Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Jan 10.
Published in final edited form as: J Affect Disord. 2012 Aug 1;144(1-2):101–105. doi: 10.1016/j.jad.2012.05.066

A multilevel analysis of association between neighborhood social capital and depression: Evidence from the first South African National Income Dynamics Study

Andrew Tomita a,*, Jonathan K Burns b
PMCID: PMC3513630  NIHMSID: NIHMS399249  PMID: 22858263

Abstract

Background

As neuropsychiatric disorders account for a great proportion of the total burden of disease in sub-Saharan Africa, depression is rapidly emerging as a public health issue in South Africa. Given the divisions enforced by a legacy of the apartheid spatial and economic policies, features of communities such as neighborhood-level social capital may play a critical role in depression. However, the extent to which neighborhood-level social capital is associated with depression in South Africa at the population-level is unknown.

Methods

Data from the first wave of the South African National Income Dynamics Study (SA-NIDS) was used to examine the association between the neighborhood-level social capital and individual depression using multilevel regression models.

Results

There was a negative association between neighborhood-level social capital and depression score with social trust and neighborhood preference accounting for this association. Structural social capital, namely civic participation, was not related to depression. Individual predictors, including social class, self-rated health status and education, were strong covariates of depression.

Limitations

The cross-sectional design of the study limits our understanding of the temporal order of social capital and depression.

Conclusions

In post-apartheid South Africa, low social capital remains an important social determinant of health, including depression outcome. This is in addition to individual determinants related to class such as unemployment, education and social class which play an important role in influencing depression. Further research utilizing a longitudinal study design is warranted to examine the association between social capital and depression in South Africa.

Keywords: Depression, Social capital, South Africa, Apartheid, Multilevel analysis

1. Introduction

Over the last decade there has been growing research on contextual factors, such as features of neighborhood, and their impact on individual health outcomes. Social capital – the features of social structure such as norms, trusts, and networks that can facilitate collective action for mutual benefit (Coleman, 1990; Putnam, 1993) – is considered an important determinant of health, including mental health. Systematic reviews point to social capital being an important factor in improving mental health (De Silva et al., 2005), particularly in low-income countries (De Silva et al., 2007).

The burden associated with mental health conditions is a serious issue in sub-Saharan Africa (Lopez et al., 2006) and depression in particular is increasingly recognized as a pressing public health concern in South Africa (Ardington and Case, 2010; Tomlinson et al., 2009). While a number of recent studies point to an association between social capital and depression (Aslund et al., 2010; Jung et al., 2012; Kouvonen et al., 2008; O’Connor et al., 2011; Oksanen et al., 2010), evidence from low- and middle-income countries (LAMICs) is limited. There is also little research examining subtypes of social capital at both neighborhood- and individual-level in relation to depression. Lastly, the legacy of apartheid’s racially and geographically segregated communities in South Africa provides a unique opportunity to examine the important role of neighborhood in relation to mental health outcomes. Many challenges of the post-apartheid era still remain, as reflected in housing policies that promote the development of new houses in existing neighborhoods, thereby enhancing rather than dismantling the apartheid urban spatial legacy (Bradlow et al., 2011). In the current study, we examined the association between neighborhood-level social capital and individual depression outcome using a multilevel regression modeling technique.

2. Methods

2.1. Data source

Data (version 4) from the first wave of the South African National Income Dynamics Study (SA-NIDS) was used to examine the association between neighborhood social capital and individual depression outcome. SA-NIDS is the first longitudinal panel survey of a nationally representative sample of households in South Africa. Wave 1 of the study was carried out in 2008 by the Southern Africa Labour and Development Research Unit, University of Cape Town. The data became publically available in 2009. The first wave of SA-NIDS involved approximately 16,800 adults from 7300 households across 400 primary sampling units. There are two separate questionnaires in the SA-NIDS: the adult and household questionnaires, available in 11 official languages of South Africa. The adult questionnaire was administered to every household member aged 15 or older. Using a non-randomized method, the household questionnaire was administered to the oldest woman in the household or to another household member knowledgeable about living arrangements. The latter questionnaire captures information relevant to household issues. The sampling methods and response rates are detailed in the SA-NIDS technical report (Leibbrandt et al., 2009); the overall individual level non-response rate being 6.7% and overall household level non-response rate being 31%.

2.2. Measures

The primary health outcome of the study was depression status which was assessed using the self-reported 10-item version scale of the Center for Epidemiologic Studies Depression (CES-D) scale available in the adult questionnaire of SA-NIDS. CES-D, frequently used in studies in South Africa (Hamad et al., 2008; Myer et al., 2008), was designed to measure depressive symptoms in the general population (Radloff, 1977). The 10 item scale correlates well with the original 20 item version with little loss of psychometric properties (Shrout and Yager, 1989). The depression score was the sum of the scores of the 10 items, which ranges from 0 to 30. CES-D score was treated as a continuum of psychological distress (Steffick, 2000), with the level of depression increasing with increasing score. The alpha value of CES-D-10 was .75, consistent with another study (Cole et al., 2004).

Neighborhood-specific social capital, the main predictor of our study, was assessed using four variables in the SA-NIDS Household questionnaire: (1) support network and reciprocity, (2) association activity, (3) collective norm and values, and (4) safety. The nature of support networks and reciprocity was assessed with the question, “How common is it that neighbors help each other out?” Association activity was assessed with the question “How common is it that neighbors do things together?” To these two questions, respondents rated on a 5-point scale, with 1 being never happens, and 5 being very common. Collective norms and values were assessed with the question, “How common is it that people in your neighborhood are aggressive?” Safety was assessed with the question, “How common is burglary and theft in your neighborhood”. Respondents rated these two questions on a 5-point scale, with 1 being very common, and 5 being never happens. The social capital score was the total from the above four questions, with higher score reflecting higher social capital. Social capital was aggregated across households to create a neighborhood social capital score (ranged from 9 to 19), this being categorized into three groups: low (9–12), moderate (13–15) and high (16–19).

Two individual-level social capital predictors used in the SA-NIDS adult questionnaire were civic participation and social trust. Civic participation was assessed based on respondent’s participation in any of 18 association/groups and is an example of an objectively verifiable structural dimension of social capital which refers to what people do (Harpham, 2008). Social trust is a subjective example of cognitive social capital which refers to what people feel (Harpham, 2008). The latter was assessed by the question, “Imagine you lost a wallet or purse that contained R200 and it was found by someone who lives close by” requiring a response on a 3-point scale (1 being very likely and 3 being not likely at all to be returned with the money in it). The study included a question regarding individual preference to remain in the neighborhood; this being considered a moderator of the relationship between neighborhood social capital and health outcomes (Carpiano, 2008), incorporating an important aspect of neighborhood integration (Morrow, 1999) and a connection to networks that possess resources (Bourdieu, 1986).

A number of individual demographic, socioeconomic, and health-related covariates were utilized from the SA-NIDS adult questionnaire including: gender, race, marital status, employment, educational attainment, age group, social class, and current health status. For social class, respondents were categorized using a six-point scale, with 1 being the lowest and 6 being the highest social class. Health status required respondents to rate their overall health on a five-point scale, with 1 being excellent, and 5 being poor.

2.3. Data analyses

Multilevel models were used to analyze the association between neighborhood social capital indicators and individual depression outcome, and involved two levels: individual and neighborhood. Four random intercept models were sequentially fitted. Model 1 was a null model with only the constant term in the fixed and random parts. Model 2 provided the effect of individual-level predictors on depression outcome without individual-level social capital indicators. Model 3 examined the effect of all individual predictors on depression. Model 4 was the full model that considered all explanatory variables at individual and neighborhood levels. All models were adjusted by the post-stratification weight as estimated by Wittenberg (2009), as the SA-NIDS is survey data to match the population estimates produced by Statistics South Africa for the mid-year population estimates for 2008 for population distribution of provinces and households. Model-fit statistics using AIC (Akaike, 1974) were established as well as estimates of variance and the intra-class correlation coefficient which explains the proportion of total variance at individual and neighborhood level. The data were analyzed using STATA version 12 (StataCorp, 2011).

3. Results

The crude depression scores were stratified by gender, population group, and age group in a published study (Ardington and Case, 2010). Table 1 provides the results based on 13,469 (81%) of the 16,657 adult respondents. The mean depression score of adult individuals in our study was 7.9 out of a score of 30. Over half of the sample were female (56%), the largest age group was 35–59 years (35%), and 51% were never married. In terms of socioeconomic status, 71% were currently unemployed and 61% completed high school. Most reported excellent (31%) and good overall health (28%). For social capital perceived by the individual, 36% belonged to at least one association or organization and 72% thought a wallet containing R200 would not likely be returned with the money in it if it was found by someone who lived close by. An overview of social capital across 400 communities is provided in Table 1.

Table 1.

Descriptive statistics for 13,469 resident adults across 400 neighborhoods.

Variable Category n (%)
Individual levela
Gender Male 5423 (.44)
Female 8046 (.56)
Population category Black 10,481 (.78)
Colouredb 1980 (.09)
Asian/Indian 200 (.02)
White 808 (.11)
Age 15–20 2738 (.20)
21–24 1414 (.10)
25–34 2568 (.24)
35–59 4911 (.35)
60+ 1838 (.11)
Marital status Married 3860 (.31)
Living with partner 1101 (.08)
Widow 1173 (.07)
Divorce/Separated 358 (.03)
Never married 6977 (.51)
Employment Employed 3338 (.29)
Unemployed 10,131 (.71)
Education No education 1786 (.09)
Did not complete school 8678 (.61)
Completed high school 3005 (.30)
Health status Excellent 3827 (.31)
Very good 3522 (.28)
Good 3302 (.23)
Fair 1755 (.11)
Poor 1063 (.07)
Social class Rung 1(lowest) 2104 (.13)
Rung 2 5113 (.34)
Rung 3 4485 (.35)
Rung 4 1431 (.14)
Rung 5 282 (.03)
Rung 6 (highest) 54 (.01)
Neighborhood attachment Strong preference to stay 8 8311 (.58)
Moderate preference to stay 1930 (.15)
No preference to stay 1427 (.13)
Moderate preference to leave 809 (.06)
Strong preference to leave 992 (.08)
Social trust High 1345 (.13)
Medium 1948 (.15)
Low 10,176 (.72)
Civic participation No 8653 (.64)
Yes 4816 (.36)
Neighborhood levelc
Neighborhood social capital Low (9–12) 52 (.13)
Moderate (13–15) 322 (.80)
High (16–19) 26 (.07)
Open in a new tab
a

Percentage results are weighted using post-stratification weights.

b

The term ‘coloured’ is used by Statistics SA.

c

400 neighborhoods with social capital score range between 9 and19.

The results from the four multilevel models are presented in Table 2. Intra-class correlation analysis from Model 1 indicates that 85% and 15% of variance in depression was explained by individual and neighborhood-level differences respectively. Model 2 suggests that gender, age, economic rung, marital status, educational attainment, employment and health status seem to influence depression. Model 3 suggests that lower social trust was significantly associated with higher depression, while higher preference to remain in the neighborhood was significantly associated with lower depression, controlling for individual demographic characteristics from Model 2. Participation in civic organizations was not associated with depression. Model 4, a full model controlling for all explanatory variables, indicated that neighborhoods with high social capital were significantly associated with lower depression scores in residents. Social trust and neighborhood preference were significant predictors of depression but civic participation was not. Among the four models, model 4 had the most appropriate fit based on lowest AIC.

Table 2.

Results for the multilevel linear regression model (N = 13,469).

Variable Model 1, SE Model 2, SE Model 3, SE Model 4, SE
Individual level
Gender: (Male)
Female .46(.08)*** .46(.08)*** .46(.08)***
Age: (15–20)
21–24 1.34(.14)*** 1.34(.14)*** 1.34(.14)***
25–34 1.54(.14)*** 1.57(.14)*** 1.57(.14)***
35–59 1.56(.16)*** 1.67(.16)*** 1.67(.16)***
60+ .97(.21)*** 1.12(.21)*** 1.13(.21)***
Population category: (White)
Black 1.77(.30)*** 1.68(.29)*** 1.70(.29)***
Coloured .70(.36)** .67(.35) .67(.35)
Asian/Indian .86(.88) .84(.83) .82(.83)
Marital status: (Married)
Living with partner .22(.20) .17(.19) .17(.20)
Widow 1.09(.20)*** 1.10(.20)*** 1.10(.20)***
Divorce/Separated 1.73(.32)*** 1.71(.32)*** 1.71(.32)***
Never married .40(.13)*** .36(.13)*** .36(.13)***
Employment: (Employed)
Unemployed .48(.11)*** .53(.11)*** .54(.11)***
Education: (No education)
Did not complete school −.35(.16)** −.40(.15)*** −.40(.14)***
Completed high school −.80(.18)*** −.89(.18)*** −.91(.18)***
Health status: (Excellent)
Very good .58(.13)*** .51(.12)*** .52(.12)***
Good .97(.15)*** .92(.15)*** .92(.15)***
Fair 1.74(.18)*** 1.71(.17)*** 1.70(.17)***
Poor 3.55(.25)*** 3.50(.24)*** 3.49(.24)***
Social class: (Rung 1–Lowest)
Rung 2 −.92(.19)*** −.88(.19)*** −.88(.19)***
Rung 3 −1.29(.19)*** −1.22(.19)*** −1.22(.19)***
Rung 4 −1.42(.23)*** −1.39(.23)*** −1.39(.23)***
Rung 5 −1.67(.37)*** −1.64(.35)*** −1.64(.35)***
Rung 6 −2.01(.58)*** −2.03(.60)*** −2.03(.59)***
Neighborhood attachment: (No preference to stay)
Strong preference to stay −.73(.16)*** −.72(.16)***
Moderate preference to stay −.25(.19) −.25(.19)
Moderate preference to leave −.27(.22) −.28(.22)
Strong preference to leave .76(.25)*** .76(.25)***
Social trust: (High)
Medium .82(.21)*** .83(.22)***
Low .25(.19) .25(.19)
Civic participation: (No)
Yes −.03(.09) −.03(.09)
Neighborhood level:
Social capital: (High)
Moderate .82(.35)**
Low .99(.39)***
Variance components:
Level 2 variance, σ2 (SE) 3.34(.27) 2.15(.20) 2.18(.20) 2.61(.23)
Level 2 intra-class correlation .15 .11 .11 .11
Model fit
AIC		 79,056 77,482 77,339 77,337
Open in a new tab

Reference category is in bracket. All regression results are using post-stratification weights. The regression is across 400 neighborhoods. All SE are reported as robust SE.

**

p ≤ .05.

***

p ≤ .01.

4. Discussion

Utilizing data from SA-NIDS, this study has shown that neighborhood social capital was associated with individual depression outcome (Model 4). In addition, our finding (Model 3) suggests that individual level social capital, particularly social trust, also influence individual depression, in addition to many demographic predictors (Model 2). When comparing Models 1 and 3, the neighborhood variance declined, suggesting that variation found in Model 1 was explained by differences in the individual- and neighborhood-level predictors. Our findings are consistent with other studies which found that individual-level low neighborhood attachment (O’Brien et al., 1994) and social trust (De Silva et al., 2005; Fujiwara and Kawachi, 2008), are significant risk factor for depression.

Our study did not find a significant association between civic participation (an example of structural social capital) and depression. As indicated in a systematic review (De Silva, 2006), structural social capital may not always be positively associated with mental health outcomes. Separate studies also reported that civic participation had no impact on depression outcome in their study (Mitchell and LaGory, 2002; Veenstra, 2005). Importantly, legacies of apartheid neighborhoods still remain a reality in South Africa (Schensul and Heller, 2010) where it is argued class is gradually replacing race as the major manifestation of inequality (Seekings and Nattrass, 2005). We believe our study lends support to this statement: individual determinants such as social class, education and unemployment, in addition to characteristics of neighborhood, still matter.

Major strengths of this study include: the large, nationally representative sample; as well as the analysis that takes the hierarchical nature of the data into consideration by using multilevel analysis and controlling for important confounders (including individual-level social capital, socioeconomic status and self-rated health status.) Our study has several limitations. Firstly, the cross-sectional nature of our study means it could be susceptible to selection bias, as depression could cause people to live in neighborhoods with low social capital. Secondly, the neighborhood-specific social capital information was collected from the household questionnaire which was administered to the oldest woman or another household member with knowledge about the living arrangements of the household. The lack of variation here in demographic characteristics of respondents (such as age, gender and seniority of household) can systematically bias aggregated neighborhood-level measures. While further research, utilizing a longitudinal study design, is warranted to examine the temporal association between social capital and depression in South Africa, this study has contributed to a better understanding about neighborhood social capital and depression in that country.

Acknowledgments

Role of funding source

Dr. Tomita was supported by the National Institutes of Health Office of the Director, Fogarty International Center, Office of AIDS Research, National Cancer Center, National Eye Institute, National Heart, Blood, and Lung Institute, National Institute of Dental & Craniofacial Research, National Institute on Drug Abuse, National Institute of Mental Health, National Institute of Allergy and Infectious Diseases Health, and NIH Office of Women’s Health and Research through the International Clinical Research Fellows Program at Vanderbilt University (R24 TW007988) and the American Recovery and Reinvestment Act.

The baseline study of South African National Income Dynamics Study (SA-NIDS) was conducted by the Southern Africa Labour and Development Research Unit (SALDRU) based at the University of Cape Town’s School of Economics. The research team is led by Murray Leibbrandt (SALDRU director/University of Cape Town) and Ingrid Woolard (SALDRU’s chief research officer). The data was accessed through Southern Africa Labour and Development Research Unit. National Income Dynamics Study (NIDS) 2008, Wave 1 (dataset). Version 4. Cape Town: Southern Africa Labour and Development Research Unit (producer), 2012. Cape Town: DataFirst (distributor), 2012.

Footnotes

Conflict of interest

All authors declare that they have no conflicts of interest.

References

  1. Akaike H. A new look at the statistical model identification. IEEE Transactions on Automatic Control. 1974;19:716–723. [Google Scholar]
  2. Ardington C, Case A. Interactions between mental health and socioeconomic status in the South African National Income Dynamics Study. Tydskrif vir studies in ekonomie en ekonometrie. 2010;34:69–85. [PMC free article] [PubMed] [Google Scholar]
  3. Aslund C, Starrin B, Nilsson K. Social capital in relation to depression, musculoskeletal pain, and psychosomatic symptoms: a cross-sectional study of a large population-based cohort of Swedish adolescents. BMC Public Health. 2010;10:715. doi: 10.1186/1471-2458-10-715. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bourdieu P. The forms of capital. In: Richardson JG, editor. Handbook of Theory and Research for the Sociology of Education. Greenwood Press; Westport: 1986. pp. 241–258. [Google Scholar]
  5. Bradlow B, Bolnick J, Shearing C. Housing, institutions, money: the failures and promise of human settlements policy and practice in South Africa. Environment and Urbanization. 2011;23:267–275. [Google Scholar]
  6. Carpiano RM. Actual or potential neighborhood resources and access to them: testing hypotheses of social capital for the health of female caregivers. Social Science and Medicine. 2008;67:568–582. doi: 10.1016/j.socscimed.2008.04.017. [DOI] [PubMed] [Google Scholar]
  7. Cole JC, Rabin AS, Smith TL, Kaufman AS. Development and validation of a Rasch-derived CES-D short form. Psychological Assessment. 2004;16:360–372. doi: 10.1037/1040-3590.16.4.360. [DOI] [PubMed] [Google Scholar]
  8. Coleman JS. The Foundations of Social Theory. Havard University Press; Cambridge, MA: 1990. [Google Scholar]
  9. De Silva MJ, Huttly SR, Harpham T, Kenward MG. Social capital and mental health: a comparative analysis of four low income countries. Social Science and Medicine. 2007;64:5–20. doi: 10.1016/j.socscimed.2006.08.044. [DOI] [PubMed] [Google Scholar]
  10. De Silva MJ, McKenzie K, Harpham T, Huttly SRA. Social capital and mental illness: a systematic review. Journal of Epidemiology and Community Health. 2005;59:619–627. doi: 10.1136/jech.2004.029678. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. De Silva MJ. Systematic review of the methods used in studies of social capital and mental health. In: McKenzie K, Harpham T, editors. Social Capital and Mental Health. Jessica Kingsley Publishers; London: 2006. pp. 39–53. [Google Scholar]
  12. Fujiwara T, Kawachi I. A prospective study of individual-level social capital and major depression in the United States. Journal of Epidemiology and Community Health. 2008;62:627–633. doi: 10.1136/jech.2007.064261. [DOI] [PubMed] [Google Scholar]
  13. Hamad R, Fernald LCH, Karlan DS, Zinman J. Social and economic correlates of depressive symptoms and perceived stress in South African adults. Journal of Epidemiology and Community Health. 2008;62:538–544. doi: 10.1136/jech.2007.066191. [DOI] [PubMed] [Google Scholar]
  14. Harpham T. The measurement of community social capital through surveys. In: Kawachi I, Subramanian SV, Kim D, editors. Social Capital and Health. Springer; New York: 2008. pp. 51–62. [Google Scholar]
  15. Jung J, Ernstmann N, Nitzsche A, Driller E, Kowalski C, Lehner B, Stieler-Lorenz B, Friepörtner K, Schmidt A, Pfaff H. Exploring the association between social capital and depressive symptoms: results of a survey in German information and communication technology companies. Journal of Occupational and Environmental Medicine. 2012;54:23–30. doi: 10.1097/JOM.0b013e318237a1b6. [DOI] [PubMed] [Google Scholar]
  16. Kouvonen A, Oksanen T, Vahtera J, Stafford M, Wilkinson R, Schneider J, Väänänen A, Virtanen M, Cox SJ, Pentti J, Elovainio M, Kivimäki M. Low workplace social capital as a predictor of depression. American Journal of Epidemiology. 2008;167:1143–1151. doi: 10.1093/aje/kwn067. [DOI] [PubMed] [Google Scholar]
  17. Leibbrandt M, Woolard I, de Villiers L. Methodology: Report on NIDS Wave 1. University of Cape Town; Cape Town: 2009. [Google Scholar]
  18. Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJL. Global and regional burden of disease and risk factors, 2001: systematic analysis of population health data. The Lancet. 2006;367:1747–1757. doi: 10.1016/S0140-6736(06)68770-9. [DOI] [PubMed] [Google Scholar]
  19. Mitchell CU, LaGory M. Social capital and mental distress in an impoverished community. City and Community. 2002;1:199–222. [Google Scholar]
  20. Morrow V. Conceptualising social capital in relation to the well-being of children and young people: a critical review. The Sociological Review. 1999;47:744–765. [Google Scholar]
  21. Myer L, Smit J, Roux LL, Parker S, Stein DJ, Seedat S. Common mental disorders among HIV-infected individuals in South Africa: prevalence, predictors, and validation of brief psychiatric rating scales. AID Patient Care STDS. 2008;22:147–158. doi: 10.1089/apc.2007.0102. [DOI] [PubMed] [Google Scholar]
  22. O’Brien DJ, Hassinger EW, Dershem L. Community attachment and depression among residents in two rural midwestern communities. Rural Sociology. 1994;59:255–265. [Google Scholar]
  23. O’Connor M, Hawkins MT, Toumbourou JW, Sanson A, Letcher P, Olsson CA. The relationship between social capital and depression during the transition to adulthood. Australian Journal of Psychology. 2011;63:26–35. [Google Scholar]
  24. Oksanen T, Kouvonen A, Vahtera J, Virtanen M, Kivimäki M. Prospective study of workplace social capital and depression: are vertical and horizontal components equally important? Journal of Epidemiology and Community Health. 2010;64:684–689. doi: 10.1136/jech.2008.086074. [DOI] [PubMed] [Google Scholar]
  25. Putnam RD. Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press; Princeton, NJ: 1993. [Google Scholar]
  26. Radloff LS. The CES-D scale. Applied Psychological Measurement. 1977;1:385–401. [Google Scholar]
  27. Schensul D, Heller P. Legacies, change and transformation in the post-apartheid city: towards an urban sociological cartography. International Journal of Urban and Regional Research. 2010;35:78–109. [Google Scholar]
  28. Seekings J, Nattrass N. Class, Race, and Inequality in South Africa. Yale University Press; New Haven, CT: 2005. [Google Scholar]
  29. Shrout PE, Yager TJ. Reliability and validity of screening scales: effect of reducing scale length. Journal of Clinical Epidemiology. 1989;42:69–78. doi: 10.1016/0895-4356(89)90027-9. [DOI] [PubMed] [Google Scholar]
  30. StataCorp. Stata Statistical Software: Release 12. StataCorp LP; College Station, Texas: 2011. [Google Scholar]
  31. Steffick DE. HRS/AHEAD Documentation Report. Survey Research Center University of Michigan; Ann Arbor, Michigan: 2000. Documentation of Affective Functioning Measures in the Health and Retirement Study. [Google Scholar]
  32. Tomlinson M, Grimsrud AT, Stein DJ, Williams DR, Myer L. The epidemiology of major depression in South Africa: results from the South African stress and health study. South African Medical Journal. 2009;99:367–373. [PMC free article] [PubMed] [Google Scholar]
  33. Veenstra G. Location, location, location: contextual and compositional health effects of social capital in British Columbia, Canada. Social Science and Medicine. 2005;60:2059–2071. doi: 10.1016/j.socscimed.2004.08.064. [DOI] [PubMed] [Google Scholar]
  34. Wittenberg M. Weights: Report on NIDS Wave 1. University of Cape Town; Cape Town: 2009. [Google Scholar]

RESOURCES