Factor with only one item from EFA I did an EFA and I recovered 4 factors but one factor has only one item with a loading of 1.00 and the lowest proportion of variance of all the factors. Further, this factor with a single item is very difficult to interpret and do further analysis. Is there a rationale to delete this factor and carry on with my analysis without it? Thank you all.
 A: Typically, you should never retain a factor with a single factor loading (especially in your case when that single item shares 100% variance with the latent factor, meaning they are empirically indistinguishable). This practice is standard and present in much of published research, and it is noted in highly cited EFA guidelines on determining the number of factors. For example, Fabrigar et al. (1999) and Costello and Osborne (2005) include the following criteria for determining the optimal factor solution:
1. Each factor must only contain items explaining at least 10% of the variance in its respective factor (i.e. loadings ≥ |.32| in magnitude)
2. Each factor is recommended to have at least three items loading ≥ |.32| (this one is especially relevant for you)
3. Your factors must be interpretable in a sensible way (also relevant in your case)
4. It is recommended that your final factor solution has no items that cross-load on multiple factors with a similar magnitude. For example, if one item loads on Factor 1 and Factor 2 with the magnitude of .40, it is better to remove it as it measures two distinct factors with the same magnitude
Finally, you may benefit from reading more by using this thread
