Hi,
I'm fairly novice in statistical inference and hypothesis testing and I have a basic understanding of regression models. I'm working on my first quantitative research in organizational behavior. My research model consists of 4 variables of interest and 3 hypotheses. Suppose we have the following:
H1: increase in X has a negative effect on Y
H2: Y and Z are correlated
H3: W moderates the negative effect of X on Y and therefore Z
Suppose I gather sufficient quantitative data for each of these variables for each subject. How do I formulate each of these hypotheses to test with statistical models? More broadly, what should I be learning to understand how to design and conduct a test for each of the hypotheses.
Do you recommend any straightforward topic or material so I can focus on learning the vital concepts? Most books and online courses tend to cover everything which means it would take a lot of time for me to just know which and which I should have studied and used.
Any advice would be really appreciated
sincerely,
Ashkan
H1 and H2 could be examined by computing correlation coefficients and testing the null hypothesis that each correlation is zero in the population.
H3 (moderation) could be tested by using moderated regression where you regress Y on X and W and in addition include the interaction (product) term XW as a third predictor/independent variable to test the hypothesized moderation effect. (You can do the same for Z as an outcome variable.)
When the regression coefficient for the interaction term XW is significant, this means that there is a statistically significant moderation effect. To understand the meaning of the interaction, it useful to plot the relationship between X and Y at different levels of W. This is described in great detail with SPSS examples in Aiken & West's (1991) book on interactions in multiple regression. The procedure is somewhat complicated, also due to the fact that centering the predictors prior to forming the interaction term is often necessary for meaningful results. You therefore might want to study this topic in somewhat more detail before proceeding with your own analyses. You want to be sure that the results are correct and that they make sense and are easily interpretable.
Dear Ashkan Taravati ,
it is clear that you did not explain well
It is very important for us what are X, Y, etc.?
What is their measure? They are ordinal or nominal or Scale.
The answer to this question helps us to know if you are using Contingency analysis or, for example, GLM or LM. Or even regression and correlation
I suggest you put your data file here.
It is very important and vital to see your variables.
Christian Geiser has given you good advice--and that Aiken & West book he mentioned is very good indeed. To find relevant examples online, you might try searching on , replacing package with the name of the stats package(s) you use.
- Badges
- Science topic
- Similar topics
- Medical Science Nursing
- Dissertations