# Study on Fixed Effect versus Random Effects Modeling in a Panel Data Analysis; a Consideration of Economic and Political Indicators in Six African Countries

Individual heterogeneity in panel data analysis can be addressed to avoid bias in the final estimations. The data in a panel might be balanced or imbalanced, and the panel can be short or long. The fixed effects and random effects modelling methodologies were applied to an economic data set, “Africa,” in the R Amelia package, to determine the right model. Both models were fitted to the data, with the assumptions of both models taken into account. The random model estimates passed the Breusch-Pagan Lagrange Multiplier test, however it had a weak coefficient of determination (R2 of 0.48697). The fixed effect was then estimated using four alternative methods (Pooled, LSDV, Within-Group, and First Differentiation) and compared to the random effect model using the Hausman test. The random effect was shown to be inconsistent in all tests, indicating that the fixed effect was more appropriate for the data. The LSDV was determined to have the best fit among the fixed effects models, with an R2 of 0.8851.

**Author(S) Details**

**M. T. Nwakuya
**University of Port Harcourt, Rivers State, Nigeria.

**M. A. Ijomah
**University of Port Harcourt, Rivers State, Nigeria.

View Book:- https://stm.bookpi.org/RAMRCS-V4/article/view/4871