Exploratory Factor Analysis (EFA) is a statistical technique used to uncover unobserved (latent) structures that explain patterns of correlations among observed variables. When combined with Likert scales, it becomes a powerful tool in social science research. Given the ordinal nature of data derived from such scales, evaluating the performance of EFA under various experimental conditions is a topic of significant interest. This thesis investigates the performance of EFA with ordinal data across three key aspects. The first aspect examines the impact of symmetry conditions on EFA results. The second focuses on the differences between Pearson correlation matrix and the polychoric correlation matrix. The third explores the effect of using unmatched factor numbers in the EFA process. To address these aspects, simulation methods were used to generate normally distributed data, which were subsequently transformed into ordinal data to mimic real-world conditions. The EFA was then applied to the simulated datasets, and the root mean square error (RMSE) was employed as a metric to evaluate the accuracy of the factor loadings. Additionally, analysis of variance (ANOVA) was implemented to provide statistical insights into the effects of the experimental conditions. Overall, the results indicate that using the polychoric correlation matrix in combination with symmetric conditions yields the most reliable EFA performance for ordinal data. However, under other conditions, the performance of EFA was found to be less stable.