ABSTRACT
This chapter examines and scrutinise the principles and methods of factor analysis. This is because factor analysis is the statistical method used for the construction of many of the best psychometric tests, although there are other methods of test construction. In a critique of the article by Joel Michell, D. Laming, with the traditional Cantabrian opposition to factor analysis, has described the factor analysis of psychometric tests of intelligence as a nonsense. It assumes that the normalised scores are normally distributed, an assumption which Laming regards as deceitful. A factor is a linear sum of variables. These combinations mathematically account for the correlations in the correlation matrix, and may be thought of as constructs, dimensions or vectors. The factor pattern is the set of factor loadings as a whole - the weights on the factors to predict the variables. This, in orthogonal solutions, is the same as the factor structure.
