I am researching secondary data of panel data type using the SEM-PLS research model, can panel data use SEM-PLS? where the condition of each of my indicators has a different scale in one latent variable.
Hi Destria! Of course you can use panel data, but you can't use indicators with different scales for measuring one latent variable in SEM. You can try to change the scales of some of the indicators for applying the same scale. If it's not possible to do so, you must use another statistical method for your analysis.
thank you for your answer Vasilica Maria Margalina. Can I ask again why a latent variable cannot have indicators with different scales? Isn't the advantage of PLS that it does not assume a certain data distribution?
It's because of the PLS-SEM algorithm and it's not related to the distribution of the data. You will obtain a very low values for reliability and validity measures, such as AVE and CR. If you give more information about the variables and the scales of their indicators maybe I can suggest you a solution for the transformation of scales.
The latent variable concept that I use is company characteristics consisting of four indicators, (1) profitability as measured by earnings per share, (2) leverage as measured by debt of assets, (3) company age as measured by years of establishment, and (4) company size as measured by the natural logarithm of assets, thank you for helping Vasilica Maria Margalina .
The problem is that you're using as indicators financial ratios and other characteristics of the firm that can't bee mixed in a latent variable. It doesn't make sense not even from a theoretical point of view. My recommendation is to have one latent variable (e.g. financial characteristics of the company) and use age and size as control variables or variables to segment the sample for a multigroup analysis. You can also use these two variables a moderators.
- Badges
- Science topic
- Similar topics
- Books