I am currently doing research on the moderation effect of a certain moderator variable. I want to perform a normal linear regression analysis and thereafter a moderated regression analysis using the PROCESS Macro of Andrew Hayes. In order to perform a regression analysis my data needs to meet some assumptions (e.g. linearity, homoscedasticity and so on).
I am now wondering if each subgroup needs to meet the assumptions separately per variable that is put in the moderated regression analysis or not? Would it be fine if I test the assumptions for the whole group at once (and not making a difference between the subgroups)?
Additional information:
- My moderator variable has either value A or B, thus, I have two subgroups (Group A + Group B)
- The normal linear regression model is performed on the whole group. No differences are made between the two subgroups.
- I have several IV's and only one DV
Thanks in advance for helping me!
The assumptions are for the overall model. See the attached search for good references. Best wishes David Booth
Multiple linear regression is based on the following assumptions:
A linear relationship between the dependent and independent variables. ...
The independent variables are not highly correlated with each other. ...
The variance of the residuals is constant. ...
Independence of observation. ...
Multivariate normality.
https://corporatefinanceinstitute.com › ...
Multiple Linear Regression - Overview, Formula, How It Works
There are four assumptions associated with a linear regression model: Linearity: The relationship between X and the mean of Y is linear. Homoscedasticity: The variance of residual is the same for any value of X. Independence: Observations are independent of each other.
https://sphweb.bumc.bu.edu › otlt
Simple Linear Regression - SPH
Similar to what occurs if assumption five is violated, if assumption six is violated, then the results of our hypothesis tests and confidence intervals will be inaccurate. One solution is to transform your target variable so that it becomes normal. This can have the effect of making the errors normal, as well.Sep 8, 2020
https://towardsdatascience.com › wh...
What Happens When You Break the Assumptions of Linear Regression ...
https://statistics.laerd.com/spss-tutorials/dichotomous-moderator-analysis-using-spss-statistics.php,kindly check this link.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2908713/,u can follow this link.
https://link.springer.com/article/10.3758/s13428-018-1088-6,this link may be helpful.
In statistics and regression analysis, moderation occurs when the relationship between two variables depends on a third variable. The third variable is referred to as the moderator variable or simply the moderator.
https://en.m.wikipedia.org › wiki
Moderation (statistics) - Wikipedia
Feedback
https://statsnotebook.io/blog/analysis/moderation_interaction_regression/,this link may be useful.
Faraed Salman David Eugene Booth Thanks! The answers were really helpful. As I understand, the assumptions need to be met for the model. As such, if I, for example, test two models
1. IV1 --> DV with moderator M
2. IV2 --> DV with moderator M
the assumption need to be met for each model separately.
Great thanks dr. Ellen Kroon for ur appreciation to to my participation in this valuable discussion.
Assumptions of the moderation model include OLS regression assumptions, and homogeneity of error variance. The latter assumption requires that the residual variance in the outcome that remains after predicting Y from X is equivalent across values of the moderator variable.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2908713/
Check https://statistics.laerd.com/spss-tutorials/dichotomous-moderator-analysis-using-spss-statistics.php
https://link.springer.com/article/10.3758/s13428-018-1088-6
I am not sure whether I could help you but my guess is that you should start with the concrete task: as far as I understand it is a medical task .I find that you have two major problems : (i) what is controller (cause) and what is the effect; (ii) what is independent variable in physiology. I am afraid that in physiology there are not independent variables. If otherwise, they should be subject to Central Limit Theorem, which prescribes normal distribitution. The latter, however, allows the deviaions from the mean to be spread up to infinity. However, if so, peoples hight for example, could to vary in an uncontrolled way: thus, a peron of fhight 8 meters is possible, though it is extremely rare event. However, the probabilities are defined as current density in probabilty space. Thus, in the course of time, a degeneration of a mankind is inevitable and this is bound to happen in various ways. On the other hand, all probabilties are defined in the interval (0,1) . Thus, their generic property becomes boundedness, whatever the variations (linear, non-linear, small intensity, large intensity, rate of development, etc.) could be. About the relation cause-effect: in medicine, what is to be cured is the effect: when you have a headake, you first take a pill to ease the pain, regardless to whether it comes from bad air, bad, mood, or tumor in the brain. Next comes what could cause it: (i) whether it is accidental (a result of a lot of circumtances happended simultaneosly); (ii) a result of small but persistent deviations from homeostasis; (iii) organic(ir)reversible damages. Good luck!
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression