Factor analysis: a problem with age and factor descriptions of the motor domain [proceedings].

Factor analysis, like any other research tool, if it is to give meaningful results must be applied in an appropriate manner. A subtle yet recognised distortion of simplest factor structure is achieved by factoring a correlation matrix holding spurious correlations. Experimental dependence is most easily observed in an inter-correlation matrix when the total score of two or more variables in the battery selected for factor analysis is added to that battery. Spurious correlations can also arise when the effect of a variable, not a member of the domain selected for factoring, is recognizable across the majority of variables selected for analysis. Such a variable is labelled an 'underlying variable' and is defined as a variable that is (i ) not a member of the variable list to be factored and (ii) has significant correlational relationships with the majority of variables in the battery selected for factor analysis.