Trial Heat Question

Wording of Trial Heat Questions

Among the choices to be made are (a) whether to include the names of all candidates who are on the ballot or just the major candidates, (b) whether to mention the party affiliation of each candidate named, (c) whether to probe respondents for a choice if they say they are undecided, and (d) whether to obtain some measure of strength of support or certainty of the choice.

Nearly all trial heat questions ask respondents to react as if “the election were held today.” Most do mention the party affiliation of candidates, though party labels are not on the ballot in all states. There are many ways to probe certainty of choice or strength of support. Most pose a binary choice: strong supporter or not strong, certain or not. One question used by some organizations asks about the candidate not chosen: Do you think there is a chance that you might vote for_____in November, or have you definitely decided not to vote for (her)(him)?

The decision regarding whether or not to include third-party and minor-party candidates is often not a straightforward one. In 2000 and in 2004, Ralph Nader had no realistic chance of winning the presidency, but his candidacy did attract enough support to have arguably affected the outcome of the race in one or more states. All major polling organizations in both years included Nader and his running mate in poll questions conducted in states where Nader was on the ballot. Other third-party candidates were qualified for the ballots in some states, but most polls did not include their names in the question wording asked of respondents, although respondents could volunteer their names.

One practice common to nearly all election pollsters is the rotation of the order of presentation of candidates in the question. Experiments have shown that the order can create small biases in response, and rotation is designed to prevent this. An alternative view is that candidates should be presented in the order they will appear on the ballot (which is itself often determined by chance but then fixed for all ballots in a given jurisdiction). It is not clear whether preserving the ballot order in a telephone survey, for example, re-creates the same cognitive experience a voter will have when encountering the printed ballot or a voting machine on Election Day. In any event, it is almost never feasible for national surveys to implement a state-by-state ballot order.

...