POLS 208 Lecture Notes - Lecture 10: Random Assignment, Statistical Significance, Statistical Hypothesis Testing
Document Summary
In class question - in the iyengar et al (1982) study of agenda setting and media effects, the authors select a nonrandom sample of new haven resident to participate in their study. The sample of about 72 people is not representative of the population. In their studies, they randomly assign individuals to watch different news stories. Consider this alternative design: the researchers recruit the same people, give them a set of choices regarding which news stories to watch instead randomly assigning them to different news stories. They do everything else in the same manner as in the iyengar et al (1982) study (measurement, types of new stories, etc). What are the consequences of this problem? they are very vulnerable to bias, won"t distribute themselves randomly, which can skew the results to one side depending on the sample. In iyengar et al (1982), the authors find statistically significant effects of news stories on opinions about issues.