Principal Components/Factor Analysis

Main aspects

  • Reduces a number of questions down to a few key factors
  • Works by grouping together questions that are correlated
  • Most commonly used on attitudinal data e.g. might reduce two dozen statements concerning smoking (agree strongly ... disagree strongly) down to half a dozen factors
  • Resultant factors are usually independent of each other, measuring quite separate 'dimensions'
  • Number of factors is determined by how much variation they explain and their interpretability
  • Each respondent can be scored on each factor which can be shown as means in crosstabs and are sometimes used as input to a subsequent cluster analysis