Estimation of measurement error in binary data in surveys | ||
| Journal of Statistical Modelling: Theory and Applications | ||
| دوره 6، شماره 1، فروردین 2025، صفحه 185-192 اصل مقاله (116.72 K) | ||
| نوع مقاله: Original Scientific Paper | ||
| شناسه دیجیتال (DOI): 10.22034/jsmta.2026.23540.1189 | ||
| نویسنده | ||
| Roshanak alimohammadi* | ||
| Department of Statistics, Faculty of Mathematical Sciences, Alzahra University, Tehran, Iran | ||
| چکیده | ||
| Measurement error is an inherent and unavoidable component of nonsampling error in surveys, and its estimation is essential for assessing the quality of survey results. This paper investigates two agreement criteria, Cohen’s Kappa and Gwet’s AC1, for quantifying measurement error in binary survey data. A theoretical comparison is first presented to highlight the conceptual differences between the two coefficients. This is followed by a simulation study in which true binary values are generated under varying prevalence levels, and observed responses are obtained through controlled misclassification mechanisms characterized by specified levels of observed agreement. The performance and empirical variances of both agreement measures are evaluated and the results demonstrate that Gwet’s AC1 provides more stable agreement estimates and variance behavior than Cohen’s Kappa, particularly under conditions of extreme prevalence. | ||
| کلیدواژهها | ||
| Agreement؛ Cohen's Kappa؛ Gwet's AC1؛ Measurement error؛ Prevalence | ||
|
آمار تعداد مشاهده مقاله: 66 تعداد دریافت فایل اصل مقاله: 32 |
||