Research Ethics
Survey research

Survey research is a major component of political science research. There has been an academically-motivated and mainly SSHRC-funded Canadian Election Study in every election but one since 1965. Smaller scale work is commonly conducted in the context of provincial elections. As well, survey research is important in the realm of multicultural citizenship, as, for example, with the Equality, Security, and Community Study, funded under an MCRI. Survey research is increasingly important for policy research, especially for systematic study of discourse and framing.

1. Contact: Almost all survey research since the mid-1980s has been by telephone, although the World Wide Web is gaining ground as a mode for fieldwork. It is essentially inescapable that fieldwork initiated by random digit dialing creates an intrusion on households. Some ethical review boards might dislike this. Sometimes it is suggested that respondents first be sent a warning letter, but this raises the cost of the research and compromises the sample.

2. Consent: The biomedical model tends to require formal, written consent. This is impossible in the telephone survey context. It is reasonable to ask for the script of the “front end” of the interview (the most important element in which is a forthright indication of the identity of the sponsoring institution) and for scripts of replies to reasonable questions, e.g., about who is paying for the research or how long the interview will take. The key in telephone interviewing is that it is very hard to coerce a potential respondent. Such persons find it very easy to say no, as declining response rates attest. The situation is far less coercive than, for example, a psychology department subject pool or an in-home survey interview.

3. Content: For the most part, content issues should be covered by properly constructed consent statements. Respondents should not be blindsided. The substantive interests of the research usually dictate careful introduction of potentially difficult issues. And again, respondents do not find it hard to refuse or, indeed, to terminate the interview. But in general, review boards should not seem themselves as in the censorship business. Consent is their essential subject.

4. Confidentiality: These are potentially serious issues. The survey research firm or institution will have information, including the telephone number and, typically, the respondent’s first name that must be kept in absolute confidence. Care also has to be taken over the full complement of a respondent’s demographic and location indicators to ensure that a highly motivated third party cannot produce a vector of characteristics that enable identification of an individual respondent. It may be necessary for research purposes to have all this information. But it may be no less necessary to mask some of it in a public-use version of the data set and to put the raw individual data behind a ³wall² that a secondary analyst could only pass through by signing a confidentiality agreement.

5. Retention of data: Some review panels are confused on this and ask how the data will be kept from the public and ultimately destroyed. Of course, SSHRC requires precisely the opposite, viz., that data (anonymized as necessary) become property of the research community and do so in a timely way. Increasingly, researchers analyze merged files from multiple sources, as many survey data sets gain value as time passes and repetition becomes possible. The Canadian Election Studies are a perfect case in point.