Inferential Statistics and the Pitfalls of Nonrandomized Sampling in Nursing Research

Inferential statistics are foundational tools in health and nursing research. However, their misuse-particularly when applied to nonrandomized samples-is widespread and has serious implications for the integrity of science and evidence-based nursing practice.

The aims of this study were to examine the consequences of performing inferential statistical analysis on nonrandomized samples and provide guidance on alternative approaches when random sampling is not feasible.

This paper synthesizes evidence from statistical theory, research methodology, and nursing literature to describe the assumptions of inferential statistics and the biases introduced by nonrandomized sampling. Alternatives such as nonparametric tests, bootstrapping, and descriptive statistics are also described.

Violating statistical test assumptions, such as random sampling and independence, can lead to misleading p -values, invalid confidence intervals, and incorrect generalizations. Systemic factors contributing to misuse include institutional pressures, growing publication options, and insufficient statistical training.

Inferential statistics must be grounded in proper sampling methods. Researchers should avoid overgeneralization from biased samples, use alternative analytical approaches where appropriate, and clearly disclose methodological limitations. Reform in nursing education and publication standards is critical to maintaining the validity and trustworthiness of nursing science.