University of Isfahan Mathematics and Society 2345-6493 11 1 2025 06 29 Evaluation of the measurement error of Iranian labor force indicators due to changing data gathering mode based on the Markov latent class model Evaluation of the measurement error of Iranian labor force indicators due to changing data gathering mode based on the Markov latent class model 109 131 29531 10.22108/msci.2025.142988.1700 FA Lida Kalhori Nadrabadi Technical Designs and Statistical Methods Department, Statistical Research and Training Center, Tehran, Iran Paria Torabi Kahlan Technical Designs and Statistical Methods Department. Statistical Research and Training Center. Tehran,Iran. Journal Article 2024 10 09 One of the most important non-sampling errors is measurement error, which causes the measured value of the target variable to be different from its true value. The markov latent class model allows for the estimation of measurement error for categorical variables in panel surveys that have at least three repetition periods, with the advantage that it does not require re-interviews. In the Iranian labor force survey, which is conducted with rotational sampling panel with a 2-2-2 pattern, this model can be applied. In this article, the challenges of changing the survey mode in times of crisis and the advantages of telephone interviews compared to face-to-face interviews are reviewed. Also, in order to evaluate the results of the labor force survey in the context of the COVID-19 pandemic, when the survey mode was suddenly changed from face-to-face to telephone interviews, the markov latent class model was used to evaluate the measurement error using data from four survey periods from spring 2018 to summer 2021. The results indicate that the measurement error increased in the first period of changing the interview mode, but decreased in subsequent periods. Overall, it can be stated that telephone interviews can be a suitable method for collecting labor force survey data. One of the most important non-sampling errors is measurement error, which causes the measured value of the target variable to be different from its true value. The markov latent class model allows for the estimation of measurement error for categorical variables in panel surveys that have at least three repetition periods, with the advantage that it does not require re-interviews. In the Iranian labor force survey, which is conducted with rotational sampling panel with a 2-2-2 pattern, this model can be applied. In this article, the challenges of changing the survey mode in times of crisis and the advantages of telephone interviews compared to face-to-face interviews are reviewed. Also, in order to evaluate the results of the labor force survey in the context of the COVID-19 pandemic, when the survey mode was suddenly changed from face-to-face to telephone interviews, the markov latent class model was used to evaluate the measurement error using data from four survey periods from spring 2018 to summer 2021. The results indicate that the measurement error increased in the first period of changing the interview mode, but decreased in subsequent periods. Overall, it can be stated that telephone interviews can be a suitable method for collecting labor force survey data. https://math-sci.ui.ac.ir/article_29531_a735c1c177da64a577e03f260bf80cd6.pdf