A New Method for Simultaneous Estimation of Means of Two Sensitive Variables Using Correlated Scrambling Variables

Abstract

Recently (Ahmed et al. in Stat Theory Methods 47:324–343, 2018) have discussed the problem of estimating the means of two sensitive variables simultaneously by using two uncorrelated scrambling variables. In this study, we introduce a new idea that use of correlated scrambling variables in randomized response techniques play an important role to increase the efficiency of the estimators of population means of sensitive variables. We found that the simultaneous estimation of the means of the two sensitive variables by using two correlated scrambling variables results in gain in the efficiency of the estimators. This idea is appealing because it costs nothing to the investigator and does not effect the privacy of a respondent as well. Empirical evidences are given based on application to a real secondary data set.

Keywords

Simultaneous estimation of means / Two sensitive variables / Randomized response techniques / Relative efficiency / Correlated scrambling variables / 62D05

References

[1] Ahmed S, Sedory SA, Singh SSimultaneous estimation of means of two sensitive variables. Commun. Stat. Theory Methods, 2018, 47(2): 324-343.

[2] Ahmed S, Sedory SA, Singh SForcibly re-scrambled randomized response model for simultaneous estimation of means of two sensitive variables. Commun. Math. Stat., 2018

[3] Bar-Lev SK, Bobovitch E, Boukai BA note on randomized response models for quantitative data. Metrika, 2004, 60(3): 255-260.

[4] Chaudhuri A, Stenger HTheory and methods of survey sampling, 1992, NY: Marcel Dekker. Inc

[5] Eichhorn BH, Hayre LSScrambled randomized response methods for obtaining sensitive quantitative data. J. Stat. Plan. Inference, 1983, 7(4): 307-316.

[6] Greenberg BG, Kuebler RRJrAbernathy JR, Horvitz DGApplication of the randomized response technique in obtaining quantitative data. J. Am. Stat. Assoc., 1971, 66(334): 243-250.

[7] Huang KCUnbiased estimators of mean, variance and sensitivity level for quantitative characteristics in finite population sampling. Metrika, 2010, 71(3): 341-352.

[8] Himmelfarb S, Edgell SEAdditive constants model A randomized response technique for eliminating evasiveness to quantitative response questions. Psychol. Bull., 1980, 87(3): 525.

[9] Horvitz, D.G., Simmons, W.R., Shah, B.V.: Unrelated question randomized response model. Journal of the American Statistical Association (Vol. 63, No. 322, pp. 754-754). 732 N Washington ST, Alexandria, VA 22314-1943 USA: Amer Statistical Assoc (1968)

[10] Rosner BFundamentals of biostatistics, 2006, Belmont, CA: Thomson-Brooks/Cole

[11] Singh, S.: On the estimation of correlation coefficient using scrambled responses. In: Handbook of Statistics (Vol. 34, pp. 43-90). Elsevier (2016)

[12] Singh S, Horn SAn alternative estimator in multi-character surveys. Metrika, 1998, 4899-107.

[13] Warner SLRandomized response: a survey technique for eliminating evasive answer bias. J. Am. Stat. Assoc., 1965, 60(309): 63-69.

Rights & permissions

School of Mathematical Sciences, University of Science and Technology of China and Springer-Verlag GmbH Germany, part of Springer Nature