Dichotomous Questions – Market Survey

The dichotomous, or two way questions, is an extreme of the multiple choice question. The idea is to offer only two choices — yes or no, did or did not, cash or credit, railroad or airline, and so on. Such questions are the most widely used of the three basic types. The following are examples of dichotomous questions:

Would the service proposed by ‘X’ Lines make motor freight service more useful to you? Is any of the discount normally passed on to others? Did you buy it or was it a gift? Was it new or used when you got it?

In the last two questions the two alternatives are both stated; in the first two, one alternative is stated, while the other is implied. In the second question it seems obvious that the other alternative is that none of the discount was passed on or putting it another way that all of the discount was kept by the firms reporting. Probably there would be no confusion as to what the two alternatives were in this case. It is also likely, however that different results would have been obtained had both alternatives been stated explicitly. For example: Is any of this discount normally passed on to others or is all of it kept by your firm?

An experiment to test the effect of stating only one alternative used the following two questions. Do you think the United States should allow public speeches against democracy? Do you think the United States should forbid public speeches against democracy?

The following results were obtained:

First question Second question

Should allow 21% Should not forbid 39%
Should not allow 62 Should forbid 46
No opinion 17 no opinion 15

As the two questions ask exactly the same things one in a positive way and one in a negative way – the answers should be directly comparable. Those who say the United States should allow such speeches should also report that these speeches should not be forbidden. The results show that this did not occur. Only 21 percent wanted to allow speeches against democracy but 39 percent were against forbidding them. A close estimate of the true feelings on the subject would probably have been obtained if the question had stated both alternatives: Do you think the United States show allow or forbid public speeches against democracy?

Dichotomous questions have about the same advantages as multiple choice questions. They are quick and easy for an interviewer to handle. Editing and tabulation are relatively simple. They offer less opportunity for interviewer bias to creep into the results. The straight ‘yes’ ‘no’ type of answer makes it easy for the respondent to reply. But dichotomous questions may be deceptive in their seeming simplicity. Few dichotomous questions, for example are actually only two way. Take the following question: Do you expect to buy another diesel automobile some day? Undoubtedly some people definitely plan to buy another diesel car and others definitely plan not to but a large middle group may have no definite plans either way. Some of these might properly report, ‘don’t know’. Others might be in a ‘may be’ class. Even the ‘may be’s might fall into distinct groups those who probably would but were not sure and those who probably would not but were not sure. This would mean that instead of two possible answers there would be five: yes, no, probably, probably not, and don’t know. If the diesel question were reworded to include explicit statement of both of the original alternatives and to take into account the five alternatives actually existing, it might appear as follows:

Do you expect to buy another diesel automobile some day, or not?

Yes ____ No____

Probably ___ Don’t Know__

Probably not ____

The five alternatives would not be suggested to the respondents, but if one of them qualified an answer, the interviewer could then check the appropriate space. At the very least the ‘don’t know’ category should be provided. Then the ‘probablys’ would be classified in the ‘yes’ category and the ‘probably not’ in the ‘no’ category. As was pointed out in the discussion of multiple choice questions, however, if the alternatives are to actually stated to the respondent fewer persons will report than would otherwise therefore, if the ‘probably’ answers are not actually indicated to the respondent the number reporting them will have a downward bias and the number saying yes and no will have an upward bias.