Is standardized coefficients enough to explain the effect size or Beta coefficient or will I have to consider unstandarized as well? Could you clarify- when do we consider unstandarized coefficient and why? Is it worthwhile to consider both standardized and unstandardized regression coefficients?
Dear Rashid
Happy new year.
We suppose you have ten independent variables in your study and each of them is measured by different scales. When you want to find Independent variables with more impact on your dependent variable you must use standardized coefficients to identify them. Indeed, an independent variable with a larger standardized coefficient will have a greater effect on the dependent variable. While this is not true for unstandardized coefficients.
If measurement scale of independent variables are same, the results of the analysis for both methods will be the same.
Actually, in interpretation of your regression results the value of any unstandardized coefficient denotes the change in the dependent variable with a unit increment in the independent variable. But you can not compare them in terms of impact on the dependent variable.
Therefore, both of them are useful but each one in one field.
Unstandardized coefficients are useful in interpretation and standardized coefficients in comparison of impact of any independent variable on the dependent variable.
"Effect size" can mean many things. If you are interested in standardized effect sizes, you need a standardized coefficient; if you are interested in unstandardized effect sizes, that's what the unstandardized coefficient is.
Unstandardized coefficients are sometimes easier to understand. For example, if you can describe a regression result as saying "Group A earned $100 more than Group B", that's easier to intepret than "Group A earned .173 standard deviations of the dependent variable more than Group B"
Dear Rashid
When measurement scale of your independent variables are different, you must use standardized coefficients to interpret and compare their effects on your dependent variable. I usually prefer using standardized coefficients instead of unstandardized coefficients.
Thank you, Seyed Mahdi!
Your answer helps me a lot! Could you explain little bit more? For example, I considered ten independent variables in my study and each of them were measured by different scales, so you mean in that case it would be better to use standardized coefficient. Isn't it so?
Sometimes statisticians encourage to use unstandardized coeffient, why do you prefer using standardized one?
Neither standardized nor unstandardized coefficients are inherently better. Each is appropriate for different kinds of situations. Seyed Mahdi and I have described some situations when unstandardized coefficients are more useful and some when standardized coefficients are more useful. Which one you choose will depend on what you are comparing and what the goals of your research are.
Dear Rashid
Happy new year.
We suppose you have ten independent variables in your study and each of them is measured by different scales. When you want to find Independent variables with more impact on your dependent variable you must use standardized coefficients to identify them. Indeed, an independent variable with a larger standardized coefficient will have a greater effect on the dependent variable. While this is not true for unstandardized coefficients.
If measurement scale of independent variables are same, the results of the analysis for both methods will be the same.
Actually, in interpretation of your regression results the value of any unstandardized coefficient denotes the change in the dependent variable with a unit increment in the independent variable. But you can not compare them in terms of impact on the dependent variable.
Therefore, both of them are useful but each one in one field.
Unstandardized coefficients are useful in interpretation and standardized coefficients in comparison of impact of any independent variable on the dependent variable.
Instead of standardized regression slopes, you may consider using semipartial correlation when SE is not reported in articles. See Aloe & Becker's paper(2011).
The standardized coefficients indicate Beta values in standard deviations. You can interpret like one standard deviation change in independent variable will have Y standard deviation effect on dependent variable
Both are useful if are used appropriately. For example, if we want to compare the effect of predictor variables measured in different unit of measurement on our response variable, we are expected use standardized estimates. When we standardize our estimates the unit of measurement is going to be standard deviation regardless of the real measurement, which enable us to compare variables measured by different unit of measurement. Otherwise, as unstandardized estimates are best for interpretation.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression