7.46 Equivalence of paper and computer-based questionnaires. Computers are commonly being used to complete questionnaires because of the increased efficiency of data collection and reduction in coding errors. Studies, however, have shown that questionnaire format can influence responses, especially for items of a sensitive nature.20 Consider the small study below comparing paper and computer survey formats of a self-report measure of mental health. Each participant completed both forms on adjacent days with the order determined by a flip of a coin.
| Subject | Paper | Computer | Diff | Subject | Paper | Computer | Diff |
|---|---|---|---|---|---|---|---|
| 1 | 5 | 2 | 3 | 11 | 6 | 5 | 1 |
| 2 | 4 | 3 | 1 | 12 | 5 | 5 | 0 |
| 3 | 4 | 4 | 0 | 13 | 3 | 7 | −4 |
| 4 | 7 | 8 | −1 | 14 | 3 | 6 | −3 |
| 5 | 4 | 5 | −1 | 15 | 4 | 4 | 0 |
| 6 | 6 | 7 | −1 | 16 | 2 | 3 | −1 |
| 7 | 4 | 3 | 1 | 17 | 7 | 10 | −3 |
| 8 | 6 | 8 | −2 | 18 | 8 | 7 | 1 |
| 9 | 6 | 5 | 1 | 19 | 4 | 6 | −2 |
| 10 | 2 | 3 | −1 | 20 | 6 | 8 | −2 |
(a) Explain to someone unfamiliar with statistics why this experiment is a matched pairs design.
(b) The measure involves 10 items and produces a whole number score ranging between 0 and 20. Do you think it is appropriate to use the t procedures on the difference in survey scores? Explain your answer.
(c) Perform an equivalency test at the 0.05 level using the limits ±0.5 and state your conclusion.