2
$\begingroup$

I have performed a logistic regression with whether or not an athlete was re-contracted by their sports team as the DV. One of the significant predictors of the final model was draft order (OR 0.888). 

Coefficients:
                                  Estimate Std. Error z value Pr(>|z|)    
(Intercept)                     -120.69457   46.78377  -2.580 0.009885 ** 
Debut.first.year                   0.67977    0.21772   3.122 0.001795 ** 
Grouped.by.fives                  -0.11849    0.02361  -5.019 5.21e-07 ***
Draft.year                         0.06109    0.02334   2.617 0.008863 ** 
Maturity                          -0.65844    0.40981  -1.607 0.108118    
Games.second.season.DC             1.87716    0.34011   5.519 3.40e-08 ***
Interstate.vic.team                0.47625    0.19390   2.456 0.014044 *  
Rising.star                        1.50635    0.44429   3.390 0.000698 ***
Team.EOS.ladder.second.year.raw   -0.04403    0.02062  -2.135 0.032728 *

I understand that this indicates that being selected later in the draft results in a reduced odds of being re-contracted. There are 8 predictors overall therefore my question is, does this indicate that being selected later in the draft results in a reduced odds of being re-contracted when all other predictors are held constant? Does this then mean that draft order is associated with being re-contracted irrespective of rising star nomination, maturity, draft year etc?

The text books I have read and online forums are quite vague and I am struggling to understand the relationship between the regression coefficients.

Thank you!

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

No, what you are doing is forced entry regression, to be able to interpret it as you did, you will need a hierarchical regression. So you should first fit a model without draft order and then fit a model with draft order and compare if the fit is significantly better in the second case. If it is, you can then look at how much additional variability is being explained by including draft order.

Improve this answer
$\endgroup$
1
  • $\begingroup$ What does the OR for draft order tell me then? $\endgroup$
    – Courtney
    Commented Feb 29, 2016 at 22:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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