Inside the Black Box of Interactions Between Programs and Participants: Re-conceptualizing Subgroups for Fatherhood Program Evaluation. VI. FINDINGS: PSYCHOSOCIAL PREDICTORS OF FATHERHOOD-RELATED OUTCOMES

10/12/2012

In this section, we present findings from the second stage of the literature review on the extent to which theoretically informed psychosocial determinants of behavior change have been examined in research on low-income fathers, and what studies show. In Table VI.1, we summarize the number of studies that examined predictors in each of the categories from our conceptual framework and the number of studies that yielded statistically-significant results.

Table VI.1. Number of Studies Examining Each Set of Psychosocial Variables (and the Number Finding Statistically Significant Associations): Total, and by Fatherhood Outcome1

Total Studies

Number of studies examining outcomes

(Number with significant associations2)

Predictor Categories

Parenting

Partner Relationship

Employment

Child Support

Well-Being

Personal history

18

17 (13)

2 (1)

0

0

4 (2)

Identity

21

17 (11)

5 (4)

1 (1)

0

7 (6)

Values and lifestyle

14

9 (6)

2 (1)

1(1)

2(1)

2(1)

Stress and coping

26

21 (13)

5 (5)

0

0

7 (7)

Knowledge

1

1 (1)

0

0

0

0

Cognitions

15

10 (5)

3 (2)

1(1)

0

5 (2)

Social norms

3

3 (1)

0

0

0

0

Relationship with child's mother

36

27 (19)

5 (4)

2 (1)

1(1)

7 (3)

Social support

14

6 (4)

2 (2)

0

1 (1)

5 (5)

Total studies

64

51

11

2

1

15

Notes: 1. Some studies examined multiple predictors and/or multiple fatherhood outcomes.

2. We define results as statistically significant if the author reported a p-value less than or equal to 0.05.

 

In Appendix Tables E.1 to E.9, we present findings at the variable level for each category of predictors. We document whether a statistically significant association (that is, a p-value less than or equal to 0.05) was reported between each variable and the fatherhood outcome shown. We note whether studies utilized multivariate methods, such as multiple regression or structural equation modeling (as opposed to bivariate methods such as bivariate correlations or comparisons of unadjusted means). Although these methods are not sufficient for establishing a causal relationship, multivariate analyses provide stronger evidence of the relationship between a predictor and outcome because such methods help to minimize (though not eliminate) omitted variable bias; that is, controlling for other variables that are correlated with both the predictor and outcome of interest helps to isolate each predictor's unique effects.[2] We also note whether analyses were longitudinal; that is, whether the predictor was measured in a time period before the outcome was observed (prospective) or if information about a predictor was collected at the same time as the outcome but refers to a time period preceding it (retrospective). Evidence of prediction is stronger in longitudinal studies because the direction of the effect is clearer. Findings from prospective longitudinal studies provide especially strong evidence of prediction given recall errors that can plague retrospective longitudinal studies.

When presenting findings, we discuss all findings from studies using both multivariate and longitudinal methods because such studies provide stronger evidence of relationships between predictors and outcomes. Studies using only multivariate methods are discussed selectively due to space limitations; however, in providing illustrative examples of findings from multivariate studies we are careful to place these findings in context of other studies' findings (or non-findings) to reflect the weight of the evidence across all studies. Studies using only bivariate methods (including longitudinal studies with no statistical controls) are not discussed, as these findings may overstate the importance of the predictor or, worse, may reflect spurious associations reflecting the effects of another variable altogether. In discussing cross-sectional studies, we typically describe findings as "associations" or "correlations," even as we continue to refer to the psychosocial factors examined as "predictors" of fatherhood outcomes.

View full report

Preview
Download

"rpt_insidebb.pdf" (pdf, 1.35Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®