Log transform data in PLS step by step guide is needed

Frequently asked questions about PLS path modeling.
Post Reply
aznur
PLS Junior User
Posts: 2
Joined: Thu Sep 19, 2019 10:18 pm
Real name and title: Aznur

Log transform data in PLS step by step guide is needed

Post by aznur » Thu Sep 19, 2019 10:34 pm

Hi all,

Need your help.

I read respond by Dr. Becker in smartpls forum to log transform a single item construct that is non normal.

I ran within subject experimental design in problem solving activity, n =84 with 3 raters that provide the scores . Rater 2 shows p value less. 05 indicate non normality. The p value of rater 2 is 0.003. This score is the DV of the study. Rater 3 shows slight below 0.05 with p value of 0.04.

After log transform the data of rater 2 and 3 are normal.

Questions:

1.should i log transform all values of the 3 raters ?

2.Can I proceed with log transform values in PLS and use it to test hypothesis with IVs that are not using log transformation value? Or I should log transform all constructs? What is a step by step correct process should be done in PLS?

3. How to interpret results from log transform values in PLS.

3.I read log transform data is hard to interpret. See journal by Feng, C., Wang, H., Lu, N., Chen, T., He, H., Lu, Y., & Tu, X. M. (2014). Log-transformation and its implications for data analysis. Shanghai archives of psychiatry, 26(2), 105-109.

I am concern on the effects as listed in the journal especially on the methodology and interpretation of results. How it effects my results later on in PLS?

Can anyone recommend readings related to log transformation in PLS. Sample of papers that use log transforms in PLS.

Thank you for your time.

Post Reply