#factor loadings

The current study used Principal Component Analysis, a multivariate statistical technique, to examine the physicochemical parameters impacting the density and distribution of two clam species in the estuary. For two years, the Kayamkulam estuary was monitored for twelve physicochemical characteristics. During the study period, there were variations ranging from 72.689 percent to 85.411 percent. Only significant variables with factor loadings greater than 0.60 were evaluated for interpretation. Salinity, sediment texture, and pH are significant guiding elements that determine clam density, according to the factor loadings on various parameters. Sand, temperature, organic carbon, and salinity were identified as the four main characteristics in the estuary using PCA. The PCA results revealed the environmental parameters that influence macrobenthos population and dispersion in the estuary.

In a confirmatory factor analysis, sometimes we need to decide whether an observed variable should load onto a given factor. One way to determine this is by examining the factor loadings. If the observed variables have low loadings on the given factor, it may indicate that the variable doesn't belong there. But how low is low? 

Kline (1995) suggests that .30 (standardized) or less is considered a low factor loading.

Kline also says that a standardized factor loading of .2 or lower in a CFA means you should re-specify the model; have the variable load onto another factor or eliminate it entirely. (p 186)

Kline, R. B. (2005). Principles and practice of structural equation modeling. Guilford press.