Hi everyone!
In my studies, I usually have small size (less then 100), and all these studies have typical design: choosing variables and setting them into a theoritical model and then testing if the observed data fits the model. Can I still use PLS- SEM since this is testing not predicting. And what does predicting actually consist of in a reasearch in Social Social (management or marketing)??? predicting how people would react??? It seems difficult to implement!! Please help me! Thank you!
Hello Soumaya,
Yes, PLS-SEM may be used to evaluate SEM models with respect to how well a given model allows explanation of the observed relationships in a data set. The PLS-SEM advocates claim it can work well with limited cases, but with fewer than 100 and more than 30 parameters to be estimated, I think you'd be asking quite a lot of your data.
The last questions you pose make it seem as if you may not have fully articulated your specific research aims. I would suggest that be your first order of business, perhaps working in consultation with someone more experienced at your institution. From there, the choices about data analysis methods, data collection requirements, and choices of specific variables and measures would follow.
Good luck with your work.
When in qualitative research, the number of variables is large or the subject is large, the research method moves towards formulating it. Because its results and stages can be predicted to some extent and a significant range can be defined for them. If you are sure of the correctness of your formula, it is better in the stage of interpreting and analyzing the data, to make several models based on acceptable and predictable assumptions of your statistical community, and based on the types of mediating variables. And compare the data based on all of these hypotheses and modeling. If the results are for qualitative research, you can compare them with the results of previous research, but if they are for quantitative research, you can select the results of each of your hypotheses by logical analysis and interpretation, in a more probabilistic way. That is, steer the modifier variables in a direction that results in quantitative, predictable results. Try not to interfere in this way in a meaningful way to your results and assumptions. But lead to statistically and relatively predictable results.
Your research should be driven by research questions. The answer to them will come from the analysis. If you have a project supervisor, discuss with them the development of research aims/objectives.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling