Article Using Analysis of Variance
Junker, Beeler and Bates (2005) made use of the Inmate Observer Program (IOP) for suicides in a federal correctional setting using the inmates as observers. The goal of the study was to determine the effects of using trained inmate observers on suicide watch. The dependent variable was the length of time that observed inmates were on the suicide watch. This variable (i.e., hours) was in ratio scale. There were two dependent variables: 1) the type of disorder of the observed inmates based on diagnosis (psychotic disorder, personality disorder, mood disorder) and 2) time (pre-IOP and post-IOP). These dependent variables were in categorical data and therefore in nominal scale. The study seeks to know whether or not there was difference in the length of suicide watch in the pre-IOP with post-IOP for the separate diagnostic categories.
The main effects for diagnosis and time were done separately per category. Length of suicide watch by type of disorder was tested using one-way analysis of variance and differences of means among types of disorder were done by t-tests. It shows that psychotic disorder had significantly longer suicide watch over other disorders. Length of suicide watch by time was done by t-test, where suicide watch was significantly higher prior to initiation of IOP. Interaction effects between type of disorder and time were determined by constructing an analysis of variance table, which rather showed the same conclusion that can be derived for the main effects. Graph showed there was significant decrease in suicide watch for psychotic disorders after initiation of IOP, which effect interacts with the slight increase of suicide watch for personality disorders. Analyzing the main effects separately was rather inappropriate as the chance of significant result would be much larger (Dallal, 2001). The ANOVA table was complete enough to show main and interaction effects. Pair-wise analysis among type of disorders can be done by post-hoc tests such as Tukey’s HSD , which is rather sensitive for unequal sizes (Dallal, 2001), which was most likely the case for this study.
Dallal, G.E. (2001). Multiple comparison procedures. Retrieved May 23, 2007. http://www.tufts.edu/~gdallal/mc.htm
Junker, G., Beller, A., & Bates, J. (2005). Using trained inmate observers for suicide watch in a federal correctional setting: A win-win solution. Psychological services. 2(1). 20-27.
No related essays.