Once we have decided on the number of factors, we retrieve the “factor loadings”. A factor loading is basically a correlation coefficient (see Correlation analysis) and, thus, it varies between -1 and +1 (where a value closer to -1 or +1 indicates a stronger correlation).

Factor loadings are given for each variable, for each factor separately. In other words, a factor loading shows how strongly a certain variable correlates with the given factor. There are no exact rules for deciding on when a loading is strong enough, but one suggested rule of thumb is below -0.5 or above 0.5.

Sometimes, however, a variable has strong loadings for more than one factor (called “cross-loading”). This can for example happen if you have not extracted enough factors, or if the factors are correlated. Sometimes a variable has weak loadings for all factors; this may suggest that this variable is weakly related to all other variables or that you need to explore an additional factor (or maybe even exclude this specific variable).