UGC NET Questions (Paper – 1)

P-hacking refers to manipulating data analysis or selectively reporting results to obtain statistically significant p-values, often without a strong theoretical rationale. Researchers may try multiple models, subsets or transformations and only present analyses that produce significant findings. This practice inflates Type I error rates and undermines the credibility of research. Because the stem describes repeatedly analysing the same data until significance is found, p-hacking is the correct term.

Option A:
Stratification is a sampling technique that divides a population into homogeneous subgroups for more precise representation and is unrelated to manipulating analyses for significance. It is a legitimate methodological procedure rather than a questionable practice. Hence, stratification does not fit the stem.

Option B:
Triangulation involves using multiple methods, data sources or investigators to improve the credibility of findings and is generally considered a strength, not an unethical practice. It seeks converging evidence rather than repeated fishing for significance. Therefore, triangulation is not appropriate here.

Option C:
Pruning may refer to trimming decision trees or simplifying models in statistical or machine learning contexts, and is not commonly used to describe unethical searching for significant results. It also does not capture the selective reporting dimension implied in the stem. Thus, pruning is not the correct completion.

Option D:
P-hacking undermines the integrity of scientific literature because published significant findings may simply reflect exploratory searching rather than real effects. It highlights the need for transparent preregistration and replication. These concerns align with the scenario described in the question, confirming p-hacking as the accurate term.