Planned missingness with multiple imputation: enabling the use of exit polls to reduce measurement error in surveys

Marco Morales sent me this paper of his with Rene Bautista:

Exit polls are seldom used for voting behavior research despite the advantages they have compared to pre and post-election surveys. Exit polls reduce potential biases and measurement errors on reported vote choice and other political attitudes related to the time in which measurements are taken. This is the result of collecting information from actual voters only minutes after the vote has been cast. Among the main reasons why exit polls are not frequently used by scholars it is the time constraints that must be placed on the interviews, i.e., short questionnaires, severely limiting the amount of information obtained from each respondent. This paper advances a combination of an appropriate data collection design along with adequate statistical techniques to allow exit polls to overcome such a restriction without jeopardizing data quality. This mechanism implies the use of Planned Missingness designs and Multiple Imputation techniques. The potential advantages of this design applied to voting behavior research are illustrated empirically with data from the 2006 Mexican election.

This sounds cool. I’d only add that all surveys are “planned missingness.” That’s what makes it a survey rather than a census. Also I want to take a look at their data and see if their results are consistent with what we found in our analysis of the 2006 Mexican presidential election.

(I’d also make some comments on the Tables, but I think you know what I’d say, so I won’t say it, and I’d say that in Figure 3 they should remove the little numbers on the lines and just label the y-axis, and I’d say that in Figure 4 they should remove that second decimal place and compress the scale of the axes and give full names or party labels rather than abbreviations on top of the plots, and on figure 5 I’d recommend just displaying “positive minus negative” rather than separately showing both (and getting rid of those silly vertical lines at the ends of the error bars, and getting rid of the confusing labels at -0.3 and +0.3 that appear in part of the graph but not all of it), but that would be a distraction from the important contributions of the article, so I won’t waste your time on that sort of picky comment, instead simply appreciating the effort that did go into this paper.)

1 thought on “Planned missingness with multiple imputation: enabling the use of exit polls to reduce measurement error in surveys

  1. this just reminds me of a thought i had back when i was taking a course on statistical quality control…

    so in quality control, you tend to get poorer results if you attempt to inspect every item, due to increasing the likelyhood of adding testing errors, particularly through human error (fatigue, laziness, and so on). instead, inspecting a sample of items produces better results.

    by that logic, why not count only a random sample of those ballots which are actually cast in elections? randomly selecting 30% of all cast ballots and using them to construct an estimate of the actual vote totals would probably be just as effective.

    of course, there is a flaw in my logic (and not the "antidemocratic" one): the machines used to count ballots are much more resistant to fatigue and inconsistancy as people inspecting industrial products.

Comments are closed.