Dear Naw

you have only got four variables? There is not a whole lot you can do with them, I am afraid. You may run an EFA to check if they can be considered one factor (if that is what is intended in the first place). Correlation analyses should be carried out. As far as multiple regressions are concerned, ask youself this: Can one of the four logically be "explained" by the other three? Do you even want to "explain" one of them? There is no point attempting SEM.

Please clarify if I understood you correctly.

Kind regards

Marcel

The statistical approach you use will be determined by your hypotheses (then the nature of your data). From your question I assume that your hypotheses concern the relationships between two or more variables (eg extroversion, happiness). In this case, as stated by Marcel, a correlation analysis will be appropriate. I further assume that the four variables you have are made up of multiple items (each item being measured on a likert scale). If this is the case and your data fit the assumptions for parametric analysis then you could use Pearson for your analysis.

Regression, path analysis and SEM are all used to determine relationships between variables but in most cases these deal (at least in part) with the impact of multiple variables (independent variables) on another (dependent variable).

Dear Mr. Grieger and Mr. Davies,

concerning the fact that I have four variables and that they are insufficient to "do a lot" with. Would you clarify for me this point? Let's say I had 6 or 8 variables, what difference would this make in the selection of statistical method?

Concerning the hypotheses/ objectives, they in part to try and examine the extent to which each variable is associated with another. A variable may be a mediator and another may be a moderator. These are interconnected and the aim is to examine exactly how.

Concerning the question of whether one variable can logically be explained by another or others, I am assuming this is the case of mediation/moderation or even simple correlation? This is an assumption based on my scarce knowledge of statistics.

I apologize for the insufficient information provided initially. I hope this makes things more clear.

Dear Naw

I believe I rather missunderstood you. We use "variables" synonymously with "items", so I thought you only had four items. That is why I said you could not do much with them. I figure you have four variables (we would call "factors" or "constructs") consisting of various items.

In that case, I have not much to add to Jason's answer. Structural Equation Modeling will definitely answer all questions concerning " the extent to which each variable is associated with another".

One additional question just so that I know we are on the same page: Have you validated your four variables already or are the still at an exploratory stage? If you are past that, I would go directly for CFA, resp. SEM (if your variables are latent). If they are not, path analysis might be a good idea.

Please come back to me if you are interested in literature on specific SEM-related questions. I will try my best to provide you with useful sources.

Best wishes

Marcel

Mr.Grieger,

Thank you for the clarification. I have a better idea where what path to take in the analysis of my data. Concerning validation, I am still at the exploratory stage.

As for the literature, I would be thrilled if you provided me with such sources.

Thank you

Dear Naw,

in that case, I would recommend reading up on Exploratory Factor Analysis before anything else. I am now assuming that your variables have not been used widely resp. that you created them. I have a great body of literature on various EFA-related topics, but if I had to name of few books/articles to start with, it would be these:

Bandalos, D. L. & Boehm-Kauffman, M. R. (2009). Four Common Misconceptions in Exploratory Factor Analysis. In C. E. Lance (Hg.), Statistical and Methodological Myths and Urban Legends. Doctrine, Verity, and Fable in the Organizational and Social Sciences (S. 61–87). New York NY u.a.: Routledge.

Field, A. P. (2017). Discovering statistics using IBM SPSS statistics (5th ed.). London: SAGE.

Hair, J. F., Black, W. C., Babin, B. J. & Anderson, R. E. (2010). Multivariate Data Analysis: A Global Perspective (7. ed.). Upper Saddle River NJ u.a.: Pearson.

Young, A. G. & Pearce, S. (2013). A Beginner’s Guide to Factor Analysis: Focussing on Exploratory Factor Analysis. Tutorials in Quantitative Methods for Psychology, 9(2), 79–94.

Feel free to contact me once you have had a look and these. Should there be a specific question I can help you out with, just let me know.

Kind regards

Marcel

Thank you for all your help, Mr.Grieger. I am sure the sources you have provided will also be of great help to me.

Thank you