Survey research and design in psychology/Tutorials/Psychometrics/Exploratory factor analysis

This tutorial provides an introduction to conducting exploratory factor analyses (EFAs) using SPSS. The SPSS data, syntax and output is available for each analysis, along with screencasts on youtube.

Here are the recommended steps when conducting an exploratory factor analysis using SPSS:
 * 1) Get to know the data:
 * 2) Univariate distributions: Explore the distribution and central tendency for each of the variables statistically (e.g., use descriptive statistics to examine the M, SD, skewness, and kurtosis) and graphically (e.g., histograms)).
 * 3) Bivariate distributions: Examine the bivariate correlations statistically (e.g., correlation matrix) and graphically (e.g., scatterplots)
 * 4) Examine EFA assumptions:
 * 5) Sample size
 * 6) Linearity
 * 7) Homoscedasticity
 * 8) Factorability
 * 9) Conduct EFA (repeat until a good model is identified - may require many different analyses involving different numbers of factors, different types of rotation, and different sets of items):
 * 10) Via Pull-down menus: Analyze - Data Reduction - Factor Analysis - Put target variables into the Variables box (order doesn't matter, but the output will be neater to read if the variables are sequentially ordered from 1 to X)
 * 11) Descriptives - check these options:
 * 12) Initial solution
 * 13) Correlation matrix
 * 14) Coefficients
 * 15) KMO and Bartlett's
 * 16) Anti-image
 * 17) Extraction - check these options:
 * 18) Analyse - Correlation matrix
 * 19) Display - Screen plot
 * 20) Extract - Eigenvalues over 1 - or usually better is to specify the number of factors
 * 21) Rotation - Varimax (uncorrelated factors) or Direct Oblimin (correlated factors)
 * 22) Options - check these options:
 * 23) Sorted by size
 * 24) Suppress values less than .20
 * 25) After checking internal consistency and creating composite scores,  work out the correlations between the factors: In SPSS - Analyze - Correlate - Bivariate - and put all of the composite factor scores in.
 * 26) Report the factor correlations in the EFA results

Exercise 1: Smoking attitudes
What are the underlying dimensions of attitudes towards other people smoking?

Data file: data_14_1.sav | .sps | .spo Allen & Bennett Ch 14 (pp. 193-204)

Rating scale

Items


 * PAF, Varimax or Oblimin rotation (5 items; N = 107), 2 factors:
 * Smoking acceptance (3 items)
 * Smoking in food/restaurants (2 items)

Exercise 2: Classroom behaviour
Data file: behav.sav | .sps | .spo | .doc | .pdf Francis 5.6 (pp. 157-166)

Rating scale (1 to 5)

Items


 * Solution: 3-factor PAF, Oblimin rotation (13 items; N = 1500), Drop items 6 and 7
 * Sociability
 * Task orientation
 * Settledness
 * See also: Lecture slides: 33-36, 38, 40, 42, 52, 54, 56, 58, 78-80

Exercise 3: Managerial skills
Data file: manage.sav | .sps |  .spo | .doc | .pdf Francis 5.12 (pp. 166, 258)

Rating scale

Items


 * Solution: 3 factor PAF Oblimin (13 items; N = 114):
 * Feedback
 * Time Management
 * Problem Solving