Consider the following surveys and questionnaire items. Which administration method would you consider more appropriate in terms of (a) reduction of nonsampling error and (b) cost?
Survey objective/questionnaire item Administration method
Binge drinking during weekdays; over last month, how often did you have more than five drinks in one evening?
Lunch habits; where have you had lunch yesterday (canteen, restaurant, etc.)?
Taxes: how much did you earn last year?
Attitudes toward football; describe your feelings when you watch a football match
Customer satisfaction for train users; how satisfactory do you rate your last train trip?
Use of Internet at work; how many times did you check your personal e-mail this morning?
With a budget of $\$ 20,000$, you are asked to run a survey. Consider the following fixed costs, variable costs and response rates:
$$
\begin{array}{lrcc}
\hline \text { Survey type } & \text { Fixed cost } & \text { Cost per questionnaire } & \text { Response rate } \\
\hline \text { E-mail } & 1,000 & 0.50 & 10 \% \\
\text { CATI } & 10,000 & 10.00 & 50 \% \\
\text { Mail } & 100 & 1.00 & 1 \% \\
\text { Mall-intercept } & 2,000 & 15.00 & 40 \%
\end{array}
$$
a. Compute the expected number of valid responses for each of the administration types.
b. Your client pretends that you use mall-intercept on at least part of the sample and that the sample size should not be less than 700 . What is the most effective way to proceed with mixed administration methods?