The Beck Depression Inventory in 28 Countries: A Meta-Analysis Dianne A. van Hemert
Fons J. R. van de Vijver Ype H. Poortinga
Tilburg University, the Netherlands
Running head: THE BECK DEPRESSION INVENTORY IN 28 COUNTRIES
Abstract
The Beck Depression Inventory (BDI) is one of the most widely used instruments to assess depression. Studies using the BDI in normal populations from 28 countries were collected. Depression showed the same correlates at individual and country level, which suggests that depression has the same meaning at the two levels. BDI scores were negatively correlated with economical and political variables and with indicators of societal equality. Splitting the sample in two subsets of more and less affluent countries resulted in significant differences on various correlations with the BDI. For example, BDI correlated significantly different with suicide rate and
Hofstede’s Masculinity for more and less affluent countries. Finally, after controlling for GNP, correlations with well-being, Confucian Work Dynamism and Human-Heartedness from the Chinese Value Survey, and Schwartz’s Autonomy value were significant.
Keywords: Depression
Beck Depression Inventory Subjective well-being Happiness
The Beck Depression Inventory in 28 Countries: A Meta-Analysis
The present study focuses on negative affect across individuals and countries. It is studied whether depression in normal populations, as measured by the Beck Depression Inventory (BDI; Beck et al., 1961), has a similar meaning at individual level and country level of analysis. This similarity is a prerequisite for interpreting BDI scores at country level. Further, hypotheses are tested about correlates of depression at country level.
In 1961, Aaron Beck and his colleagues (Beck et al., 1961) published the BDI to measure the ‘behavioral manifestations of depression’. The BDI is not based on a theory but on clinical observation of symptoms displayed by depressive patients (Steer et al., 1986). These symptoms include disturbed mood (sadness, loneliness, apathy) and a negative self-concept and self-punitive wishes. Further, depressive persons can suffer from physical and vegetative symptoms such as anorexia, insomnia, and loss of libido. Depressed individuals can also experience changes in activity level: They show either retardation or agitation (Beck, 1967). The BDI
disorder. In the first instance the extent of depression can be taken as a continuous variable with a mean level that is a characteristic of a population as a whole. In the second instance there is a proportion of individuals scoring above a threshold.
According to Rose (1992), ‘’the visible part of the iceberg (prevalence) is a function of its total mass (the population average)’’ (p. 72).
BDI scores of normal individuals can be aggregated to obtain country level data. However, in order to be able to interpret such data, one has to know whether the meanings of individual-level and country-level BDI scores are equivalent. This is a question of functional equivalence (Van de Vijver & Leung, 1997). One way to examine similarity of meanings is to make use of external variables that are part of the nomological networks of depression across levels. In this method, correlations of context variables with BDI scores are compared at individual and country level. This strategy focuses on functional equivalence, as it refers to the functional context of the concept (see also Van Hemert et al., submitted). An example of different correlations at individual and country level was described by Myers and Diener (1996). Within most nations there is hardly a relationship between income and happiness; although Americans earn on average twice as much today as they did in 1957, they have not become happier. At the same time people in rich countries are generally happier than people in less prosperous countries.
the basis of cross-cultural hypotheses about correlates of depression cross-culturally. A first theory, labeled livability theory by Veenhoven (1995), states that happiness is influenced by objective living conditions. These are broadly defined, ranging from education and income to equality, stability and freedom. Maslow's (1970) need-gratification theory underlies this perspective (Oishi et al., 1999). The comparison
theory claims that happiness is dependent on comparison with other times' or other
people's conditions (Veenhoven, 1995). This theory stresses relative living conditions as contrasted to absolute living conditions. Two types of comparison can be
distinguished: Social comparison involves comparison with other people and lifetime comparison involves comparison with other times in life. The folklore theory
(Veenhoven, 1995) states that happiness is a reflection of the national character of a country and is unrelated to the actual quality of life in that country.
In general, subjective well-being and happiness have been found to depend on absolute (objective) quality of life (Diener et al., 1993; Veenhoven, 1995),
supporting the livability theory. Arrindell et al. (1997) related country scores on subjective being to a number of country-level characteristics. Subjective well-being was found to be higher in wealthy countries with a high level of civil rights. Diener et al. (1995) reported positive correlations between subjective well-being and economical and political variables, as well as variables related to equality and
individualism, in 55 nations. Similar findings were reported by Veenhoven (1999), who examined happiness in 43 countries. He also found happiness to be related to educational variables, variables concerning media availability, and Schwartz's Autonomy. Inglehart (1997, 2000) found stability of democracy and to a lesser degree level of democracy to go together with well-being. Further, a strong
Inglehart added that above a threshold of about US$6,000 (in 1991 dollars) there was hardly a relationship. Diener and Diener (1995) reported similar findings.
With respect to the comparison theory, Hagerty (2000) found that a more positively skewed national income is associated with lower happiness. As for lifetime comparison, Easterlin (1995) discusses evidence for a positive link between
economical growth and the increase of happiness. However, studies by Diener et al. (1993) and Veenhoven (1995) did not find support for the idea that well-being is dependent on relative quality of life (for instance through social comparison).
Some evidence has been found to support the folklore theory. In the study by Arrindell et al. (1997) the Neuroticism and Psychoticism scales from the Eysenck Personality Questionnaire (EPQ-N respectively EPQ-P) were negatively correlated with subjective well-being (r = -.41, p < .05 respectively r = -.40, p < .05).
Furthermore, of the well-known dimensions of Hofstede (1980) uncertainty avoidance and power distance were negatively correlated with subjective well-being, while a high level of individualism went together with high subjective well-being. Although no significant correlation was found with masculinity for the total sample, in poorer countries masculinity and subjective well-being were positively correlated whereas in richer countries the opposite was found.
Additionally, studies on depression and well-being suggest that depression might also be positively correlated with marital status (Diener et al., 2000;
Mastekaasa, 1994; Myers, 1992; Myers & Diener, 1995; Veenhoven, 1983),
religiosity (Myers, 1992; Myers & Diener, 1995; Van den Eeden & Braam, 1999), and employment (Myers, 1992).
of such relationships at the individual level. In the following discussion we review individual-level studies using context variables which will be used later on in the country-level analyses. Several authors reported a negative correlation of depression with socioeconomic status (Gibbs, 1985) and with level of education (Oliver &
Burkham, 1979; Williamson, 1989), as would be predicted from livability theory. As for personality, numerous authors report a significant, positive correlation between BDI scores and EPQ-N (e.g., Heaven & Shochet, 1995; Meites et al., 1980;
The present study examines two questions. The first involves the equivalence of depression in normal populations as measured by the BDI at individual and
country level. Correlations with context variables at individual and country levels are compared to assess equivalence. Positive evidence with respect to this first question is conditional for the second part of the study. This second part examines the
correlates of depression at country level. Based on the theories and literature discussed, the following country-level hypotheses are tested.
General hypotheses:
1. Subjective well-being and happiness are negatively related to depression1. 2. Prevalence rates of (clinical) depression are positively related to
depression.
3. Suicide rate is positively related to depression.
Livability theory:
4. More affluent countries show lower depression scores. (According to the livability theory affluence pertains to economical well-being, as well as education and political stability.)
Comparison theory:
5. Equality of income distribution is negatively related to depression (social comparison).
6. Economical growth is negatively related to depression (lifetime comparison).
Folklore theory:
7. National character traits explain a significant portion of variance in depression.
8. The percentage of married people is negatively related to depression. 9. Religiosity is negatively related to depression.
10.Unemployment is positively related to depression. Method
Data Collection
PsycInfo (formerly known as PsycLit) was searched for reports, using the keywords BDI and Beck and Depression (time period covered: 1960- July 1999). Two strategies were used to find additional studies. First, several authors of obtained references were asked by letter or e-mail to provide all reports they had available on the Beck Depression Inventory. Second, by looking at references in literature already identified another part of the data set was found. Studies were included in the data set if mean scores on the BDI and sample sizes were reported. If available, the standard deviation of BDI score distribution, the internal consistency of the BDI measured by Cronbach’s alpha, and the mean and standard deviation of the age distribution were recorded. There was an important restriction concerning the type of sample used; clinical samples and all other groups were excluded that were
expected to produce scores deviating from a representative sample in a national population. Studies reporting ethnic diverse samples were only included if no significant differences in means were reported between the ethnicities. This restriction was needed because ethnic differences within a country were not the focus of our study. Each separate group of participants in a single report, for instance men and women or different age groups, was regarded as an independent study.
their feelings at the moment of administration, the revised version asks to report on the “past week including today”. Also, the wording of several items was changed. Although the number and global meaning of the 21 items remained the same, and high correlations between the versions were found, significant differences in means have been reported (Beck & Steer, 1984; Lightfoot & Oliver, 1985). The 1961 version generally shows lower means. Most articles did not explicitly state what version was used. To obtain this information various strategies were followed. First, if the version was specified in the list of references, either 1961 or from 1978, this was used as an indicator of the version. Nevertheless, Beck et al. (1988) acknowledge that “there is no feasible way of determining which version researchers have actually employed from the descriptions presented in the majority of studies”. They add that most researchers cite the 1961 reference, although some of them actually used the 1978 version. Second, when no reference was given it was assumed that other studies by the same author(s) and from the same country used the same version. In some countries, different versions have been used. In countries in which several samples were studied the ambiguous studies were eliminated, in others the year was decisive: All studies published before 1981 were regarded to concern the 1961 version, the others were considered to have used the 1978 version. A short form (Beck & Beck, 1972) was also introduced, consisting of 13 items from the original BDI. Gould (1982) and Reynolds and Gould (1981) found this short version for patients to be reliable and valid, and highly correlated with the standard from. Studies using this version were also included. Generally, it was clear when this version was used.
or adding items or modifying the content of existing items. Thus, several studies had to be ignored in setting up the data set. In one case, the deletion only concerned one item (Tashakkori et al., 1989). Here the mean total score was adjusted by adding one-twentieth of the total score.
Data Sets
<Table 1>
Three data sets were composed. The first data set included the studies using the 1961 version of the BDI. It consisted of 72 studies, providing data for 117
separate samples. Twenty-five countries were represented in this data set. Table 1 shows all countries in this data set with their means, standard deviations, and
reliabilities. The second data set contained the studies using the 1978 version of the BDI (Table 2). Thirty-seven studies, with 56 samples from 13 countries made up this data set. Finally, 13 studies using the short version of the BDI formed the third data set (Table 3). It consisted of 33 samples and included 8 countries.
<Table 2> <Table 3>
Publication years ranged from 1974 to 1999 in the 1961-version data set, from 1988 to 1999 in the 1978-version data set, and from 1984 to 1997 in the
short-version data set. Medians were 1993, 1995, and 1985 for the 1961-short-version, 1978-version and short-1978-version data set, respectively.
1978-version data set the corresponding figures were 7% men and 13% women. In 71% of the studies in this data set, either the samples were mixed or no separate information on sex was mentioned. Finally, the short-version data set contained 15% men and 16% women; 36% of the samples were mixed or had an unknown gender
distribution.
Age ranged from 15.00 to 74.10 years in the 1961-version data set, from 15.90 to 74.04 years for the 1978-version data set and from 17.27 to 84.50 years in the short-version data set. Participants were on average 22.99 (SD = 11.21) year in the 1961-version data set, 25.42 (SD = 8.66) year in the 1978-version data set and 39.57 (SD = 21.12) year in the short-version data set. Data on age were available in 83% and 89% of the studies in the 1961-version and the 1978-version data sets
respectively, and in all studies of the short-version data set.
Correlations across versions were calculated by correlating the means (weighted for sample size) of all data sets at country level: r = .64 (ns) between the version and the 1978-version data sets ; r = .79 (p < .05) between the 1961-version and the short-1961-version data sets; and r = .42 (ns) between the 1978-1961-version and the short-version data sets. Compared with the correlation of .94 between the 1961 and the 1978 version reported by Lightfoot and Oliver (1985), correlations are low. However, it should be noted that correlations were calculated at country level instead of individual level. Unfortunately the data were not suited for individual-level analyses.
Analysis and Testing of Hypotheses
correlations with the BDI at individual level had to be drawn from existing literature on the subject. Country-level correlations were derived from the total data set collected for the present study.
To test the hypotheses as formulated in the introduction several indicators at country level were used.
General Hypotheses
Subjective well-being. Diener’s subjective well-being was derived from Diener
et al (1995). This measure combines scores from several surveys and is available for 25 countries. Inglehart’s (1997) measure of subjective well-being was derived from questions concerning happiness and satisfaction with life as a whole (23 countries).
Happiness. Veenhoven (1999) reported data on happiness from the World
Database of Happiness (update 1996). His measure of happiness was used for 27 countries.
Prevalence rates. The World Health Organization (1999) reported 1998 data
Suicide rates. Death rates per 100,000 inhabitants by suicide were collected
using the Demographic Yearbook 1987 (United Nations, 1990a). Data were available for 25 countries.
Livability Theory
Affluence. An overall score on Affluence (similar to Georgas et al., submitted)
was derived from factor analyzing (for the countries involved in the present study) all variables constituting five ecosocial factors as used by Georgas and Berry (1995): Ecological factor (temperature and precipitation), Economical factor, Educational factor, Mass communication factor (number of telephones, newspapers, etc.) and Population factor (mortality and birth rates). The scores on the resulting single factor were used as a measure of Affluence. Where data on the indicators were not
available for a country in the data set, they were added by using the Demographic
Yearbook 1987 (United Nations, 1990a), the UNESCO Statistical Yearbook 1990
(United Nations, 1990b) and the historical weather data web site (Washington Post, n.d.). Data on Affluence were available for all 30 countries.
Gross National Product. Data on the GNP per capita were obtained from the
Georgas and Berry (1995) database.
Political Rights and Democracy. We combined four variables concerning
(percentage of total population who voted in the election). This index was available for 29 countries. We factor analyzed all four variables. A one-factor solution was found with an eigenvalue 3.33, explaining 83.24% of variance. Factor scores on this Political Rights and Democracy factor were used for further analyses (27 countries).
Civil rights. For 24 countries Gupta’s measure for observance of civil rights, as
cited by Diener et al (1995), was used. The original sign was reversed so as to obtain a higher score for more observance of civil rights.
Additional measures of democracy. Inglehart’s (1997) measure of the level of
democracy in 1990, as well as the measure for stability of democracy (expressed in number of years of continuous democracy) were available for 23 countries.
Deaths from political violence. As a measure of political instability the data by
Taylor and Jodice (1983) on deaths of political violence between 1948 and 1977 were used. This measure was corrected for population size, and available for all countries but one.
Comparison Theory
Gini index. The Gini index expresses the degree of income inequality in a
society. Indices for 21 countries, ranging from 1986 to 1995, were collected from the
World Development Report (World Bank, 1999). A higher value denotes more
inequality.
Economic growth. Average annual growth of the Gross Domestic Product
(GDP) between 1980 and 1987 was retrieved from the World Development Report
1989 (World Bank, 1989) and available for 27 countries. Folklore Theory
Eysenck Personality Questionnaire. Studies reporting means on the four
were collected across 38 countries. The four scales are the EPQ-P (Psychoticism), the EPQ-E (Extraversion), the EPQ-N (Neuroticism), and the EPQ-L (Social
Desirability). The data set includes 153 studies, providing data for 333 separate samples, and is described in more detail in Van Hemert et al. (submitted). Data on the EPQ scales were available for 24 countries in the BDI data set.
Hofstede’s measures. Data for 27 countries were available on Hofstede’s
(1980, 2001) work-related values Individualism, Masculinity, Power Distance, and Uncertainty Avoidance.
Values from Chinese Value Survey. The Chinese Culture Connection (1987)
reported factor scores of 22 countries (overlapping with 13 countries in the present data set) on four dimensions of the Chinese Value Survey. The four dimensions included Integration ('broadly integrative, socially stabilizing emphasis', but not familial bonding), Confucian Work Dynamism (hierarchical work ethic), Human-Heartedness ('gentleness, compassion', but also a 'legalistic approach'), and Moral Discipline (moral self-control and moderation ('following the middle way')).
Schwartz’s Values. Schwartz (1994) provided data on values in 19 countries
of our sample. Factor analysis by Georgas et al. (submitted) yielded two bipolar factors, labeled Autonomy and Hierarchy. Scores on these two dimensions were used as indicators.
Demographic Hypotheses
Marital Status. Percentages of married females and males between 25 and 44
years old were taken from the Georgas and Berry (1995) database. Data were available for 27 countries.
Religiosity. An analysis was done on a part of the 1990-1991 World Values
countries. It provided data on religion, including a range of topics related to religion, like the meaning of life, religious services, the role of churches and praying. We selected six items to fit in a scale measuring religiosity. Examples of items are “Do you find that you get comfort and strength from religion?”, “How important is God in your life?”, and “How often do you pray to God outside of religious services?”. The six items yielded a one-factor solution in factor analysis at country level, with an
eigenvalue of 5.39 (89.8% of variance explained). Cronbach’s alpha was .85. Data on religion were available for 21 countries.
Unemployment. The percentage of unemployment in 1987 was available for
23 countries (United Nations, 1993).
Results
Descriptives
Means, standard deviations, and values of Cronbach’s alpha for the different versions of the BDI in every country appear in Table 1, Table 2, and Table 3.
Moderate levels of depression according to Beck and Steer (mean score between 17 and 29; cited by Apostolou et al., 1999) were found in Israel and Kuwait, whereas mild levels of depression (mean score between 10 and 16) were found in India, Iran, Hong Kong, Australia, Japan, and Turkey. Cronbach’s alpha was higher than .80 in all countries but former West-Germany.
showing a z-score higher than 2.0. These countries (also) scored above the
threshold of 17 for mild depression as defined by Beck and Steer (Apostolou et al., 1999). The Kuwait data were based on three samples from two studies (N = 274), which were both published after the Gulf War. Possibly the influence of post-war stress was visible in these studies (Al-Issa & Ismail, 1994; Lester & Abdel-Khalek, 1998). The Israel data (N = 574) were based on only one study in a Palestine sample (Abdallah, 1997). Particularly in this group, the political situation is likely to enhance feelings of unsafety and depression. The outliers used the 1961 version of the BDI, but different translations. Kuwait and Israel were deleted from the data set, and standardizations were carried out again in all three data sets before combining them into the total data set, that was used for further analysis.
<Figure 1>
Correlations at Individual and Country Level: Establishing Functional Equivalence
<Table 4>
differences between individual level and country level occur for suicide rate and alcohol use. The overall similarities seem to suggest that functional equivalence of the BDI can be assumed.
Correlations with Country Level Variables: Hypothesis Testing
Pearson correlations and significance levels were calculated both in the regular way and with a bootstrapping procedure. For each correlation coefficient 10,000 bootstrapped samples were drawn with replacement from the original data set. The advantage of this procedure is not having to rely on the bivariate normality of the data and applicability to small samples.
General Hypotheses
<Table 5>
Measures of both subjective well-being and happiness showed negative correlations with depression scores (Table 5). Thus, the first hypothesis was
supported. Although both depression prevalence rates correlated positively with BDI scores, only one reached significance. Thus, the second hypothesis remains open. Suicide rate was negatively related to depression, a rather unexpected finding, that contrasts with other reports (Lester, 1988, 1986). This disconfirms the third
hypothesis.
Livability Theory
In general a higher level of Affluence was related to a lower level of depression. This finding is in agreement with prior findings on the relationship
supported. A factor analysis of all variables testing the livability theory resulted in a one-factor solution with an eigenvalue of 7.20, explaining 79.98% of variance. The significant negative correlation between BDI score and factor scores on this Livability factor (Table 5) confirms this conclusion.
Comparison Theory
The correlation between the Gini index, indicating equality of income
distribution, and BDI scores was not significant, contrary to expectation. Hypothesis 5 could not be confirmed. However, the correlation between economical growth and depression was significant in the bootstrap procedure: Hypothesis 6 was confirmed. In sum, social comparison does not appear to be associated with depression, but lifetime comparison does.
Folklore Theory
None of the EPQ scales correlated significantly with BDI score. Hofstede’s measures were related both positively (Masculinity and Power Distance) and
negatively (Individualism) to BDI scores, largely replicating earlier mentioned findings by Arrindell et al. (1997). The Chinese Integration value correlated negatively to BDI, and the Confucian Work Dynamism dimension showed a significant positive
correlation, but only when bivariate normality was assumed. Finally, Schwartz’s Autonomy factor was negatively related to BDI. This finding is in agreement with results reported by Sagiv and Schwartz (2000), who found values concerning achievement, self-direction and stimulation to be positively correlated with the
affective component of subjective well-being across three cultural groups. It is difficult to conclude whether this evidence does or does not support the rather unspecific predictions of the folklore theory.
None of the demographic hypotheses were confirmed. Percentages of married females and males were positively correlated with the level of depression,
contradicting expectations. Correlations with religiosity and the percentage of unemployed persons were not significant.
In sum, it can be concluded that wealthy, egalitarian, individualistic, and politically stable countries appear to be less depressed. These findings mainly support the livability theory.
Further Exploration of Country-Level Correlations
affluent countries the opposite was true (r = -.48). Hofstede’s Masculinity score also showed different correlations (p = .05) in the two subsamples; in affluent countries a high level of masculinity went together with a high level of depression (r = .77), whereas in less affluent countries the opposite was the case (r = -.42). This finding is congruent with Arrindell et al.'s (1997) results on subjective well-being, reported before. Human-Heartedness correlated positively with BDI scores for affluent countries (r = .84), whereas in less affluent countries a negative (although non-significant) correlation of -.27 was found. Correlations with Moral Discipline also differed significantly for the two subsamples (p = .05). Both correlations were positive (resp. r = .12 and r = .79). A relationship between wealth and depression could not be established in either subsample, in contrast to findings on subjective well-being in poor and rich countries (Inglehart, 1997).
Briefly summarized, in more affluent countries depression levels correlated positively with the Gini index, Neuroticism, Masculinity, Human-Heartedness, and the percentage of married males, while negative correlations were found with variables related to happiness, prevalence of depression, political rights, livability and with suicide. Less affluent countries only showed significant correlations of depression with the percentage of married females (positive) and Autonomy (negative). Thus, Inglehart’s (1997) finding, that only for less affluent countries GNP shows a
relationship with well-being, could not be replicated.
Partial correlations, controlling for GNP, left only four significant correlations. Inglehart’s well-being (r = -.53, p < .01), Confucian Work Dynamism (r = .64, p < .05), Human-Heartedness (r = .60, p < .05), and Schwartz’s Autonomy (r = -.56, p < .05) correlated with mean BDI score.
In the present study differences between countries in the level of depression, as measured by the Beck Depression Inventory, were investigated. Two research questions were addressed. First, the equivalence of BDI scores across different levels of analysis was examined. Patterns of correlations with context variables were compared at individual level and country level. In general, meanings of depression appeared to be reasonably similar at individual and country level. Differences occurred for suicide rate and alcohol use. Myers (1992) reported a similar
sufficiently positive on functional equivalence across levels to allow analysis of the second question.
Second, an attempt was made to explain country differences in BDI score in terms of contextual factors. As for the concepts of depression and subjective well-being, BDI scores were negatively correlated with subjective well-being measures. Furthermore, prevalence rates of clinical depression were positively correlated with depression score levels. The relatively low level of correlations can be explained by considering the nature of the populations involved. As self-reported depression inventories rely on the voluntary participation of individuals, extremely depressed participants are likely to be missed in questionnaire studies of non-clinical samples.
Mean depression scores were lower in affluent countries, that is countries with a high GNP, a high level of education, and a high level of information available through media. Also, countries with a low level of depression were more egalitarian, more democratic, and had more civil, human, and political rights. All these findings confirm the livability theory, explaining happiness through objective living conditions like income and education. This theory gets most support from the present results. These findings are in line with results reported by Veenhoven (1995, 1999), Diener et al. (1995), and Arrindell et al. (1997).
For the comparison theory, claiming that happiness is dependent on
comparison with other times' or other people's conditions, little support was found, as only economic growth was related to depression. This latter finding is not in
Also for the folklore theory, stating that happiness is a reflection of the national character of a country, only mixed evidence was found. Some variance could be explained by national character of a country, but mostly this was a function of GNP. Some value indices, Confucian Work Dynamism, Human-Heartedness and
Schwartz's Autonomy, were related to depression when GNP was controlled for. An obvious difficulty with folklore theory is the lack of precision of the predictions that can be derived from it. Any significant correlation provides supporting evidence, even independent of sign.
In general, several reasons can be named for the sometimes unclear results. First, many of the samples in the data set consisted only of students. Possibly, these samples are not representative. In poor countries students are relatively rich, in rich countries they are relatively poor (Diener et al., 1993). Apart from their extreme economic position (either relatively rich or relatively poor), student samples tend to be uniform in many other aspects. This argument was supported by Veenhoven (1995), who found the correlation between average happiness and GNP per capita to be larger in a general population than in a student population. Second, besides differences in means a shift in symptoms could take place, with more somatic and psychological symptoms scoring higher in one country than another. No raw data were available to do analyses on this point. A third reason is the presence of cultural bias. Bias can limit the validity of findings considerably (cf. Van de Vijver & Leung, 1997). For example, item bias should be taken seriously in cross-cultural
could emerge from the composition of the sample. It is clear that western, affluent countries are overrepresented in the present study. Splitting the sample in more and less affluent countries partly controls for this, but at the same time implies a
restriction of range. A third kind of bias pertains to the indicators. Some indicators like GNP and suicide rate are proxies for psychological variables with partly unknown ramifications.
Acknowledgement
Footnote
References
A list of references included in the meta-analysis may be obtained through the authors.
Abdallah, T.: 1997, ‘Reliability and validity of Palestinian Student Alienation Scale’, Adolescence 32, pp. 367-371.
Al-Issa, I., and S.J. Ismail : 1994, ‘Social support and depression of male and female students in Kuwait: Preliminary findings’, Anxiety, Stress, and Coping 7, pp. 253-262.
Apostolou, M., A. Asimakopoulou, M. Kopakaki, A. Kouri and A. Stalikas: 1999, ‘Beck Depression Inventory: Preliminary findings from a standardization process into the Greek language’, Poster session presented at the 5th European Conference on Psychological Assessment (Patra, Greece).
Arrindell, W.A., C. Hatzichristou, J. Wensink, E. Rosenberg, B. Van Twillert, J. Stedema and D. Meijer: 1997, ‘Dimensions of national culture as predictors of cross-national differences in subjective well-being’, Personality and Individual Differences 23, pp. 37-53.
Beck, A.T.: 1967, Depression: Clinical, Experimental, and Theoretical Aspects (Harper & Row, NY), pp. 3-9.
Beck, A.T., A.J. Rush, B.F. Shaw and G. Emery: 1979, Cognitive Therapy of Depression (Guilford: NY).
Beck, A.T. and R.A. Steer: 1984, ‘Internal consistencies of the original and revised Beck Depression Inventory’, Journal of Clinical Psychology 40, pp. 1365-1367.
Beck, A.T., R.A. Steer and M.G. Garbin: 1988, ‘Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation’, Clinical Psychology Review 8, pp. 77-100.
Beck, A.T., C.H. Ward, M. Mendelson, J. Mock and J. Erbaugh: 1961, ‘An inventory for measuring depression’, Archives for General Psychiatry 4, pp. 561-571.
Byrne, B.M. and T.L. Campbell: 1999, ‘Cross-cultural comparisons and the
presumption of equivalent measurement and theoretical structure: A look beneath the surface’, Journal of Cross-Cultural Psychology 30, pp. 555-574.
Carr, J.G., F.D. Gilroy and M.F. Sherman: 1996, ‘Silencing the self and depression among women’, Psychology of Women Quarterly 20, pp. 375-392.
Carrillo, J.M., N. Rojo and A.W. Staats: 1996, ‘Vulnerable personality in depression: Investigating commonality in the search for unification’, European Journal of
Chinese Culture Connection: 1987, ‘Chinese values and the search for culture-free dimensions of culture’, Journal of Cross-Cultural Psychology 18, pp. 143-164.
Costa, P.T. and R.R. McCrae: 1980, ‘Influence of extraversion and neuroticism on subjective well-being: Happy and unhappy people’, Journal of Personality and Social Psychology 38, pp. 668-678.
DeNeve, K.M. and H. Cooper: 1998, ‘The happy personality: A meta-analysis of 137 personality traits and subjective well-being’, Psychological Bulletin 124, pp. 197-229.
Diener, E.: 1994, ‘Assessing subjective well-being: Progress and opportunities’, Social Indicators Research 31, pp. 103-157.
Diener, E. and M. Diener: 1995, ‘Cross-cultural correlates of life satisfaction and self-esteem’, Personality Processes and Individual Differences 68, pp. 653-663.
Diener, E., M. Diener and C. Diener: 1995, ‘Factors predicting subjective well-being of nations’, Journal of Personality and Social Psychology 69, pp. 851-864.
Diener, E. and S. Oishi: 2000, ‘Money and happiness: Income and subjective well-being across nations’, in E. Diener and E.M. Suh (eds.), Culture and Subjective Well-Being (MIT, Cambridge, MA), pp. 185-218.
Diener, E., E. Sandvik, L. Seidlitz and M. Diener: 1993, ‘The relationship between income and subjective well-being: relative or absolute?’, Social Indicators Research 28, pp. 195-223.
Easterlin, R.A.: 1995, ‘Will raising the incomes of all increase the happiness of all?’, Journal of Economic Behavior and Organization 27, pp. 35-47.
Freedom House: n.d., ‘Annual survey of freedom country scores 1972-1973 to 1998-1999’ (Author: Washington, DC), retrieved September 27, 1999 from the World Wide Web: http://www.freedomhouse.org/rankings.pdf.
Georgas, J. and J.W. Berry: 1995, ‘An ecocultural taxonomy for cross-cultural psychology’, Cross-Cultural Research 29, pp. 121-157.
Gibbs, J.T.: 1985, ‘Psychosocial factors associated with depression in urban
adolescent females: Implications for assessment’, Journal of Youth and Adolescence 14, pp. 47-60.
Hagerty, M.R.: 2000, ‘Social comparisons of income in one's community: Evidence from national surveys of income and happiness’, Journal of Personality and Social Psychology 78, pp. 764-771.
Hays, W.L.: 1988, Statistics (Holt, Rinehart and Winston, NY).
Heaven, P.C.L. and I.M. Shochet: 1995, ‘Dimensions of neuroticism: Relationships with gender and personality traits’, Personality and Individual Differences 18, pp. 33-37.
Hofstede, G.: 1976, Nationality and Organisational Stress (European Institute for Research in Management, Brussels).
Hofstede, G.: 1980, Culture’s Consequences (Sage, Beverly Hills, CA).
Hofstede, G.: 2001, Culture’s Consequences. Second Edition. Comparing Values, Behaviors, Institutions, and Organizations Across Nations (Sage, Thousand Oaks).
Inglehart, R.: 1993, World Values Survey 1990-1991. WVS Program (J.D. Systems, S.L. ASEP S.A.).
Inglehart, R. and H.-D. Klingemann: 2000, ‘Genes, culture, democracy, and happiness’, in E. Diener and E.M. Suh (eds.), Culture and Subjective Well-Being (MIT, Cambridge, MA), pp. 165-183.
Lester, D: 1986, ‘Level of depression in nations and their suicide rates’, Psychological Reports 58, p. 930.
Lester, D: 1988, ‘National character and rates of personal violence (suicide and homicide)’, Personality and Individual Differences 9, p. 423.
Lester, D. and A.M. Abdel-Khalek: 1998, ‘Depression in college students in the United States and Kuwait’, Psychological Reports 83, p. 410.
Lightfoot, S.L. and J.M. Oliver: 1985, ‘The Beck Inventory: Psychometric properties in university students’, Journal of Personality Assessment 49, pp. 434-436.
Lynn, R: 1981, ‘Cross-cultural differences in neuroticism, extraversion and
psychoticism’, in R. Lynn (ed.), Dimensions in Personality. Papers in Honour of H. J. Eysenck (Pergamon, Oxford), pp. 263-286.
Maslow, A.H.: 1970, Motivation and Personality (Harper & Row, NY).
Myers, D.G.: 1992, The Pursuit of Happiness: Discovering the Pathway to Fulfillment, Well-Being, and Enduring Personal Joy (Avon, NY).
Myers, D.G. and E. Diener: 1995, ‘Who is happy?’, Psychological Science 6, pp. 10-19.
Myers, D.G. and E. Diener: 1996, ‘The pursuit of happiness’, Scientific American 274, pp. 54-57.
Meites, K., W. Lovallo and V. Pishkin: 1980, ‘A comparison of four scales for anxiety, depression, and neuroticism’, Journal of Clinical Psychology 36, pp. 427-432.
Oishi, S., E.F. Diener, R.E. Lucas and E.M. Suh: 1999, ‘Cross-cultural variations in predictors of life satisfaction: Perspectives from needs and values’, Personality and Social Psychology 25, pp. 980-990.
Oliver, J.M. and R. Burkham: 1979, ‘Depression in university students: Duration, relation to calendar time, prevalence, and demographic correlates’, Journal of Abnormal Psychology 88, pp. 667-670.
Reynolds, W.M. and J.W. Gould: 1981, ‘A psychometric investigation of the standard and short form Beck Depression Inventory’, Journal of Consulting and Clinical
Richter, P., J. Werner, A. Heerlein, A. Kraus and H. Sauer: 1998, ‘On the validity of the Beck Depression Inventory’, Psychopathology 31, pp. 160-168.
Rose, G.: 1992, The Srategy of Preventive Medicine (Oxford University Press, Oxford).
Sagiv, L. and S.H. Schwartz: 2000, ‘Value priorities and subjective well-being: direct relations and congruity effects’, European Journal of Social Psychology 30, pp. 177-198.
Saklofske, D.F., I.W. Kelly and B.L. Janzen: 1995, ‘Neuroticism, depression, and depression proneness’, Personality and Individual Differences 18, pp. 27-31.
Salamero, M., T. Marcos, F. Gutiérrez and E. Rebull: 1994, ‘Factorial study of the BDI in pregnant women’, Psychological Medicine 24, pp. 1031-1035.
Schwartz, S.H.: 1994, ‘Beyond individualism/collectivism: New cultural dimensions of values’, in U. Kim, H.C. Triandis, C. Kagitcibasi, S.C. Choi and G. Yoon (eds.),
Individualism and Collectivism (Sage, Thousand Oaks, CA), pp. 85-119.
Tanaka-Matsumi, J. and V.A. Kameoka: 1986, ‘Reliabilities and concurrent validities of popular self-report measures of depression, anxiety, and social desirability’, Journal of Consulting and Clinical Psychology 54, pp. 328-333.
Tashakkori, A., J. Barefoot and A.H. Mehryar: 1989, ‘What does the Beck Depression Inventory measure in college students?: Evidence from a non-Western culture’, Journal of Clinical Psychology 45, pp. 595-602.
Taylor, C.L. and D.A. Jodice: 1983, World Handbook of Political and Social
Indicators: Vol. 2. Political Protest and Government Change (Yale University Press, New Haven).
Ulusahin, A. and B. Ulug: 1997, ‘Clinical and personality correlates of outcome in depressive disorders in a Turkish sample’, Journal of Affective Disorders 42, pp. 1-8.
United Nations: 1990a, United Nations Demographic Yearbook 1987 (Author, NY).
United Nations: 1990b, Unesco Statistical Yearbook 1990 (Author, NY).
United Nations: 1993, United Nations Statistical Yearbook 1990/1991 (Author, NY).
International Association for Cross-Cultural Psychology (Swets & Zeitlinger, Lisse), pp. 336-347.
Van de Vijver, F.J.R. and K. Leung: 1997, Methods and data analysis for cross-cultural research (Sage, Newbury Park, CA).
Veenhoven, R.: 1983, ‘The growing impact of marriage’, Social Indicators Research 12, pp. 49-63.
Veenhoven, R.: 1995, ‘The cross-national pattern of happiness: Test of predictions implied in three theories of happiness’, Social Indicators Research 34, pp. 33-68.
Veenhoven, R.: 1999, ‘Quality-of-life in individualistic society: A comparison of 43 nations in the early 1990’s’, Social Indicators Research 48, pp. 157-186.
Washington Post: n.d., Historical Weather Database (Author, Washington, DC), retrieved August 9, 1999 from Historical Weather Database on the World Wide Web: http://www.washingtonpost.com/wp-srv/weather/historical/historical.htm.
Williams, P.G., R.S. Surwit, M.A. Babyak and C.C. McCaskill; 1998, Personality predictors of mood related to dieting’, Journal of Consulting and Clinical Psychology 66, pp. 994-1004.
World Bank: 1989, World Development Report 1989 (Oxford University Press, Washington DC).
World Bank: 1999, World Development Report: Knowledge for Development (Oxford University Press, NY).
Table 1
Study Characteristics of the 1961-Version Data Set
Number of BDIa
Country Resp. Samples Mean SD Alpha
Australia 225 3 7.57 6.50 --Bulgaria 691 1 9.14 -- --Canada 2140 13 8.85 5.84 0.82 China 100 1 6.68 4.18 --Denmark 217 1 5.00 0.40 --Finland 285 4 2.59 1.17 --Germany West 42 1 5.30 -- 0.60 Greece 37 1 4.89 5.40 --Iceland 187 2 5.77 5.50 --India 490 3 10.71 7.02 0.88 Iran 405 1 11.35 -- 0.82 Ireland 359 3 7.32 6.57 0.85 Israel 574 2 17.75 10.02 0.91 Italy 445 6 8.16 6.46 --Kuwait 274 3 16.59 9.27 --Netherlands 359 3 4.62 4.98 0.82 New Zealand 844 4 7.64 7.03 0.88 Nigeria 180 1 7.30 7.90 --Norway 100 1 5.92 5.33 --Spain 1524 3 7.34 7.09 --Sweden 3913 5 6.18 6.37 0.89 United Kingdom 4481 14 7.25 5.85 --USA 5456 34 6.75 6.37 0.85 Hong-Kong 2780 7 11.44 8.29 0.86 Total mean 26108 117 7.98 6.50 0.86
Table 2
Study Characteristics of the 1978-Version Data Set
Number of BDIa
Country Resp. Samples Mean SD Alpha
Argentina 608 2 8.46 7.26 0.86 Australia 1343 3 9.55 7.50 0.86 Brazil 3203 6 8.47 6.69 0.83 Canada 139 3 7.53 4.61 --Greece 98 1 8.17 6.69 0.86 Japan 116 2 10.99 8.04 --Mexico 1012 12 7.64 7.58 --Netherlands 10 1 1.00 1.10 --Spain 2407 5 6.44 5.81 0.83 Turkey 1514 4 13.06 7.88 --United Kingdom 47 1 4.60 4.20 --USA 1602 13 8.56 6.63 0.86 Hong-Kong 675 3 10.86 7.90 0.86 Total mean 12774 56 8.81 6.88 0.85
Table 3
Study Characteristics of the Short-Version Data Set
Number of BDIa
Country Resp. Samples Mean SD Alpha
Australia 1659 5 3.45 4.25 0.88 Canada 499 2 7.28 6.07 --Finland 1070 4 2.69 2.88 0.73 France 376 3 2.85 2.50 --Germany West 60 1 3.72 3.24 --Iran 305 1 8.03 3.29 0.83 Netherlands 156 1 2.80 3.50 --New Zealand 1091 16 2.50 3.09 0.81 Total mean 5216 33 3.67 3.68 0.84
Table 4
Correlations Between BDI Score and Context Variables Within Countries and Between Countries
Context variables Individual levela Country levelb
Income (SES) - -.46* Educationc - -.54** Suicide + -.46* Alcohol used + -.25 Anxietye + -.02 EPQ-P + .29 EPQ-E - -.05 EPQ-N + .23 EPQ-L - .23
aResults are based on literature reviewed in the introduction section, and studies reported by Beck et al. (1988)
bResults are based on analyses from the present study
cAt country level measured by the Education factor (Georgas & Berry, 1995), consisting of the teacher-pupil ratio in the first level, the gross enrollment in the second and the third level, and the percentage of adult illiterates
dAt country level measured by death rate through liver cirrhosis (United Nations, 1990a)
eAt country level measured by mean scores on the question ‘How often do you feel nervous or tense at work?’ (Hofstede, 1976) as cited in Lynn (1981), with reversed sign
Table 5
Correlation Coefficients and Significance Levels (Regular and After Bootstrapping) Between Standardized BDI Scores and Context Variables for More Affluenta and Less Affluent Countries
BDI
Context variables N
countries
Total sample More affluent countries (Nmax = 14) a
Less affluent countries (Nmax = 14) General hypotheses Subjective Well-Being Diener 24 -.46*† -.42 -.29 Inglehart 23 -.66**†† -.74**†† -.49 Happiness (Veenhoven, 1999) 26 -.24†† -.59*† -.03 Prevalence rates
Unipolar major depression 28 .19 -.52†† -.23
Bipolar affective disorder 28 .33† -.50 .24
Suicide rate 23 -.46*† -.70*†† .22
Livability theory
Affluence 28 -.50**†† -.18 -.31
Gross National Product 28 -.46*†† -.14 -.18
Political Rights and Democracy 25 -.46*†† -.73**†† -.33
Civil rightsb 23 -.46*† -.62*†† -.40
Level of democracy 23 -.29 .15 -.04
Stability of democracy 23 -.46*† -.28 -.12
Death by political violence (corrected) 28 -.02 .12 -.16
Factor score Livability 20 -.40† -.62*† -.11
Comparison theory
Economic growth 25 .34† .45 .34 Folklore theory
Eysenck Personality Questionnaire
Psychoticism (EPQ-P) 23 .29 .33 .13
Extraversion (EPQ-E) 23 -.05 .18 -.19
Neuroticism (EPQ-N) 23 .23 .49† -.11
Social Desirability (EPQ-L) 22 .23 -.10 -.10
Hofstede's Measures Individualism 26 -.45†† -.07 -.26 Masculinity 26 .38† .77**†† -.42 Power Distance 26 .50**†† .40 .30 Uncertainty Avoidance 26 .07 .36 -.26 Chinese Values Integration 13 -.67*† -.41 -.68
Confucian Work Dynamism 13 .64* .33 .76
Human-Heartedness 13 .48 .84**†† -.27 Moral Discipline 13 .40 .12 .79 Schwartz’s Values Autonomy factor 18 -.62**†† -.36 -.72*†† Hierarchy factor 18 .35 .44 .05 Demographic hypotheses
Percentage of married females 25 .65**†† .41 .74**††
Percentage of married males 25 .52**† .40† .43
Religiosity (Inglehart, 1997) 21 .29 .36 .10
Percentage unemployed 22 -.12 .07 -.36
aMore affluent countries have a GNP (1987) higher than 9210 US$, less affluent countries have a GNP lower than 9210 US$ bOriginal sign changed
Figure Caption
30 N =
