Some problems caused by not having a conceptual foundation for health research: An illustration from studies of the psychological effects of abortion

A quantitative review of 24 empirical studies on the psychological impact of abortions revealed a strong relationship between effect size and whether a pre/posttest design or a comparison group design was used. Although these relationships were overlooked by the authors of the reports reviewed, strikingly different conclusions were implied by the two sets of studies: pre/posttest designs appeared to show that abortions led to improved psychological state while comparison group designs showed little change or slightly negative outcomes. An examination of the possible reasons women may be affected by undergoing abortions shows that the relationship between the results and the research design selected could have been anticipated by an examination of the uncontrolled threats to internal validity. It is argued that meta-analytic methods help make such systematic biases apparent.