Dear Colleagues
I have three aspects in my conceptual model such as: 1- organiazational culture and its element , 2- national culture and its elements , and 3- personal tendencies and its elements.
Should i put all the independent variables together in the spss while testing regression analysis , or it is acceptable to put the independent variables for each aspect or category alone? that means i put elements of organizational culture in spss , then elements of national culture , then elements of personal tendencies - and is it acceptable not putting all these elements together in spss?
I believe that would depend on your research question and hypotheses. I would probably include all at once and maybe even, if I was interested in prediction, do a lasso regression to get the order of the predictors. Best, D. Booth
I think it depends on the aim and hypothesis of yours research paper. if you mention all the dimensions of your variables you should test them. however if you consider your variables without their dimensions it is better to neglect dimensions.
I disagree if we are talking about experimental data. I believe that the entire experiment should be described. Best, D. Booth BTW in your case why did you collect information on variables you didn't want to study. If you studied them for any reason other than "what's on the shelf today" report it. If it was "what's on the shelf" get a better research design. D. Booth
It depends on the main and sub-hypotheses of your research. You did not explain the nature of the relationship among the variables (e.g.organizational culture and national culture ), whether it is a causation or association. You may use LASSO regression which takes place in three popular techniques; stepwise, backward and forward technique. You may use one of the variables as a moderator or a mediator.
I would prefer that you do path analysis structural equation modeling using (Analysis of Moment Structures) AMOS or Lisrel or Mplus.
Thank you for all colleagues - i am using the realtion ship of causing between independent variables and the dependent variable.
Can i use multicollinearity instead of regression analysis?
Waleed, sorry but multicollinearity is the name of a common error not a method. With no description of your DVs and IVs I am at a loss to see how anyone could suggest a method.at this point. BTW since lasso was mentioned I have attached my most recent lasso papers. One is in press the other is finishing review. This is a very good method if you are building a predictive model. So please describe your variables and if you want a predictive or explanatory model(see the references in my papers). Best, David Booth
It depends on your objective task, what exactly you'd like to test by performing regressions.
If a regressor can be taken as a reasonable cause for another variable, then, yes you can use it.
But you have to be careful with the output, maybe sometime it will not be a quality one...
No doubt that it depends on your research question(s). However, if you test each aspect in separation actually you are omitting the explanatory power of other aspects which is statistically wrong.
For more detail see....
Halcoussis, D. (2005)
IMHO, please write your regression equation first, and we might help you further. Or at least define the indicator of each variable. Base your model through some prior research and add your novelty.
It depend on your research, but we take independent variable independently and dependent variable may be combine or consider separately it depend on you.So take all independent variable separately.
Waleed, your variables are at two different level of analysis -- group and individual levels, therefore you will have to do a cross-level analysis to test your model.
Look at this article's procedure as an example:
Chen, S. L. (2018). Cross‐level effects of high‐commitment work systems on work engagement: the mediating role of psychological capital. Asia Pacific Journal of Human Resources, 56(3), 384-401.
- Badges
- Science topic
- Similar topics
- Social Science
- Sociology