The choice of a data-collection tool depends on who the respondents are, how many there are, how widely they are spread and how sensitive the subject is. Each of those points here favours the same instrument.

A questionnaire is the preferred tool. The study is in an urban area, where literacy is high enough for respondents to read and complete a written form without assistance. Assessing a party's prospects requires a large sample if the findings are to be representative, and a questionnaire can be distributed widely at low cost and answered at the respondent's convenience. Most importantly, political preference is a sensitive matter, and the anonymity of a self-administered form encourages honest answers where a face-to-face encounter might not.

An interview would give richer detail but is costly and slow for a large sample, and the presence of an interviewer can bias replies on a politically sensitive question. A schedule is a set of questions filled in by an enumerator who visits the respondent, which is the appropriate choice where respondents cannot read - typically in rural or low-literacy settings - and it carries the same interviewer-effect problem.

A rating scale is not a data-collection tool in this sense at all; it is a device for recording the degree of an attitude, and it would be used as a component within a questionnaire rather than in place of one.

Hence the researcher should prefer a questionnaire.