|
 |
|
|||||||
|
|||||||||
© 2004 American Public Health Association
RESEARCH AND PRACTICE |
At the time the study began, Eric Dearing was with Judge Baker Children’s Center, Harvard Medical School, Boston, Mass; he now is with the Department of Psychology, University of Wyoming, Laramie. At the time the study began, Beck A. Taylor was with the Harvard Graduate School of Education, Cambridge, Mass; he now is with Baylor University, Waco, Tex. Kathleen McCartney is with the Harvard Graduate School of Education.
Correspondence: Requests for reprints should be sent to Eric Dearing, Department of Psychology, Biological Sciences, University of Wyoming, Laramie, WY 82071 (e-mail: deariner{at}uwyo.edu; beck_taylor{at}baylor.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. We examined within-person associations between changes in family income and women’s depressive symptoms during the first 3 years after childbirth.
Methods. Data were analyzed for 1351 women (mean baseline age = 28.13 years) who participated in the National Institute of Child Health and Human Development Study of Early Child Care. Nineteen percent of these women belonged to an ethnic minority, and 35% were poor at some time during the study.
Results. Changes in income and poverty status were significantly associated with changes in depressive symptoms. Effects were greatest for chronically poor women and for women who perceived fewer costs associated with their employment.
Conclusions. Given that women head most poor households in the United States, our findings indicate that reductions in poverty would have mental health benefits for women and families.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Researchers interested in links between poverty and depression have begun using longitudinal data to account for volatility in family income.1–4 For example, it has been demonstrated that adults who are chronically poor are at greater risk for developing depression later in life than those who are transiently poor, when researchers control for current psychological functioning.4 Because longitudinal studies have generally relied on between-person comparisons (e.g., rates of depression for chronically poor vs transiently poor populations), it remains unclear whether economic resource changes are associated with depressive symptom changes within individuals. If a person’s income rises, then do depressive symptoms diminish? Conversely, if income falls, then do depressive symptoms increase? Longitudinal analyses of within-person associations between income and depressive symptoms would answer these questions by estimating whether individuals’ income and depressive symptoms covary over time.
In our study, within-person associations between changes in family income and changes in women’s depressive symptoms were examined periodically from 1 to 36 months after giving birth. A large body of literature has documented an increased risk of depression among women compared with men.5 The postpartum period is a time of heightened vulnerability, in part because of hormonal changes that increase women’s psychological reactivity to high-stress conditions.6,7 In fact, approximately 10% of postpartum women experience clinical levels of depressive symptoms within the first few weeks after delivery, with the majority of these episodes lasting 6 months or less.8 However, postpartum depression appears similar to depressive disorders occurring at other times in life with regard to both symptoms and precursors.9
In general, depressive episodes may be transient, lasting a period of days, or chronic, lasting years.7,10 Life stressors such as financial strain and marital discord, as well as previous depressive episodes, are associated with an increased risk for depression, regardless of timing.10 Yet the public health relevance of women’s depression during the first 3 years after childbirth is exceptional, primarily because maternal depression during infancy and early childhood has been well documented as a risk factor for children’s social-emotional development.11–13
Not surprisingly, women living in lowincome families are more likely than other women to be exposed to high-stress living conditions such as overcrowding, noise, and violent communities.14 Thus, we predicted that changes in family income would be associated with changes in women’s depressive symptoms throughout the first 3 years after childbirth, such that increases in income would be linked with decreases in depressive symptoms. We also predicted that changes in poverty status would be associated with changes in depressive symptoms, such that moving out of poverty would be associated with decreases in depressive symptoms, above and beyond the effects of the corresponding income changes.
Although reciprocal causation between income and depressive symptoms is possible, examining the role of employment offers the opportunity to test the direction of effect. Employment changes are, in fact, the most common reason for income gains and losses among poor families.15 If income directly influences depression, then changes in hours of employment should be indirectly related to symptom changes. That is, associations between employment and depression should be due to income gains or losses resulting from work changes. However, if depression influences income, then events that affect earnings, such as hours of employment, should mediate the link. We examined 2 pathways linking changes in hours of employment, income, and depressive symptoms: (1) changes in hours of employment 
changes in income changes in depressive symptoms, and (2) changes in depressive symptoms changes in hours of employment changes in income. We predicted that our results would be consistent with the first pathway. Thus, we expected our results to be more consistent with social causation theories in which economic status is hypothesized to causally influence mental health rather than health selection theories in which mental health is hypothesized to causally influence economic status.16–17
Interactions between income changes and characteristics of women, their children, and their families were also examined to determine whether the association between changes in income and depressive symptoms varied across women. We expected the association between income and depressive symptoms to be larger for poor women than for nonpoor women, primarily because income gains and losses would have greater relative impacts on the economic resources of poor families (e.g., a $10 000 increase in income would be a 100% gain for families earning $10 000 per year and a 20% gain for families earning $50 000 per year).18 We also expected the association to be larger for women who believed that the costs (i.e., detrimental effects) of maternal employment were low for their children compared with women who believed these costs were high. That is, we predicted that the positive psychological impact of income gains and the negative psychological impact of income losses would be limited by the belief that maternal employment is harmful to children, a belief that has been associated with low rates of maternal employment.19 Thus, the goal of the present study was to examine pathways linking within-person changes in income and women’s depressive symptoms during the first 3 years after childbirth, as well as variations in the association between income and depressive symptoms across demographic and psychological characteristics of women.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Sample
Data from the first phase of the National Institute of Child Health and Human Development Study of Early Child Care (SECC) were used in this investigation. Shortly after giving birth in 1991, 1364 women living in or near 10 urban and suburban sites in the United States were recruited to participate in this study; a conditional random sampling method was used (recruitment and sampling details have been published by the Early Child Care Research Network).20–22 Of this group, 99% (1351) had sufficient nonmissing data for analysis. Designed to study the developmental implications of early child care, the first phase of the SECC includes longitudinal data (collected at 1, 6, 15, 24, and 36 months postpartum) on family economic and psychological well-being. Sample demographics are presented in Table 1
.
|
Measures
Demographics. At 1 month postpartum, women reported their age, ethnicity, years of education, and child’s gender. At all 5 measurement occasions, mothers reported their partner status and number of children living in the home. Ethnicity (African American vs Other and Latino American vs Other), child’s gender, and partner status were coded as dummy variables.
Family Income and Hours of Employment. At all 5 measurement occasions, women reported their total household income, which for purposes of analysis was divided by $10 000. At these time points, women also reported their own and their partner’s weekly hours of employment from all jobs. The ratio of family income to family needs, an index often used by poverty researchers, was computed by dividing total family income by the poverty threshold for the appropriate family size.23,24 Women with income-to-needs ratios less than 1 at 3 or more assessments were coded as chronically poor (15% of sample). Women with income-to-needs ratios less than 1 at only 1 or 2 assessments were coded as transiently poor (20% of sample).
Maternal Depressive Symptoms. At all 5 measurement occasions, women completed the Center for Epidemiological Studies Depression Scale (CES-D), one of the most widely used measures of depressive symptoms. The 20-item checklist measures the presence and frequency of depressive symptoms during the previous week.25 Note that scores on the CES-D can range from 0 to 60, and scores of 16 or higher are generally associated with clinical depression. In standardization samples, reliabilities ranged from 0.84 to 0.90.25 Validity has been established by means of correlations with clinical diagnoses.26 In the SECC sample, reliability ranged from 0.85 to 0.90.
Costs of Maternal Employment.
At 1 month, women completed the Beliefs About the Consequences of Maternal Employment for Children, an 11-item scale.27 Items were summed so that higher scores reflected greater perceived costs to children (mean = 34.14, SD = 7.13). The scale’s validity has been established by means of significant associations with women’s employment status and gender-role traditionalism.19,28 In the SECC sample, this measure was internally consistent (
= .94).
Statistical Analyses
Hierarchical linear modeling is an extension of repeated measures analysis of variance, with the advantage that change at the level of the individual is estimated directly rather than by means of the interaction of time by subject.29 Considering the longitudinal design of the SECC and our study questions regarding within-person change, hierarchical linear modeling is an ideal method for analyzing these data. Two-level hierarchical linear modeling was used to estimate trajectories of depressive symptoms from 1 to 36 months postpartum and within-person associations between changes in family income and changes in depressive symptoms. In the first level of analysis, linear and nonlinear changes in depressive symptoms were estimated, as well as associations between time-varying predictors (e.g., changes in income) and changes in depressive symptoms. In this first level of analysis, predictors were centered around their group means so that associations between predictors and outcomes were within-person estimates. In the second level of analysis, a set of time-invariant predictors of changes in depressive symptoms was estimated. These predictors were used to examine cross-level interactions. For example, we estimated variations in the association between changes in income and changes in depressive symptoms as a function of whether women had been chronically, transiently, or never poor.
Because the meaningfulness of changes in CES-D scores may not be intuitive when coded as a continuous scale, a conditional (i.e., fixed-effects) logistic regression model was used to estimate the likelihood of within-person changes in clinical depression status (a dichotomous indicator) as a function of changes in income and poverty status.30 Recall that CES-D scores of 16 or higher are indicative of clinical depression. Two aspects of this analysis are important to highlight. First, the outcome variable was a dummy variable (i.e., depressive symptoms below the clinical threshold vs depressive symptoms above the clinical threshold). Thus, the estimated conditional logistic regression model coefficients were interpreted as associations between changes in the predictor variables (e.g., income) and changes in the odds of experiencing an episode of clinical depression. Second, conditional logistic regression model estimates were interpreted correctly as average, within-person estimates (e.g., the average within-person association between changes in income and changes in the odds of experiencing clinical depression).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Time-Varying Predictors of Depressive Symptoms: Mediation Pathways
To provide a descriptive account of change in income and change in depressive symptoms from 1 to 36 months postpartum, 2 hierarchical linear models were estimated. The first estimated patterns of change in income, and the second estimated patterns of change in depressive symptoms. Both income and depressive symptoms significantly varied between 1 and 36 months postpartum, such that there were significant linear, quadratic, and cubic time trends. Sample averages for these time trends are illustrated in Figure 1
. Note that depressive symptoms and family income covaried negatively over time. In fact, the 2 curves appear to be mirror images. In addition to significant group averages, patterns of change in both depressive symptoms and income significantly varied across women (e.g., for linear change in income, 
2 = 2285.09, P < .001, and for linear change in depressive symptoms, 2 = 1603.52, P < .001). Some women, for example, experienced greater gains in income than the sample average, and others experienced losses.
|
Next, a series of hierarchical linear models was estimated that examined 2 possible pathways linking income, hours of employment, and depressive symptoms. Pathways were estimated using the product of the coefficients approach to testing mediation.31 To test the first pathway (i.e., hours of employment
family income depressive symptoms), 2 models were specified: (specification 1) women’s and their partner’s hours of employment were estimated as time-varying predictors of family income, and (specification 2) family income and hours of employment (both women’s and partner’s) were simultaneously estimated as time-varying predictors of depressive symptoms. The products of the coefficients for maternal and partner hours of employment in specification 1 and income in specification 2 were estimated using the Sobel test of indirect effects (i.e., ab divided by the square root of b2sa2 + a2sb2 — sa2sb2, where a represents the association between hours of employment and family income, b represents the association between family income and depressive symptoms, sa represents the standard error of a, and sb represents the standard error of b; resulting values were treated as z-test statistics).31
Coefficients, standard errors, and P values from specifications 1 and 2 are displayed in Table 2. Note that changes in partner status, changes in number of children in the home, and time parameters were included in these models as time-varying covariates. With respect to interpretation, the estimated coefficients for hours of employment represent the average change in depressive symptoms resulting from a 1-hour-per-week increase in employment; the estimated coefficient for income represents the average change in depressive symptoms resulting from a $10 000 increase in annualized income.
|
In specification 1, increases in both mothers’ and partners’ hours of employment were significantly associated with increases in family income (i.e., coefficients of 0.04 and 0.02). In specification 2, increases in family income were significantly associated with decreases in depressive symptoms (coefficient of –0.14). Note, however, that when we controlled for family income, neither maternal nor partners’ hours of employment were significantly associated with depressive symptoms in specification 2. These results were consistent with the first pathway: change in hours of employment
change in family income change in depressive symptoms. In fact, the products of the coefficients for employment (specification 1) and income (specification 2) were significantly different from zero (i.e., for maternal employment, z = –3.45, P < .001; for partner’s employment, z = –3.30, P = .001). In other words, the path from hours of employment to income and the path from income to depressive symptoms were estimated to be jointly significant, providing evidence that family income mediates the link between hours of employment and depressive symptoms.31
To test the second pathway (i.e., change in depressive symptoms change in hours of employment change in family income), 2 models were specified: (1) depressive symptoms as a time-varying predictor of hours of employment, and (2) depressive symptoms and hours of employment simultaneously estimated as time-varying predictors of family income. Family income was a significant predictor of depressive symptoms, even when we controlled for maternal and partners’ hours of employment. Further, the path from depressive symptoms to hours of employment and the path from hours of employment to family income were not jointly significant. Therefore, these results were not consistent with the second pathway.
Poverty Status.
To determine whether changes in poverty status were associated with changes in depressive symptoms above and beyond the effects of hours of employment and family income, change in poverty status was added to the 8 predictors from specification 2 in Table 2
. Changes in poverty status were associated with changes in depressive symptoms (coefficient = 0.55, SE = 0.30), above and beyond accompanying changes in income and employment. However, this association was only marginally significant (P = .07).
Changes in Clinical Depression.
To further investigate the public health significance of family economic changes, a conditional logistic model was used to estimate the likelihood of within-person changes in clinical depression status as a function of the time-varying predictors including family income and poverty status (Table 3). Increases in income were significantly associated with the odds of a change from clinical to nonclinical status. When we controlled for the effect of income, women who moved out of poverty were 1.48 times more likely to experience a shift from clinical to nonclinical depression status than if they had remained in poverty. Thus, changes across the poverty threshold increased the probability of within-person changes across the clinical depression threshold.
|
A 2-level Hierarchical Linear Model: Interaction Effects
As a final step, a 2-level hierarchical linear model of depressive symptoms was estimated to examine interactions between time-varying and time-invariant predictors (i.e., chronic poverty, transient poverty, maternal education, costs of maternal employment, maternal age, maternal ethnicity, and child’s gender). The magnitude of the association between changes in family income and depressive symptoms significantly varied by 2 time-invariant predictors: chronic poverty and costs of maternal employment. The negative association between changes in income and depressive symptoms was significantly larger for chronically poor women (i.e., estimated income coefficient = –0.70) compared with never-poor women (i.e., estimated income coefficient = –0.12), even when we controlled for the fact that both transiently and chronically poor women reported, on average, more depressive symptoms than never-poor women. Note, however, that for both groups, the association between changes in income and changes in depressive symptoms was significantly different from zero.
In addition, the association between changes in family income and depressive symptoms was significantly smaller for women who reported relatively high costs of maternal employment (estimated income coefficient = –0.11 for women who were 1 SD above the mean costs) compared with women who reported relatively low costs (estimated income coefficient = –0.29 for women who were 1 SD below the mean costs). However regardless of women’s beliefs about employment, the association between changes in income and changes in depressive symptoms was statistically significant.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
With approximately 17% of families in the United States living in poverty and most of these households headed by women, the mental health of poor women remains a pressing topic for both public health science and public health policy.32 The present study extends previous work by demonstrating that changes in income and poverty status were associated with changes in women’s depressive symptoms in the first 3 years after childbirth. Income gains resulted in the alleviation of symptoms, especially when these gains were substantial enough to lift families out of poverty. In fact, women were 1.48 times more likely to experience a shift from clinical to nonclinical status after transitions out of poverty. There is now between-person1–4 and within-person evidence that changes in family economics are associated with changes in depression.
In the present study, changes in employment were indirectly associated with depressive symptoms by means of changes in income. These results were consistent with a path of influence leading from employment changes to income changes, and, in turn, to depressive symptom changes. There was no evidence that depressive symptom changes influenced income by means of employment changes. Therefore, our results were consistent with social causation hypotheses.16,17 During the first 3 years after childbirth, women’s economic well-being appeared to influence their psychological well-being, rather than the opposite.
Further, associations between income and depressive symptoms varied by demographic characteristics of women. Women who were chronically poor experienced the strongest effects of changes in income on their depressive symptoms, perhaps because income gains and losses for these women were associated with the largest relative changes in economic well-being. Policies that increase the economic resources of these women are likely to have the greatest mental health impacts. The associations between income and women’s depressive symptoms also varied by women’s concerns about the ramifications of maternal work. The belief that work was costly to children limited the ameliorative effects of income gains and the deleterious effects of income losses.
Although some researchers have failed to detect intervening effects of income between employment and depressive symptoms, their results were based on between-person residual change analyses (i.e., regressing depressive symptoms at time 2 on depressive symptoms at time 1 along with employment and income indicators).3 The within-person analyses of change estimated in the present study are preferable, because they avoid statistical problems that plague residual estimates of change (i.e., biased, imprecise, and unreliable coefficients) and because changes in employment, income, and depressive symptoms have been linked within individuals rather than across individuals.33
However, within-person analyses that capture more of the life course may help further disentangle the links between income and depression for women with children. Data both before and after childbirth would be particularly helpful in this regard. In addition, inclusion of dynamic processes such as social support, marital quality, and partner’s depression would be useful controls to the extent that changes in these variables may influence both income and women’s depression. It is also important to note that postpartum depression, per se, was not uniquely identifiable in the present study, although past research has indicated that economic strain is associated with an increased risk of depression, regardless of timing.10
Taken together, the results of our study indicate that increasing the economic resources of women living in poverty would have substantial mental health benefits during the first few years after childbirth. These findings are of added relevance to public health considering that maternal depression poses a substantial risk to the psychological well-being of children during the first 3 years of life.11–13 Intervention efforts that are sensitive to both demographic and psychological characteristics of individuals are likely to be most successful. More specifically, our results indicate that policies and interventions targeting chronically poor mothers will yield the greatest improvements in public health, especially if these efforts lead to financial gains without increased child-rearing anxieties. Future studies of the mechanisms by which changes in income and poverty status lead to changes in depressive symptoms could further inform intervention efforts.
|
Acknowledgments |
|---|
This research was supported in part by a postdoctoral fellowship awarded to E. Dearing from the National Academy of Education/Spencer Foundation and a grant to K. McCartney from the National Institute of Child Health and Human Development (HD 25451).
Human Participant Protection
This study was approved by the Judge Baker Children’s Center institutional review board and was provided an exemption by the University of Wyoming institutional review board because it was secondary data analysis.
|
Footnotes |
|---|
Contributors
E. Dearing had primary responsibility for data analysis and writing the article. B. A. Taylor also contributed to the data analysis and writing, particularly with regard to the conditional logistic models. K. McCartney contributed to writing the article.
Accepted for publication July 18, 2003.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
1. Miech RA, Shanahan MJ. Socioeconomic status and depression over the life course. J Health Soc Behav. 2000;41:162–176.[Web of Science]
2. Murphy JM, Olivier DC, Monson RR, Sobol AM, Federman EB, Leighton AH. Depression and anxiety in relation to social status. Arch Gen Psychiatry. 1991;48:223–229.
3. Dooley D, Prause J, Ham-Rowbottom KA. Underemployment and depression: longitudinal relationships. J Health Soc Behav. 2000;41:421–436.[Web of Science][Medline]
4. Lynch JW, Kaplan GA, Shema SJ. Cumulative impact of sustained economic hardship on physical, cognitive, psychological, and social functioning. N Engl J Med. 1997;337:1889–1895.
5. Culbertson FM. Depression and gender: an international review. Am Pschol. 1997;52:25–31.
6. Miller LJ. Beyond the "blues": hypotheses about postpartum reactivity. In: Miller LJ, ed. Postpartum Mood Disorders. Washington, DC: American Psychiatric Press; 1999:3–19.
7. Miller LJ. Postpartum depression. JAMA. 2002;287:762–765.
8. Cooper PJ, Murray L. Course and recurrence of postnatal depression: evidence for the specificity of the diagnostic concept. Br J Psychiatry. 1995;166:191–195.
9. Cooper PJ, Murray L, Stein A. Postnatal depression. In: Seva A, ed. European Handbook of Psychiatry and Mental Health. Barcelona, Spain: Anthropos; 1991:1255–1262.
10. Keller MB, Klerman GL, Lavori PW, Coryell W, Endicott J, Taylor J. Long-term outcome of episodes of major depression: clinical and public health significance. JAMA. 1984;252:788–792.
11. Lyons-Ruth K, Easterbrooks MA, Cibelli CD. Infant attachment strategies, infant mental lag, and maternal depressive symptoms: predictors of internalizing and externalizing problems at age 7. Dev Psychol. 1997;33:681–692.[Web of Science][Medline]
12. Field T. Psychologically depressed parents. In: Bornstein MH, ed. Applied and Practical Parenting. Mahwah, NJ: Lawrence Erlbaum; 1995:85–100. Handbook of Parenting; Vol 4.
13. Petterson SM, Albers AB. Effects of poverty and maternal depression on early child development. Child Dev. 2001;72:1794–1813.[Web of Science][Medline]
14. Evans GW, English K. The environment of poverty: multiple stressor exposure, psychophysiological stress, and socioemotional adjustment. Child Dev. 2002;73:1238–1248.[Web of Science][Medline]
15. Corcoran ME, Chaudry A. The dynamics of childhood poverty. Future Child. 1997;7(2):40–54.[Web of Science][Medline]
16. Dohrenwend BP, Doherwend BS. Social Status and Psychological Disorder: A Causal Inquiry. New York: John Wiley & Sons; 1969.
17. Ritsher JE, Warner V, Johnson JG, Dohrenwend BP. Inter-generational longitudinal study of social class and depression: a test of social causation and social selection models. Br J Psychiatry. 2001;40:84–90.
18. Taylor BA, Dearing E, McCartney K. Incomes and outcomes in early childhood. J Hum Resources. In press.
19. Greenberger E, O’Neil R. Parents’ concern about their child’s development: implications for fathers’ and mothers’ well-being and attitudes toward work. J Marriage Fam. 1990;52:621–635.[Web of Science]
20. National Institute of Child Health Early Child Care Research Network. Nonmaternal care and family factors in early development: an overview of the National Institute of Child Health Study of Early Child Care. Appl Dev Psychol. 2001;22:457–492.
21. National Institute of Child Health Study of Early Child Care and Youth Development. Online instrument documentation for the National Institute of Child Health Study of Early Child Care and Youth Development. Available at: http://public.rti.public.org/secc. Accessed October 6, 2002.
22. National Institute of Child Health Early Child Care Research Network. Child outcomes when child care center classes meet recommended standards for quality. Am J Public Health. 1999;89:1072–1077.
23. US Census Bureau. Poverty Thresholds. Available at: http://landview.census.gov/hhes/poverty/histpov/hstpov1.html. Accessed May 16, 2003.
24. Dearing E, McCartney K, Taylor BA. Change in family income-to-needs matters more for children with less. Child Dev. 2001;72:1779–1793.[Web of Science][Medline]
25. Radloff LS. The CES-D Scale: a self-report depression scale for research in the general population. Appl Psychol Measurement. 1977;1:385–401.
26. Cho MJ, Moscicki EK, Narrow WE, Rae DS. Concordance between two measures of depression in the Hispanic Health and Nutrition Examination Survey. Soc Psychiatry Psychiatr Epidemiol. 1993;28:156–163.[Web of Science][Medline]
27. Greenberger E, Goldberg WA, Crawford TJ, Granger J. Beliefs about the consequences of maternal employment for children. Psychol Women Q. 1988;12:35–59.
28. Goldberg WA, Greenberger E, Hamill S, O’Neil R. Role demands in the lives of employed single mothers with preschoolers. J Fam Issues. 1992;13:312–333.
29. Bryk AS, Raudenbush SW. Hierarchical linear models: applications and data analysis methods. Advanced Quantitative Techniques in the Social Sciences. Vol 1. Newbury Park, Calif: Sage Publications; 1992.
30. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons; 2001.
31. MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V. A comparison of methods to test mediation and other intervening variable effects. Psychol Methods. 2002;7:83–104.[Web of Science][Medline]
32. US Census Bureau. 2001 Poverty Statistics. Available at: http://www.census.gov/hhes/www/poverty01.html. Accessed October 6, 2003.
33. Rogosa D. Myths and methods: "Myths about longitudinal research" plus supplemental questions. In: Gottman JM, ed. The Analysis of Change. Mahwah, NJ: Lawrence Erlbaum; 1995:3–66.
This article has been cited by other articles:
|
|
 |
|
  A. L. Kozyrskyj, G. E. Kendall, P. Jacoby, P. D. Sly, and S. R. Zubrick Association Between Socioeconomic Status and the Development of Asthma: Analyses of Income Trajectories Am J Public Health, March 1, 2010; 100(3): 540 - 546. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. J. Conron, W. Beardslee, K. C. Koenen, S. L. Buka, and S. L. Gortmaker A Longitudinal Study of Maternal Depression and Child Maltreatment in a National Sample of Families Investigated by Child Protective Services Arch Pediatr Adolesc Med, October 1, 2009; 163(10): 922 - 930. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Kurtz Landy, W. Sword, and R. Valaitis The Experiences of Socioeconomically Disadvantaged Postpartum Women in the First 4 Weeks at Home Qual Health Res, February 1, 2009; 19(2): 194 - 206. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |