# 1Questions about Mixed Model ANOVA Autism and Risperidone

.docx
Math 227 Name(s):____________________________________________________ Article Questions Answer the following questions about the article Aman, M.G. et. al. (2008). Cognitive effects of Risperidone in children with autism and irritable behavior. Journal of Child and Adolescent Psychopharmacology 18(3), 227 - 236. 1. What is the population to which the authors generalize their results? The population to which the authors generalize their results from was children and adolescents with mental ages greater than 18 months who had autism and severe behavioral disturbance. 2. What is the sampling frame? The sampling frame was 38 people who had scorable cognitive measures at a baseline and at least had one visit out of the 101 study participants found. 3. What are three things about how the sample was selected from the sampling frame that could be improved in order to better generalize from the sample to the population? Can be improved in the variety of the severity of their autism, not using all children that have sever behavioral disturbances, and probably changing mental age to not start at 18 months but around 48 months and greater. 4. Give an example of a variable that was omitted from the study but might have been worthwhile to include. An example of a variable that was omitted from the study that might have been worthwhile to include was children with past medical history of neuroleptic malignant syndrome. 5. How many ANOVA's were performed in this study? What type of ANOVA was performed? What were the factors and how many levels did they have? How could they have done all of these ANOVA's as a single analysis?
There was two ANOVAS performed in this study. They completed a repeated measures ANOVA. The factors were cognitive data was analyzed by function of drugs, placebo vs risperidone, which had two levels and by time which had 3 levels, within participants. 6. Looking at the statistics provided in Table 1 for the variable AGE, can you tell what the shape of the distribution is (negatively skewed, roughly symmetric, or positively skewed)? Explain. The shape of the distribution was positively skewed because the testable participants had a higher mean average of 9.0 versus the mean of the untestable participants at 7.0. 7. On page 230, in the section on "Cognitive measures," the authors report that the partial eta squared for the interaction between time and drug for the Cancellation Task is .16. Interpret that value. According to the value of partial eta squared np^2 =.16, approximately the Risperidone resulted in a large effect om more correct detections than placebo on the Cancellation Task 8. Although the authors state in the "Study Participants" section on page 229 that the sample size is 38, in two places the inferred sample sizes are different from 38. On page 230, in the section on "Cognitive measures", the authors report the results for the interaction between time and drug for the Cancellation Task as F (1, 17) = 3.18, p = .05. On page 232 in Table 2 the sample sizes are given in the footnotes. What is the sample size according to these two analyses? Explain. The sample size is 34 because not everyone was able to complete cognitive testing. 9. On page 230, in the section on "Cognitive measures", the authors mention that they will study the drug effects by looking at the interaction between time and drug. In fact, they hope for an interaction between the two. Why is that? In other words, why not just report the main effect of drug?
There hoping for interaction in order understand the drug and time to figure out if it has correlation to schizophrenia treatment in adults. 10. In the "Limitations" section on page 234, the authors wrote, "Had we corrected for multiple comparisons, none of the comparisons would have exceeded alpha." What did they mean? They mean that if they would have corrected, they would see any real comparison difference, and probably would see any affects between the interactions shown. Probably would have resulted in false positive answers if corrected, not true values shown.