An Empirical Test of the Utility of the Observations-To-Variables Ratio in Factor and Components Analysis

TLDR

Many researchers propose a minimum observation‑to‑variable ratio or an absolute minimum number of observations to achieve stable factor configurations, yet few empirical studies have tested these claims. The authors systematically examined two large phobic samples (N = 1104 and N = 960) using the Fear Survey Schedule‑III and Fear Questionnaire, randomly splitting the data into subsamples with ratios from 1.3:1 to 19.8:1. Neither the observation‑to‑variable ratio nor an absolute minimum number of observations influenced factor stability.

Abstract

Many researchers have proposed a minimum ratio of observations to variables or an absolute minimum of observations in order to obtain stable factor config urations. However, hardly any empirical studies em ploying real data are available that attest to the tena bility of these proposals. A systematic analysis of the problem was undertaken, using self-report data from two large phobic samples on the Fear Survey Sched ule-III (N = 1104) and the Fear Questionnaire ( N = 960). The data sets were randomly split into subsam ples with ratios varying from 1.3:1 to 19.8:1. Neither the observations to variables ratio nor an absolute minimum of observations had any influence on factor stability.