Unmatched count technique

The unmatched count technique (sometimes also item count technique ) is a method used in psychology and empirical social research to reduce certain falsifications of interview responses.

If the scientific question is only interested in the number of people who have a certain characteristic, opinion or attitude, the unmatched count technique offers an opportunity to estimate the actual number of bearers of the critical characteristic through anonymization . Embarrassing or delicate questions, but also those (e.g. Tom Bradley's loss of election ), where unexpected changes of opinion are empirically known with sufficient anonymity , are considered special areas of application .

Procedure

The respondents are randomly divided into two groups. In one group ( control group ) only a few harmless questions are asked, in the other group the " sensitive question ", that is one more question (hence the name "Unmatched Count"). The respondent should only state how many "yes" answers they would have given. Since the interviewer does not know how the sum mentioned came about, the critical "yes" answer, the admission of the embarrassing quality, is protected. Because of the unequal number of questions, the proportion of "yes" answers to the question of interest can then be deducted.

example

The control group should indicate how many of the following statements are true:

I've moved before.
I have a pet.
I like to go to the cinema.
I like to watch sports on TV.

The number of "yes" answers in this group is 31.

The other group is also asked about the characteristic of interest:

I've cheated on an exam before.

The number of "yes" answers in this group is 34.

evaluation

The number of "yes" answers in the control group forms the so-called baseline . The assumption is that the second group would have had the same number without the critical question, that is, that the difference from 3 in the example is based on the number of "yes" answers to the critical question. The percentage of these "yes" -sayers (for example, with 100 respondents per group it would be 3%) is used as an estimate for the percentage of cheaters in the population .

However, one must note that the answer to the non-critical questions without the critical question already has considerable fluctuations due to stochastic processes. If the rate of answering these uncritical questions with "yes" is around 50%, their value fluctuates in this example by 5% ( ), i.e. H. a rate of 3% is not detectable because it is within the margin of error (i.e., less than the standard error). ${\ displaystyle {\ sqrt {0.5 * 0.5}} / {\ sqrt {100}}}$

criticism

The main point of criticism of the method is that the desired information can only be identified by part of the sample: In the example, the 'control group' does not receive the sensitive question at all. One possibility to remedy this weak point is to ask the control group with new harmless questions plus the tricky question, while the previously 'experimental group' becomes the control group in this case. While the averaging of the two resulting values is not a problem, this improvement (which also means a longer questionnaire) makes statements about the quality of the statistical estimator (i.e. the percentage value) and the calculation of the scatter significantly more complicated (Ben Jann 2013) .

literature

Elisabeth Coutts, Ben Jann: Sensitive Questions in Online Surveys: Experimental Results for the Randomized Response Technique (RRT) and the Unmatched Count Technique (UCT) , General Online Research 2009 in Vienna
Ben Jann: Asking Sensitive Questions in Online Surveys: An Experimental Comparison of the Randomized Response Technique and the Crosswise Model , Conference of the Department of Methods in the Social Sciences, 2013 in Vienna, p. 33
Allison M. Ahart, Paul R. Sackett (2004). A New Method of Examining Relationships between Individual Difference Measures and Sensitive Behavior Criteria: Evaluating the Unmatched Count Technique . Organizational Research Methods, Vol. 7, No. 1, pp. 101-114
Dan R. Dalton, James C. Wimbush, Catherine M. Daily (1994). Using the Unmatched Count Technique (UCT) to estimate base rates for sensitive behavior . Personnel Psychology 47, pp. 817-829

Web links