Evaluating Normality in Continuous Data: Evidence from LASI Wave-1

Abstract

Abstract

Background

Descriptive statistics are indispensable in empirical research as they furnish concise summaries that clarify the fundamental attributes of study variables. These summaries generally include measures of central tendency and variability, which are essential for precise data interpretation. Nonetheless, whether to use parametric or non-parametric statistical tests depends mostly on how well the dataset satisfies the normality assumption. Consequently, the evaluation of the normality of continuous variables emerges as a pivotal preliminary phase in statistical analysis.

Method

The present study examines the assessment of normality utilizing both quantitative and visual methodologies on continuous variables sourced from the Longitudinal Aging Study in India (LASI) Wave-1 dataset. Quantitative approaches incorporate the Anderson-Darling test, while visual evaluations are performed through histograms, Q-Q plots, and P-P plots. The data were subjected to analysis utilizing SPSS and Minitab software to compare the efficacy and interpretability of each technique.

Results

The findings of the study suggest that visual methods offer intuitive understanding of data distribution, especially when used alongside statistical tests. The Anderson-Darling test is particularly effective for medium to large sample sizes, providing a reliable assessment of deviations from normality. Descriptive and visual analyses indicate that the data largely follow a normal distribution. However, no single method is universally best, as the choice of method depends on factors such as sample size, data characteristics, and the research context.

Conclusion

This study underscores the necessity of performing normality assessments prior to hypothesis testing to ensure the validity and reliability of research findings. The utilization of an integrated approach combining both quantitative and visual methods facilitates a more comprehensive understanding of data distribution, especially when engaging with extensive secondary datasets such as LASI.

Keywords

LASI, Normality testing, Descriptive statistics, Visual and Numerical methods, Statistical tests