Socio-Economic Status of Individuals in Tehran University of Medical Sciences Employees` Cohort Study Using PCA, MCA and FAMD Methods

Abstract

Introduction: Determining socio-economic status (SES) can greatly help decision makers in the field of social health. Because SES can play an important role in accessing medical services or welfare amenities. We aimed to determine the SES using principal component analysis (PCA), multiple correspondence analysis (MCA), and factor analysis of mixed data (FAMD) methods.

Methods: In this cross-sectional study (2023), 4448 employees aged 19 to 75 years were included to the study from Tehran University of Medical Sciences employees` cohort (TEC). Demographic variables and socioeconomic factors were considered. Considering the weaknesses of PCA and MCA methods, we calculated the SES score using PCA, MCA and FAMD methods, and the percentile of people was determined. These weaknesses include normality assumption and considering only linear relationship for PCA, inability to interpret the relationships between variables and considering each level of classification variables as a new variable for MCA.

Results: We studied 4448 people (39.3% men) with mean age of 42.3 and a standard deviation of 8.7.The correlation between the percentiles obtained through PCA, MCA and FAMD methods was very high, and the highest correlation was related to the percentiles obtained through PCA and FAMD methods with a value of 0.994. The intraclass correlation coefficient value was 0.996. Also, this value was 0.996 and 0.994 in the random samples of 250 and 100 individuals from the original data, respectively.

Conclusion: All of the three methods worked similarly on determining the SES and calculating the percentile of people. PCA and FAMD methods had better agreement than others. Therefore, in studies that have both quantitative and qualitative variables, the choice of analysis method depends on the opinion of the researcher.