An integrative research review: Meta-analysis of psychological characteristics of adolescent pregnancy 1964 through 1994




Gilliam, Stephen

Journal Title

Journal ISSN

Volume Title



Pregnancy is a normal and healthy event in the life of a woman; for an adolescent , pregnancy is a complex event adding to a demanding time of life. For this reason many researchers from divergent fields of study have focused on the phenomenon . The purpose of this research was to collect and summarize the available research and to "determine what can be said with confidence" about psychosocial aspects of the phenomenon . This research addressed the question: In research from 1964 through 1994 , what are the relative effect sizes of psychosocial factors influencing adolescent pregnancy , and do demographic attributes of study participants or study characteristics moderate these effects? A collection of 290 research reports were identified from the literature that dealt with psychosocial aspects of adolescent pregnancy . Inclusion criteria of a control group narrowed the field of studies included in the review to 68 which represented 12,106 subjects including 3,881 pregnant teens. Conceptually similar variables from the 68 studies were grouped into 31 clusters which were subjected to a comprehensive analysis. This analysis included but was not limited to: determination of frequency, mean and standard deviation of study and sample characteristics, Weighted Effect Size (zr), and 95% Confidence •Interval. After hypothesis testing and homogeneity analysis, the cluster variables that remained and were most strongly correlated with the pregnant adolescents included: an identification with traditional female roles (Zr = 0.45), positive beliefs about parenting (Zr = 0.15), and sexual activity (zr = 0.14). The cluster variables most strongly correlated with the non-pregnant control group were contraception use (Zr = 0.16), educational expectations (Zr = 0.21), future orientation (Zr = 0.15), school grades (Zr 0.24), and occupational expectations (Zr = 0.18). During the meta-analysis of each cluster, study characteristics and study subject demographic variables were analyzed as potential moderator variables. Moderator variables indicate the need to look for sources of variance within a meta-analysis other than the cluster variable. No pattern of variables were found to act as moderators across all or groups of the clusters. The implications of moderators were briefly considered; however, theoretical inference was left for future research.



Health and environmental sciences, Psychology, Teen pregnancy, At-risk teens