Abstract
The high school completion rate for the U.S. has increased only slightly over the last quarter of a century. A promising area of focus for dropout prevention efforts may be found in research that assesses how messages about educational attainment may affect school completion rates. The supportive communication framework is used to address how communication in the home and in school may affect student educational attainment. An extensive search of the student dropout literature was followed by standard meta-analytic procedures. The average weighted effect for communication in the home on high school completion was r=.21, and the effect for communication in school on high school completion was r=.14. The age of subjects and region of the country in which subjects lived were addressed as possible moderators of the relationship between communication in the home and school and educational attainment. No evidence that these variables moderated the relationship was found.
This manuscript is based on the first author's preliminary paper
This manuscript is based on the first author's preliminary paper
Notes
This manuscript is based on the first author's preliminary paper
1. Conventionally, race was thought to be the most important predictor of school dropout, but research has demonstrated that when socioeconomic status is held constant, race no longer matters (Kaufman & Bradby, 1992; Rumberger, 1995).
2. Fifteen additional studies fit the criteria for inclusion in this review, but the statistics needed to calculate an effect size were not available in these manuscripts. Thus, they could not be included in the quantitative analysis.
3. In order to assess the reliability of items for the effects sizes calculated, correlations of effects with the number of items for each effect were run. These correlations were not sufficiently large. Thus, the variance found in the effect sizes cannot be attributed to differential reliability in the items.
4. The average weighted effect for communication in the home (without the Velez study, r=−.18) on school completion was r=.23 (unweighted r=.23), k = 8 (number of studies), N=10,582, =.009 (weighted variance), =0007 (variance due to sampling error). This sample of correlations was not homogeneous, χ2(7, N=10,582) = 106.17, p<.05.