How to detect data collection fraud using system properties approach"?
Answers
Answer:
Data collection through survey comes with its own limitations. Among these limitations, two important ones are supervision of survey work done by enumerators and quality of data collected through survey work. Quality of data erodes when enumerators take a casual approach to survey work and exploits the loopholes in supervision of survey work. To fill these loopholes this study has attempted to devise procedure. This approach exploits the unique property of ranked responses that population of ranked responses for a set of objects ranked by respondent is fully known. This helps in devising a weighted average which serves as test statistic. Kolmogorov Smirnov (KS) test is deployed to test the significance of difference enumerator generated series of test statistic and randomly generated series from population of same size. Through simulation study it was found that KS test has type one error probability of nine percent. Thus we determine the probability of KS test can go wrong for an enumerator generated series against very large number of samples of same size drawn from population. If the probability of type one error exceeds 9 percent then we have reason to question the enumerator. The limitation of this approach is that population from which enumerator is doing sampling must be very large for ranked responses to fully reflect the population of true responses.
Explanation: