Ahmed Shiyam,

Hello again.

Different methodologists provide quite different advice about the recommended size of communalities. Some, for example, say that they should ideally be > .50.

I have found such a high level to be counterproductive as a way of selecting items. My usual strategy is to begin by identifying the likely number of factors in the data (by using the scree plot in conjunction with parallel analysis), then to eliminate items in the pattern matrix that have very low factor loadings (e.g., < .30), then re-run the EFA again and eliminate items with low loadings (raising the bar as I go, perhaps eliminating items with loadings < .35, then < .39), and also eliminate items that cross load - and keep doing that until there are clear factors, each with items that load > .40 on them alone.

At the end of the process, I think that the lowest and highest communalities, and the mean of the communalities, should be reported.

Feel free to get back with questions - and I would not mind if you are given different kinds of advice from others.

In Exploratory Factor Analysis (EFA), communalities represent the proportion of variance in each observed variable that can be explained by the factors. Generally, communalities above 0.6 are considered ideal, while values between 0.25 and 0.4 are often deemed acceptable. However, the minimum accepted value can vary depending on the context of the research and the specific criteria of the analyst. For instance, some sources suggest that communalities less than 0.5 may be too low, as they indicate that less than half of the variable's variability is shared with other variables. In practice, items with communalities less than 0.2 are typically considered for removal, especially in the context of dimension reduction techniques which seek to identify items with shared variance. Nevertheless, it's important to consider the overall research project and the significance of each item before making a decision to exclude based on communalities alone. In cases where the common factors explain a significant amount of variance, a lower threshold might be justified. Ultimately, the decision should be informed by both statistical guidelines and the substantive meaning and importance of the items within the questionnaire.

I hope you are doing good!

https://www.jstor.org/stable/pdf/23033534.pdf

Refer to this article PLS-SEM: INDEED A SILVER BULLET (Joe F. Hair, Christian M. Ringle, and Marko Sarstedt).

Check Page for related details.

I hope it helps :)