Longitudinal studies provide valuable insights into changes and factors influencing responses over time, but inappropriate methods can lead to erroneous results. This study evaluates longitudinal data analysis methods for estimating antibody titres, focusing on correcting inappropriate commonly used methods and providing recommendations for optimal statistical inference. The study contributes to the knowledge base in Malawi and addresses the gap in appropriate longitudinal modelling techniques. Using inappropriate statistical methods in longitudinal data analysis can yield misleading results, affecting the validity and reliability of research findings. Addressing this issue is crucial for ensuring accurate estimation of antibody titres. In this, study a comparative approach was employed, analyzing both real-world and simulated data to assess the performance of different modelling techniques. A longitudinal censored mixed model used in the simulated data to account for lower limits of detection and contrasted this to imputations of censored values to 0, DL/2, DL, and complete case analysis. Censored regression models and imputations were used for non-linear, non-longitudinal PCVPA data. Raw data used arithmetic means, while log-transformed data used geometric means. The longitudinal aspect of the data is accounted for through random effects. By simulating ELISA data with known vaccination and age effects evaluated the effectiveness of statistical models in estimating antibody concentrations. The analysis of both real-world and simulated data reveals significant insights into the performance of different statistical methods. Findings indicate that certain models perform poorly in capturing the effects of age, exposure, and gender. However, the censored model stands out by providing estimates closer to the true values and narrower confidence intervals, particularly in intercept estimation. The comparison between real-world and simulated data underscores the importance of selecting appropriate statistical methods for longitudinal data analysis. The study's results emphasize the significance of the censored model in improving estimation accuracy and reducing bias. Thereby, enhancing understanding of longitudinal data analysis for antibody titres, contributing to advancing statistical inference in research.