1. A researcher has an automobile driving simulator in which the difficulty of the driving tasks encountered can be controlled and manipulated by the computer that controls the simulator. The researcher wants to test two hypotheses: (1) that alcohol drinking disrupts driving performance on difficult conditions more than it disrupts performance in easier conditions, and (2) that for difficult conditions, the more alcohol consumed the greater the overall disruption in performance. She chooses to use a mixed design, with alcohol condition manipulated as a between-subjects factor and driving condition manipulated as a within-subjects factor. Subjects were randomly selected and randomly assigned to the different alcohol treatment groups formed, with ten subjects in each group. There were four levels of alcohol consumption: no alcohol placebo drink, one drink, two drinks, and four drinks. There were two driving conditions: easy and difficult. The computer automatically recorded the number of mistakes (driving errors) during the 10-minute simulated driving session in each driving condition, starting 30 minutes after the drink(s) were consumed. Answer each of the following questions. (25 points)

A. What is your DV? What type of measurement scale does it represent (and indicate why)?

B. Identify each of the factors in the experimental design, giving each factor a name reflecting the relevant dimension being manipulated. Then, identify each of the levels for each factor. Be sure to identify which factor includes repeated measures.

C. How should the 40 subjects (24 male, 16 female) be assigned to each treatment group? Be specific about how group equivalence will be assured for all key subject variables. In particular, how would you control for potential confounding variables that may result from the different prior experience with alcohol of subjects in different treatment groups?

D. Since the design includes a within-subjects factor, describe how would order and carry-over effects be controlled for that factor. BE SPECIFIC

E. Using the names of factors chosen in “B” above, write out all of the terms that will be included in the mixed-model ANOVA, including all main effects and interactions. How many degrees of freedom are there for each main effect and interaction?

What single term in the above ANOVA must reach statistical significance for the study to support hypothesis (1) above?

Assume that an ANOVA was performed on just the data from the difficult driving conditions, to test hypothesis (2) of the study. What kind of ANOVA would that be? How many factors? How many levels?

Draw a graph of the outcome of the predicted effects of hypothesis (1). Be sure to label clearly the x-axis (name of variable and levels) and the y-axis. Be sure to include (and provide a legend for) the alcohol treatment condition in your plot. Assume that in addition to the predicted outcome, there is a main effect of driving condition, too.

2. Internal validity is the cornerstone to good experimental science, but the study of subject variables by definition must use quasi-experimental approaches. Why? Compare "true" experiments with quasi-experiments in terms of the potential threats to internal validity, including a description of the means available in true experiments to increase internal validity (reduce confounding variables).