Exploratory Factor Analysis in Urban and Regional Planning

Exploratory Factor Analysis (EFA) is a powerful and commonly-used tool for investigating the underlying variable structure of an urban phenomenon. EFA is commonly employed to reduce many variables into a smaller set of dimensions, hoping to extract dimensions which will explain a good portion of the variance in the original data. Despite the widespread use of EFA, researchers often make questionable decisions when conducting these analyses. Appropriate use of EFA requires a series of thoughtful analytical decisions on the appropriateness of data for EFA, sufficiency of sample size, best extraction techniques to be used, the decision about the number of factors to extract and interpret, the appropriateness of particular rotation techniques, and the calculation method for the factor scores. This article reviews the major design and analytical decisions that must be made when conducting an EFA. The following six steps are proposed for an EFA application in urban and regional research area: 1. Selection of variables/indicators
Variable selection should be guided by theory and/or the findings from the past research.
2. Controlling the adequacy of the data for EFA
The researcher must decide how large the sample must be. This is a necessary consideration to obtain a stable factor structure. KMO and Bartlett’s Test of Sphericity check the adequacy of the data for EFA
3. Deciding the method and the number of factors to extract
There are several factor analysis extraction methods to choose from. The PCA is generally used for factor extraction. However, it is suggested that if data are relatively normally distributed, maximum likelihood is the best choice and if the assumption of multivariate normality is severely violated, then it is suggested to use the principal factor methods. After extraction the researcher must decide how many factors to retain for rotation. Both over-extraction and under-extraction of factors retained for rotation can have harmful effects on the results. The default in most statistical software packages is to retain all factors with eigenvalues greater than 1. Another test for factor retention is the scree test. The number of factors chosen should be such that they explain at least 60 percent of the data variation.
4. Selecting a factor rotation The goal of rotation is to simplify and clarify the data structure. There are two extraction orthogonal and oblique. Orthogonal rotations produce factors that are uncorrelated; oblique methods allow the factors to correlate.
5. Interpreting and naming the factors
Interpretation involves the researcher examining which variables are attributable to a factor, and giving that factor a name. Usually, at least two or three variables must load on a factor so it can be given a meaningful interpretation. The naming of factors is a subjective, theoretical, and inductive process. Therefore, the researcher should be very familiar with the issue at hand.
6. Computing the factor scores
Finally, factor scores are computed for the extracted factors.
In this paper, after reviewing the EFA processes and procedures, a real application of the method in assessing the social sustainability at community level