Are IIM interviews fake

The influence of fake interviews on survey results

The effect of interviewer cheating on survey results
[Journal article]

Abstract

"One of the many possible criticisms of survey results is distortion caused by fake interviews. These concerns are examined analytically and empirically. Interview falsification is a special case of 'missing data problems' and can therefore be estimated using the same formulas. ... more

"One of the many possible criticisms of survey results are biases caused by fake interviews. These concerns are examined analytically and empirically. Interview falsification is a special case of 'missing data problems' and can therefore be estimated using the same formulas. The corresponding analytical results are simple Statistics only close to small biases. Since such analytical estimates are hardly possible for multivariate statistics, possible bias effects are empirically investigated using various methods. The results of a study on the 'quality' of fake interviews with 22 'interviewers' who each conducted 10' interviews shows, among other things, a greater consistency of fake interviews compared to real interviews, but no major differences to the real data.These falsifications would have had no effect on the calculation of univariate statistics or on multivariate analyzes if they had been 5 percent Bes would have been part of a data set. Finally, some simulations demonstrate the robustness of a regression model even with regard to higher proportions of interview falsification. If it appears necessary to estimate the possible effects of falsified interviews, similar simulations must be carried out in each individual case. "(Author's presentation) ... less


"One of the many possible criticisms of survey results is bias due to interviewer cheating. This objection is tested by analytical and empirical means. The effect of interviewer cheating is a special case of missing-data problems in general. Therefore, the effects can be estimated with similar formu ... more

"One of the many possible criticisms of survey results is bias due to interviewer cheating. This objection is tested by analytical and empirical means. The effect of interviewer cheating is a special case of missing-data problems in general. Therefore, the effects can be estimated with similar formulas. For simple statistics the results indicate only small biasing effects. For multivariate statistics such analytical estimates are impossible; therefore these are empirically tested by different means. The results of a study of the quality of 220 faked interviews show a greater consistency in the faked interviews, but only minor differences in relation to real interviews. If these faked interviews had been a small part (5 percent) of a dataset, the effect of interviewer cheating would have been neglectable. Finally, a few simulations show the robustness of a regression model in spite of a large amount of interviewer cheating. If an estimation of the possible effects of interviewer cheatin g seems to be neccessary, similar simulations will have to be done in each specific case. " (author's abstract) ... less

Thesaurus keywords
Measuring instrument; Estimate; Survey; Interview; Problem; Statistics; Scam; Survey research

classification
Survey techniques and analysis techniques in the social sciences

method
empirically; Method development; Basic research

Language document
German

Year of publication
1991

Page reference
Pp. 25-35

Magazine title
Journal of Sociology, 20 (1991) 1

License
Deposit License - No Redistribution, No Processing

This metadata set was created by the special collection area social sciences (USB Cologne).