negative Factor loading- what does it mean

juneloweelyn · Post by **juneloweelyn** » Mon Sep 28, 2009 5:11 pm

Hello guys,

I am new here. Hope u can help.

If factor loading get as a negative figure, what does it mean?

Do i have to delete the item because it has a negative factor loading?

Diogenes · Post by **Diogenes** » Mon Sep 28, 2009 11:31 pm

Hi,

If i have a negative factor loading, what does it mean?
Formative indicators (outer weight): It could be a effect of multicollinearity between the indicators.
Reflective indicators (outer loading): It could be a reverse item (its meaning is the contrary of the others items of the same latent variable). If it is the case, you could invert the scale, for instance, in a five point Likert scale 1 is changed to 5, 2 to 4, …

Do i need to delete this item with negative factor loading?
Formative indicators: If you delete some indicator, the meaning of the latent variable could change.
Reflective indicators: You could delete, but if the absolute value is greater than 0.7, it is better to understand why the factor loading is negative. The calculation of AVE will be correct, but the composite reliability should be computed by hand with a positive signal (absolute value).

I hope this help.

Best regards,

Bido

ruchi · Post by **ruchi** » Fri Oct 05, 2012 4:41 am

Hello Sir

what if the 1) loading and weight of the formative indicator is negative and significant (checked also for multicollinearity, VIF is less than 2 and also the items are coded properly?). Should that indicator be removed. How should we interpret it.

2)what if loading is negative and significant for formative indicator but weight is not significant?

Thanks
Ruchi

Hengkov · Post by **Hengkov** » Mon Dec 31, 2012 11:29 pm

Hi,

Outer loading cannot negative: only 0-1.
If weight not significant but loading high (> 0.5), you retain it.

Regards,
Hengky

mloose · Post by **mloose** » Sat Jan 12, 2013 12:22 pm

Diogenes wrote:Hi,

If i have a negative factor loading, what does it mean?
Formative indicators (outer weight): It could be a effect of multicollinearity between the indicators.
Reflective indicators (outer loading): It could be a reverse item (its meaning is the contrary of the others items of the same latent variable). If it is the case, you could invert the scale, for instance, in a five point Likert scale 1 is changed to 5, 2 to 4, …

Do i need to delete this item with negative factor loading?
Formative indicators: If you delete some indicator, the meaning of the latent variable could change.
Reflective indicators: You could delete, but if the absolute value is greater than 0.7, it is better to understand why the factor loading is negative. The calculation of AVE will be correct, but the composite reliability should be computed by hand with a positive signal (absolute value).

I hope this help.

Best regards,

Bido

Currently, I am faced exactly with this problem.
Can you recommend me a literature source for your described remarks?

Thank you in advanced for your help!

Michael

Hengkov · Post by **Hengkov** » Sun Jan 13, 2013 9:13 am

Hi,

Check recent articles below:

Hair, J.F., Sarstedt, M., Pieper, T., Ringle, C.M. 2012. “Applications of partial least squares path modeling in management journals: a review of past practices and recommendations for future applications,” Long Range Planning, (45:5-6), pp. 320-340.

Kock, N., and Lynn, G.S. 2012. “Lateral Collinearity and Misleading Results in Variance-Based SEM: An Illustration and Recommendations,” Journal of the Association for Information Systems (13:7), pp. 546-580.

Latan, H. 2013. "Reporting Result of Partial Least Squares Structural Equation Modeling Analyses," Working Paper, pp. 1-38.

Best Regards,
Hengky