Hi, Anak Agung Bagus Wirayuda, for reporting the results of PLS-SEM, I would like to recommend you the following paper. Also it does not require a path coefficient greater than 0.2.

Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to report the results of PLS-SEM. European business review.

Hello Anak,

I think Wang Zheng 's response is pointing you in the right direction.

In factor models, the issue of just how strong the relationship between scores on a variable and scores on a factor need be to be judged as "salient," has many suggested thresholds and guideline for them, but there is no single, definitive, threshold that is universally accepted.

Raymond Cattell, a noteworthy contriibutor to factor analysis literature, discussed a data set in which he considered a variable loading of .15 as salient. That's quite low compared to many of the usual guidelines, but helps to make the point that context, the specific sift of variables, and the research purpose all contribute to what might be a sensible threshold. Assuming otherwise is, I believe, foolish.

Good luck with your work.

Dear Anak

My experience shows that if the path coefficients are way too high, you usually have an issue with the discriminant validity between the latent constructs. I recommend performing a discriminant validity test before you run the analysis in order to show that you are not regressing the latent variable on (a perceived by your survey participants) on itself.

Best

Eugene (Yevgen)

A minimum level is usually not required. However, you want to see a significant coefficient based on bootstrapping. For the significant paths, you may want to assess the magnitude for relevance (a path can be significant but hardly relevant if it is very weak).

Dear Anak Agung Bagus Wirayuda , I support Christian M. Ringle and other professors that "A minimum level is usually not required, however, you want to see a significant coefficient based on bootstrapping for the significant paths"

According to my papers, many times the coefficient is lower than 0.15 but significant when running with the bootstrap.

Kind Regards,

Hi,...

When the desired path is placed in a network of relationships, the path coefficient decreases or increases depending on the number of variables, it is important that the path coefficient is significant, not the value of the coefficient. The path must first be meaningful to talk about its value.

Very useful to follow. Good inquiry and answers. Thank you, professors.

Kind Regards,

The issue of path values of significance does not much feature in the literature but the level of significance ( expressed in terms of p-values). However, it is important to add that the ' meaningfulness criterion' should always be important because the stats alone do not make sense. Understanding your data in light of relevant theory helps address Simpson’s Paradox ( in which case inverse relationships emerge that defy causally logical inter-factor relationships ). In WarpPLS , a sign change indicates the presence of Simpsons Paradox.

http://warppls.blogspot.com/2012/05/simpsons-paradox-and-unexpected-results.html

Thanks for the information. Very useful indeed. What would be your take on if the value of path coefficient is negative after bootstrap. I have got few of my coefficients coming in negative after bootstrap is that a sign of big worry. Thanks