Once again, let us return to the hypothetical example of the political scientist researching conservatism. A factor analysis reveals (let us say) five distinct factors in the data (obtained from asking participants the 25 questions). For each of these factors, a factor score can be calculated (there are several ways of doing this). If we are correct in our theoretical assumption that there is a single construct that we have named conservatism, then all of those factor scores should result in a single factor when factor analyzed. If our theoretical speculation is confirmed and only one factor is identified, then we have strong support for the theoretical position that there is indeed a “thing” called conservatism, and our 25 items measure that thing. When we identify a factor composed of other factors, statisticians call this a higher-order factor.
Last Modified: 02/14/2019