Refer to Exercise 8.2.25 on accuracy in surveys. Another way to measure accuracy of responses is whether people round when giving numerical answers. The researchers considered a number to be rounded if it ended in a 0 or a 5. One question in which the subjects had to give a numeric answer was to how many movies they had seen in the past 12 months. They found that 17.1% of those that got the survey via a human text rounded their answers, 12.1% of those that got the survey via an automated text rounded their answers, 24.4% of those that got the survey via a human voice rounded their answers, and 18.2% of those that got the survey via an automated voice rounded their answers. 

a. Produce a table of observed counts comparing whether they rounded or not across the modes of survey. The totals for each interview mode are the same as in the Exercise 8.2.25 table. 

b. Are the validity conditions met for a chi-square test? Explain.

c. State the hypotheses. 

d. Using the appropriate applet, determine the theory-based p-value. 

e. State your conclusion in the context of the research question. If you get significant results, determine which interview modes are significantly different than which other interview modes and give the direction of those differences.

Data from Exercises 8.2.25

When surveys are administered, it is hoped that the respondents give accurate answers. Does the mode of survey delivery affect this? American researchers investigated this question (Schober et al., 2015). They had 634 people agree to be interviewed on an iPhone, and they were randomly assigned one of the four interview modes (human voice, human text, automated voice, automated text). One question that was asked was whether they exercise less than once per week on a typical week (an example of a question in which a yes would be considered socially undesirable). The summarized data on the answers to this question are shown in the Exercise 8.2.25 table. 

Transcribed Image Text:

Exercise? Yes No Total Human text 34 124 158 Automated text 46 111 157 Human voice Automated voice 21 139 160 20 139 159 Total 121 513 634