The threshold is mostly taken at 0.40.

But check these two references.

1) Williams B, Onsman A, Brown T (1996). Exploratory factor analysis: A five-step guide for novices. J Emerg Prim Health

Care, 19, 42–50.

2) Gie Yong A, Pearce S (2013). A beginner’s guide to factor analysis: Focusing on exploratory factor analysis. Tutor Quant Methods Psychol, 9, 79–94.

Hi Shamina,

there are already some years-old threads about factor loadings that may contain informative content. IMHO, any cut off or rule of thumb has little justification. Rather, the loadings communicate the relationship between an observed indicator and your intended factor. Hence, your theoretical judgement should help to interpret/decide whether this is support for measurement validity of the measure. I doubt that .5 has any rational basis unless I assume that there is a huge conceptual gap between the factor and the indicator and/or that the measurement error is huge (namely makes ab 75% of the observed variance in the case of a .5 loading).

Such low loadings often ocurr in measurement/factor models which are misspecified but defended anyway. Hence, select a subset of indicators which can be defended as viable candidates for the supposed factor *a priori* and whose test is hence reasonable. In such cases you will get a) a set with a prior reasonable factor structure, b) hopefully a fitting model (which supports this structure"9) and c) strong loadings with - again - support the validity of the indicator and the intended meaning of the factor.

There is no reasonable statistical analysis without clear and precise theoretical basis :)

Best,

Holger

Usually cut off at 0.49 and below. But there is no clear rule, as some state 0.3; some 0.4 or 0.5.

Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4, 272-299.

See: https://pareonline.net/pdf/v10n7.pdf

Perhaps it is better to test for the factor reliability using Cronbach Alpha (should be at least 0.7 for a good factor) to determine if an item should be loading onto the factor or not...