Letter to The Editor: Sigsearch: A New Term for Post Hoc Unplanned Search for Statistically Significant Relationships with The Intent to Create Publishable Findings

Abstract

Post-hoc secondary data analysis with no prespecified hypotheses has been discouraged by textbook authors and journal editors alike. Unfortunately no single term describes this phenomenon succinctly. I would like to coin the term “sigsearch” to define this practice and bring it within the teaching lexicon of statistics courses. Sigsearch would include any unplanned, post-hoc search for statistical significance using multiple comparisons of subgroups. It would also include data analysis with outcomes other than the prespecified primary outcome measure of a study as well as secondary data analyses of earlier research.

Keywords:

• sigsearch
• statistics
• subgroup analysis