When a meta-analysis on results from experimental studies is conducted, differences in the study design must be taken into consideration. A method for combining results across independent-groups and repeated measures designs is described, and the conditions under which such an analysis is appropriate are discussed. Combining results across designs requires that (a) all effect sizes be transformed into a common metric, (b) effect sizes from each design estimate the same treatment effect, and (c) meta-analysis procedures use design-specific estimates of sampling variance to reflect the precision of the effect size estimates.

Previous research has recommended several measures of effect size for studies with repeated measurements in both treatment and control groups. Three alternate effect size estimates were compared in terms of bias, precision, and robustness to heterogeneity of variance. The results favored an effect size based on the mean pre-post change in the treatment group minus the mean pre-post change in the control group, divided by the pooled pretest standard deviation.

Future research is needed to identify moderators of the effects of both education and contact.

People with serious mental illness have higher rates of mortality and morbidity due to physical illness. In part, this occurs because primary care and other health providers sometimes make decisions contrary to typical care standards. This might occur because providers endorse mental illness stigma, which seems inversely related to prior personal experience with mental illness and mental health care. In this study, 166 health care providers (42.2% primary care, 57.8% mental health practice) from the Veteran's Affairs (VA) medical system completed measures of stigma characteristics, expected adherence, and subsequent health decisions (referral to a specialist and refill pain prescription) about a male patient with schizophrenia who was seeking help for low back pain due to arthritis. Research participants reported comfort with previous mental health interventions. Path analyses showed participants who endorsed stigmatizing characteristics of the patient were more likely to believe he would not adhere to treatment and hence, less likely to refer to a specialist or refill his prescription. Endorsement of stigmatizing characteristics was inversely related to comfort with one's previous mental health care. Implications of these findings will inform a program meant to enhance VA provider attitudes about people with mental illness, as well as their health decisions.

When conducting a meta-analysis on studies with repeated measures, a useful measure of effect size is Becker's (1988) standardized mean change. This paper examines the distributional properties of the standardized mean change, and discusses potential problems with the variance formulae given in Becker (1988). First, an error in the exact variance formula can lead to severe underestimation of the actual sampling variance. Second, Becker (1988) recommended the use of an approximation which is shown to underestimate the sampling variance when sample size is small. Using the approximation can decrease the accuracy of meta-analysis results; however, the degree of error is generally small.