Mary Fischer

A Longitudinal Study of Graduates of Special Education Preschools: Educational Placement After Preschool

The purpose of this study was to continue the collection of data on graduates of public special education programs in Washington State to determine the stability of initial placement. Placement data were collected on 582 children from three cohorts of children who left preschools in 10 Washington State school districts. Data indicate that 36% of these children were initially placed in regular education settings and 64% were placed in special education or resource settings. The stability of these placements 1 year and 2 years after graduation were high, with stability rates ranging from 75% to 100%. Of the children who made placement changes, 28% moved to less restrictive settings. Further questions to be asked as the data collection continues over the next few years are whether the current trends will continue and if placement in a less restrictive setting leads to better achievement.

Co-twin Control Methods†

The co-twin control method is an elegant research design that uses differences within twin pairs to examine the association between a putative environmental risk factor and an outcome variable. The design is extremely efficient for examining risk factor–outcome associations compared to unmatched designs. This is especially true when using monozygotic (MZ) twins who share 100% of their genetic material. Co-twin control studies can be used in both the experimental and observational settings. Statistical analysis for continuously distributed outcome variables focuses on the within-pair mean difference, while for dichotomous outcomes, the matched pair odds ratio is used. Whenever possible, it is informative to examine differences in the relationship between the risk factor–outcome association by zygosity; findings that are only present in dizygotic (DZ) pairs suggest that the observed association may be due to genetic influence on both the risk factor and outcome. We present an illustration of the analysis of a co-twin control study, looking at the association between military service in Vietnam and posttraumatic stress disorder, with data from the Vietnam Era Twin Registry. A wide range of new regression modeling methods, which account for nonindependent observations, can be readily adapted for co-twin control studies. Keywords: matched pair; confounding; odds ratio; within-pair difference