Questionable single component factors may include only 4 factors, I recommend that you perform confirmatory factor analysis to determine model fit values

Eigenvalue ≥1 , scree plot are important. another important metric to keep in mind is the total amount of variability of the original variables explained by each factor solution. t is also important that the rotated factors make theoretical sense to the researcher.

We want to see in factor analysis a smaller number of interpretable factors that explain the maximum amount variability in the data. So , you are the best judge to consider to take either 4 or 6 factors according to your study needs.

You can take 4 or 6 based on above criteria. Don't think only about analysis, think about reality or theory and study context too.

Afnan Ghanyem, first up, please note that it is "exploratory", not "explanatory" factor analysis that you appear to be using. The difference in those words might be important in terms of what you believe you are doing.

Second, can you indicate how you decided that you had six factors in your data, please. A lot of people believe that the number of factors is indicated by the number of eigenvalues > 1, but for about 30 years that has been shown to be a very poor way of determining number of factors, often suggesting that there are more factors than is the case.

The scree plot might provide a better idea of the number of factors in many cases, but a procedure called parallel analysis is more definitive. That procedure is available through freely available software on the web and, although it looks daunting at first, is really quite simple to use.

There are some YouTube presentations that could be helpful in relation to parallel analysis.

Hi Afnan Ghanyem

The interpretation of Factor analysis (FA) is based on rotated factors, rotated loadings, and rotated eigenvalues. Generally, factors with an eigenvalue greater than one (with a variance greater than the original standardized variables) are kept since these components explain most of the variance (Davis 1986).

Davis JC (1986) Statistics and data analysis in geology, 2nd edn. Wiley, New York

F. Bouteldjaoui, please forgive me if I seem to be disagreeable, but I think there are other perspectives to the ones you've outlined above. For example, the notion that the number of eigenvalues > 1 indicates the number of factors in a data set, although very widespread, has been criticized for approximately 50 years, usually because it suggests that there are too many factors in a data set.

Instead, for perhaps 40 or 50 years (forgive me for not being precise), the scree plot has been recommended as a better alternative, and, more recently still, parallel analysis (and some other procedures) have been strongly recommended for identifying the number of items in data.

Parallel analysis is not difficult to conduct, and, in my experience, has generally been very informative and reassuring. The following site contains an "engine" for conducting parallel analysis:

https://analytics.gonzaga.edu/parallelengine/

I'm not sure what you mean by "rotated eigenvalues". I am pretty sure that eigenvalues are robust / stable whether or not factors are rotated.

Robert Trevethan thank you for your explanation

F. Bouteldjaoui, you are very welcome. I'm glad you don't seem to have taken offence at what I wrote. I generated my post with the best of intentions.

All the best with your own research!