > .05

DISCUSSION but comparisons based on basic demographic

variables did demonstrate a quite satisfactory
When designing the study and decid- representativeness. Furthermore, according
ing to use a postal questionnaire approach, to results from the other section of the ques-
questions were raised on the appropriateness tionnaire on self-reported suicidal behaviors
to do so. Could it even be dangerous since it (Salander Renberg, 2001), the responders did
might provoke and evoke suicidal problems, reveal quite high prevalences of suicidal be-
and would people actually respond? The re- haviors, indicating that the group of respond-
sponse rates of 75% and 64%, respectively, ers were persons quite familiar with suicidal
indicate that people in general actually do problems. If this assumption is correct, the
feel great concern for the problem and are nonresponders could constitute a group not
prepared to respond, despite the fact that it that close to or involved with suicidal prob-
might be frightening and provoke anxiety. lems, and consequently, according to the re-
Some people responding to the questionnaire sponses from the suicidal group, there could
made comments like: “A very important be some bias concerning permissiveness, in-
study, it was not easy to fill in the question- comprehensibility, normal-common, and re-
naire but the questions are very important lation-caused. Some reasons given from the
and we must learn to talk about and reflect nonresponding group, even though just a
on these issues, I hope I have been of some few, said that this was a too difficult and hard
help.” In this situation, it could even be a problem to approach and that they felt un-
preventive intervention per se to ask the pub- comfortable to answer the questionnaire.
lic about suicide to increase general aware-
ness and hopefully the preparedness to pre- Construction of the Instrument
vent suicide in society. However, we still have
a situation with nonresponders that should Beside the question on selection bias,
not be neglected. reliability and validity issues have to be ad-
The design of the study did not allow dressed. Concerning reliability, internal con-
for a study in nonresponders in more detail, sistency was increased in the second version
RENBERG AND JACOBSSON 61

of the instrument, especially for some sepa- suicide or attempted suicide, ideation or pre-
rate factors, where probably the new ques- paredness to help persons in suicidal crisis.
tions on different referents contributed to In current study the definition of the attitude
the increase. Still the internal consistency target is quite broad, and a strong argument
was low for the whole instrument and on for including all these targets is that we ac-
some factors even very low. Possibly, the in- tually do not have sufficient knowledge on
strument is measuring a too broad area of at- whether and how attitudes are differentiated
titudes toward suicide. An alternative would regarding suicide and other suicidal behaviors.
be to include more items on each domain and Most questions concern suicide, but there are
to cover fewer domains, since internal consis- exceptions (factor 6, 7, and 8). In future stud-
tency increases with the number of items on ies it might be warranted to more clearly sep-
each factor (e.g., Guilford, 1954). However, arate the different attitude targets, in order
the psychometric properties of the instru- to make separate analysis for different tar-
ment had to be balanced with the ambition gets.
to develop a multi-dimensional as well as fea-
sible instrument for the use in large-scale
Associations Between Attitudes
studies (not too time consuming and exten-
and Suicidal Behavior
sive). The internal consistency measure is
probably also influenced by the non clear-cut
nature of the attitudes toward suicide and the Data speak in favor of both positive
ambivalence associated to the phenomenon. and negative associations, depending on atti-
A crucial issue is whether an instrument mea- tude, type of suicidal behavior, and subgroup.
suring different attitude domains really can A higher level of permissive attitudes in 1996
yield a high total internal consistency since as compared to 1986, coincided with a con-
different attitudes toward suicide might be tinuous decline in suicide rates in Sweden
conflicting and instable. Another important since the beginning of the 1980s, suggesting
factor influencing the reliability is the heter- a negative association, where suicide rates de-
ogenity of the population under study (an in- crease and permissiveness increases. Gender-
evitable condition in this study), beside the specific patterns to the contrary supported a
heterogenity of the instrument. The ques- positive association in 1986, where women
tionnaire was aimed at measuring different who had suicide rates about four times lower
domains in a very heterogenous group of in- than males, were less permissive. However, in
dividuals, and can, consequently (theoretically), 1996 the female pattern was changed, giving
hardly accomplish the demands of high con- a situation where gender differences in per-
sistency. Most questionnaire studies conducted missiveness disappeared. If permissive atti-
in the field of attitudes toward suicide are tudes actually lead to a raise in suicide rates,
tested and fielded on quite homogenous groups, as was suggested by Farberow (1989), this in-
such as students. Such samples imply other crease in permissiveness among females might
conditions for high consistency. indicate a forthcoming increase in female sui-
As for the validity aspects, obtaining an cide rates. There are indications of such a sit-
identical factor model in both 1986 and 1996, uation among younger females according to
based on the 20-item version, is satisfying and recent studies where the rates for young wom-
gives support for a high construct validity. By en are not declining in Sweden, as compared
constructing the questionnaire after consult- to rates in most other age groups (Hultén,
ing several experts in the area as well as asking 2000). Another quite surprising finding in the
laymen it is possible to guarantee a rather high gender pattern was that men in 1996 to a high-
face validity, and the criterial validity was er extent than women believed in the possi-
partly met by the association with own sui- bilities to help a suicidal person (single item
cidal behavior. However, it is not obvious to comparisons). This might reflect a possible
which behaviors attitudes should be related: change in help-seeking behavior among men,
62 QUESTIONNAIRE ON ATTS

contributing to the current decrease in sui- tempters to further investigate this attitude
cide especially among males. in a more clinically defined group. For a sui-
An interesting finding was that the gen- cidal person the interaction with the sur-
eral response-style changed toward a more rounding is of crucial importance, whether it
frequent use of the undecided option on sin- is possible for the individual to express sui-
gle items in 1996. This might reflect the non cidal feelings openly, or if there is a repres-
clear-cut nature of attitudes toward suicide, sive and tabooing climate not prepared to
but also a generally increased awareness and share or listen to the problems. This touches
knowledge in society regarding suicidal be- on how to deal with attitudes in a preven-
haviors. Changes in some attitudes were clearly tional and interpersonal context. Among young
manifested in that sense (e.g., how to assess women, there is probably a relatively high
suicidal communication), where almost half correspondence between the attitudes in the
of the respondents, both genders, chose the surrounding peer group and that of a suicidal
undecided alternative in 1996. person, giving a climate where it is possible
Although there were no gender differ- to talk about the problem. However, we
ences in 1996 concerning permissiveness, such know too little about what is protective and
differences were found on factors that can be what is provoking in such a situation to draw
considered as central preventional ones: com- any definite conclusions on what constitutes
munication, tabooing, normality, and resig- a preventive situation. Further research is war-
nation. These results support the importance ranted.
of measuring different attitude areas and not To interpret attitude changes over
only unidimensional attitudes. time, the role of more general or official sui-
Since the relative rates of attempted cide prevention strategies have to be consid-
suicide in different subgroups are nearly the ered. In Sweden, as in many other western
reverse as compared to suicide rates, the as- countries, a great number of initiatives in the
sociation between suicide attempt rates and field of suicide prevention have been taken
attitudes toward suicide are reverse as well. during the last two decades, and a national
We find the most permissive attitudes in program for suicide prevention was estab-
groups with the highest rates of attempted lished in 1995. The program was followed by
suicide (younger women) (Salander Renberg, a more general activity in the field of suicide
1999). These age- and gender specific pat- prevention, and among other things attitudes
terns are important to consider, not only be- towards suicide were addressed in a public
cause they differ per se but as they probably health context. Therefore, the attitude changes
can take on different meaning in different might be due partly to the increased efforts
groups. in the suicide prevention area, that after so
In correspondence with other studies many years of tabooing there is an ongoing
(Kerkhof & Natawat, 1989; Platt, 1989; Stein, change in attitudes toward suicide—not only
Brom, Elizur, & Witztum, 1998), positive as- concerning permissiveness, but also concern-
sociations at the individual level were found ing prevention.
between permissive attitudes toward suicide
and self-reported suicidal behavior. The fac- Conclusions
tor includes items on the right to commit sui-
cide if ones suffering becomes unbearable, thus The current study shows that it is pos-
guaranteeing a way out. A group of suicidal sible to approach the general population with
persons might be close to that experience, a postal questionnaire on attitudes toward
and may psychic pain is too unbearable, that suicide, reflected in the relatively high re-
they have considered or even made an act to sponse rates. This study demonstrates that at-
escape from that suffering. It would be inter- titudes toward suicide is not just a matter
esting to also approach a group of suicide at- of hostility-sympathy or liberalization-re-
RENBERG AND JACOBSSON 63

pression, but consist of various domains, espe- 1. The increase in permissiveness indi-
cially concerning preventional issues. Data cate a forthcoming increase in sui-
also show that the direction of the association cide rates, especially among younger
between suicide and suicidal behaviors and dif- women.
ferent attitudes toward suicide varies according 2. A more open and permissive attitude
to subgroup, type of behaviors and aggregated gives allowance for the individual to
or individual levels. The results speak in fa- express suicidal problems, and conse-
vor of two divergent hypothesis: quently be more in the reach for help.

REFERENCES

CANETTO, S. S. (1992–93). She died for JENNER, J. A., & NIESING, J. (2000). The
love and he for glory: Gender myths of suicidal construction of the SEDAS: A new suicide-atti-
behavior. Omega, 26, 1–17. tude questionnaire. Acta Psychiatrica Scandinavica,
DELUTY, R. H. (1988–89). Factors affect- 102, 139–146.
ing the acceptability of suicide. Omega, 19, 315– KERKHOF, A.J.F.M., & NATAWAT, S. S.
326. (1989). Suicidal behavior and attitudes towards
DIEKSTRA, R.F.W. (1989). Introduction. In suicide among students in India and the Nether-
R.F.W. Diekstra, S. R. Maris, S. Platt, A. Schmidtke, lands. A cross-cultural comparison. In R.F.W.
& G. Sonnek (Eds.), Suicide and its prevention: The Diekstra, S. R. Maris, S. Platt, A. Schmidtke, &
role of attitude and imitation (pp. 1–4). Leiden: G. Sonnek (Eds.), Suicide and its prevention: The
Canberrs, Brill. role of attitude and imitation (pp. 144–159). Leiden:
DIEKSTRA, R.F.W., & KERKHOF, A.J.F.M. Canberrs, Brill.
(1989). Attitudes towards suicide. The development LESTER, D. (1972). Why people kill them-
of a suicide-attitude questionnaire (SUIATT). In selves: A summary of research findings on suicidal be-
R.F.W. Diekstra, S. R. Maris, S. Platt, A. Schmidtke, havior. Springfield, Illinois: Charles C. Thomas.
& G. Sonnek (Eds.), Suicide and its prevention: The MARKS, A. (1988–89). Structural parame-
role of attitude and imitation (pp. 91–107). Leiden: ters of sex, race, age, and education and their in-
Canberrs, Brill. fluence on attitudes toward suicide. Omega, 19,
DOMINO, G., MOORE, D., WESTLAKE, L., 327–336.
& GIBSON, L. (1982). Attitudes towards suicide: a NEELEMAN, J., & LEWIS, G. (1999). Sui-
factor analytic approach. Journal of Clinical Psychol- cide, religion, and socioeconomic conditions. An
ogy, 38, 257–262. ecological study in 26 countries, 1990. Journal of
FARBER, M. L. (1968). Theory of suicide. New Epidemiology and Community Health, 53, 204–210.
York: Funk and Wagnalls. OSGOOD, C., SUCI, G., & TANNENBAUM,
FARBEROW, N. L. (1989). Attitudes toward P. (1957). The measurement of meaning. Urbana:
suicide. In R.F.W. Diekstra, S. R. Maris, S. Platt, University of Illinois Press.
A. Schmidtke, & G. Sonnek (Eds.), Suicide and its PAYKEL, E. S., MYERS, J. K., LINDEN-
prevention: The role of attitude and imitation (pp. 281– THAL, J. J., & TANNER, J. (1974). Suicidal feelings
298). Leiden: Canberrs, Brill. in the general population. A prevalence study.
GIBBS, J. P. (Ed.). (1968). Suicide. New York: British Journal of Psychiatry, 124, 460–469.
Harper-Row. PLATT, S. (1989). A subculture of parasui-
GUILFORD, J. P. (1954). Psychometric meth- cide? In R.F.W. Diekstra, S. R. Maris, S. Platt, A.
ods. New York: McGraw-Hill. Schmidtke, & G. Sonnek (Eds.), Suicide and its
HALMAN, L. (1993). Suicidal attitudes and prevention: The role of attitude and imitation (pp.
modernizing society. Some methodological issues 108–143). Leiden: Canberrs, Brill.
and results from the European Values Study: RAMSAY, R., & BAGLEY, C. (1985). The
Methodological Aspects. Paper presented at the prevalence of suicidal behaviors, attitudes and as-
seminar on Suicidal Behavior in Society, Hög- sociated social experiences in an urban population.
berga Kursgård, Lidingö Sweden. Suicide & Life-Threatening Behavior,15, 151–167.
HULTÉN, A. (2000). Suicidal behaviour in ROGERS, J. R., & DESHON, R. P. (1995).
children and adolescents in Sweden and some eu- Cross validation of the five-factor interpretive
ropean countries. PhD dissertation, Karolinska model of the suicide opinion questionnaire. Sui-
Institute, Stockholm. cide and Life-Threatening Behavior, 25, 305–309.
64 QUESTIONNAIRE ON ATTS

SALANDER RENBERG, E. (1999). Parasui- SCHMIDTKE, A. (1997). Perspective: Sui-

cide in a northern Swedish county 1989–1995 and cide in Europe. Suicide and Life-Threatening Behav-
its relation to suicide. Archives of Suicide Research, ior, 27, 127–136.
5, 97–112. STEIN, D., BROM, D., ELIZUR, A., & WITZ-
SALANDER RENBERG, E. (2001). Self- TUM, E. (1998). The association between attitudes
reported life-weariness, death-wishes, suicidal plans toward suicide and suicidal ideation in adolescents.
and suicide attempts in general population surveys Acta Pychiatrica Scandinavica, 97, 195–201.
in the north of Sweden 1986 and 1996. Social Psy- STILLION, J. M., MCDOWELL, E. E., &
chiatry and Psychiatric Epidemiology, 9, 429–436. SHAMBLIN, J. B. (1984). The suicide attitude vi-
SALANDER RENBERG, E., & JACOBSSON, gnette experience. A method for measuring ado-
L. (1999). Some aspects of the dynamics in suicide lescent attitudes toward suicide. Death Education,
rates and the relevance for prevention: A northern 8, Supplement, 65–80.
Swedish study 1952–1995. Nordic Psychiatry, 53,
257–266.
SALE, I., WILLIAMS, C. L., CLARK, J., &
MILLS, J. (1975). Suicide behavior: Community Manuscript Received: July 2, 2001
attitudes and beliefs. Suicide, 5, 158–168. Revision Accepted: July 1, 2002