Does providing additional information affect responses to a survey question? Two statistics students decided to investigate this issue by asking different versions of a question about texting and driving. Fifty mall shoppers were divided into two groups of 25 at random. The first group was asked version A and the other half were asked version B. Here are the actual questions:

Version A: A lot of people text and drive. Are you one of them?

Version B: About 6000 deaths occur per year due to texting and driving. Knowing the potential consequences, do you text and drive?

Of the 25 shoppers assigned to version A, 16 admitted to texting and driving. Of the 25 shoppers assigned to version B, only 12 admitted to texting and driving.

a. State appropriate hypotheses for performing a significance test. Be sure to define the parameters of interest.

b. Explain why you should not use the methods of this section to calculate the P-value.

c. We performed 100 trials of a simulation to see what differences (Version A−Version B) in proportions would occur due only to chance variation in the random assignment, assuming that the question asked doesn’t matter. A dotplot of the results is shown here. What is the estimated P-value?

d. What conclusion would you draw?

Transcribed Image Text:

우 -0.4 -0.3 -0.2 -0.1 0.1 0.1 0.2 0.3 0.4 Simulated difference (Version A - Version B) in sample proportion of yes answers