ABSTRACT
Whereas people apparently spontaneously appreciate the impact of sample size in frequency-distribution tasks, they seem to have persistent difficulties in solving sampling-distribution tasks even when the process that creates the sampling distribution is explained to them (chapters 3 and 9; Sedlmeier, 1998b). This study was conducted to find out if the urn model, which is in accord with the adaptivealgorithms approach, and which proved to be effective in frequency-distribution tasks (chapter 10), can be successfully used to teach sample-size tasks involving sampling distributions. The current study employs two versions of sampling-distribution tasks. In choice tasks (e.g., the maternity ward task-see chapter 1) participants have to choose among the responses larger sample, smaller sample, and does not make a difference, whereas in construction tasks (e.g., the distribution-of-sexes task-see chapter 1), they have to estimate the number of sample means or proportions falling within a certain category or interval. In earlier research, construction tasks usually led to even lower spontaneous solution rates than did choice tasks (Fischhoff et al., 1979; Kahneman & Tversky, 1972; Olson, 1976; Teigen, 1974).
