Questions - Heart of All Polls - continued
The range of responses offered to respondents can also distort responses. Respondents may not like any of the options offered. One well known example is, "Answer yes or no. Have you stopped beating your wife?" With this kind of closed-end or forced-choice question the respondent is put in the difficult position of having to choose one response from two unacceptable alternatives.
From the standpoint of the polling organization the advantage of a forced-choice question is clear: the public's views are neatly divided into discrete alternative opinions. Recording responses is simplified because there are no surprises. Respondents approve or disapprove of the president's proposal. Period.
Closed-ended vs. Open-ended Questions
The disadvantage of forced-choice questions is that respondents have no good way to indicate that they do not like the options they are given. The respondent who wants to say he approves of the president's proposal, except for certain provisions, is forced to choose between approve, disapprove, and no opinion, even though none of these options accurately reflects his view. An open-end question allows the respondent to say exactly what is on his or her mind. There are no limits to what respondents can say when they are asked, "What is the most important facing the country today?" The difficulty with these questions comes in trying to cope with the variety in the responses. There could be as many different problems mentioned as there are respondents in the survey. To make sense of the 1,500 different answers, the polling organization must identify common threads running through the replies and group them accordingly, assigning them to defense, inflation, environment, crime, and others.
Closed-end questions generally produce higher percentages for any given response than open-end questions. Asked an open-end question about the causes of crime, some of the respondents might mention courts being too soft on criminals. In contrast, the percentage blaming courts would be considerably higher if that is one of only three or four options in a forced choice question.
Most people simply respond to the question put to them. If respondents are asked, "Which party do you think would do a better job of protecting Social Security, the Democrats or Republicans?" most will select one of the two parties. A few will volunteer the response, neither, or both equally. However, if the question offers the options of, "Democrats, Republicans, neither, both equally, or haven't you thought enough about that to say?" then the percentage selecting Democrats or Republicans will drop. Limiting the range of acceptable answers can bias the responses.
This can be true even when the alternative seems to be implied. For example, respondents must receive as much encouragement for a negative response as an affirmative one. Tests have found that, "Do you think we should spend more on education or not?" will yield a higher negative response than, "Do you think we should spend more on education?" Alternatives must be explicit, not implicit, if the question is to avoid biasing the responses.