My question is very simple, What the different between the standardized coefficients (this coefficient has been obtain in multiple regression, and take a scale (-1,1). And the semi PARTIAL coefficient could be obtain in the model regression,but with this number you can take a variable contribution in percent (%). But in both cases this coefficients explain the variable's contribution. Both also control for the effect of one or more another variables. Thus, you could compare model results.


marked as duplicate by jbowman, Michael Chernick, gung r Sep 21 '17 at 18:03

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • $\begingroup$ I think you will find the information you need in the linked thread. Please read it. If it isn't what you want / you still have a question afterwards, come back here & edit your question to state what you learned & what you still need to know. Then we can provide the information you need without just duplicating material elsewhere that already didn't help you. $\endgroup$ – gung Sep 21 '17 at 18:03
  • 1
    $\begingroup$ @gung, is it really a duplicate? The OP is asking about part correlation, not partial. Perhaps let them search first part semi-partial... $\endgroup$ – ttnphns Sep 21 '17 at 18:11
  • $\begingroup$ Dinosca, are you really after semi-partial correlation and not partial? Are you sure? $\endgroup$ – ttnphns Sep 21 '17 at 18:12
  • $\begingroup$ The question about the semipartial correlation has been asked before, too: stats.stackexchange.com/questions/112891/… $\endgroup$ – statmerkur Sep 21 '17 at 18:19
  • $\begingroup$ My question is about semi partial coefficient, not correlation semipartial... I think so my question is no duplicate, because I search in browser page and i dont find the answer @ttnphns $\endgroup$ – Dinosca Sep 21 '17 at 18:59

They are both measures of effect size, and what you have described is true of many measures of effect size. Otherwise, they are quite different (for multiple regression). In single regression, they are the same.

The standardized coefficient is a regression coefficient in the units of standard deviation. That is, a for a 1-standard deviation increase in X, we expect Y to increase by B standard deviations. This can be useful to compare the effect of variables if there is some reason to think moving a standard deviation in one variable is equivalent to moving a standard deviation in another (which, so argue, is not often true). This says nothing about how much variance in Y is explained by X; it just quantifies the relationship between X and Y in a unit-free way.

The squared semi-partial correlation is a type of R2 measure that described the proportion of variance in Y that is explained uniquely by X. It says nothing about the magnitude of the relationship between X and Y; that is, it tells us nothing about how Y changes as X changes. You need to have an understanding of explained and unexplained variance to make any sense of this measure; it describes the reduction of randomness in Y due to your modeling choices.


Not the answer you're looking for? Browse other questions tagged or ask your own question.