1

I have a few questions about Ordered Logistic Regression Model (OLR). In general, how do I know if this is my model is good, and how do I enhance it?

Using this example from https://stats.idre.ucla.edu/r/dae/ordinal-logistic-regression/, if these are the outputs I am looking at:

## R OUTPUTS
## Call:
## polr(formula = apply ~ pared + public + gpa, data = dat, Hess = TRUE)

## Coefficients:
##          Value Std. Error t value
## pared   1.0477      0.266   3.942
## public -0.0588      0.298  -0.197
## gpa     0.6159      0.261   2.363
## 
## Intercepts:
##                             Value  Std. Error t value
## unlikely|somewhat likely     2.204  0.780      2.827 
## somewhat likely|very likely  4.299  0.804      5.345 
## 
## Residual Deviance: 717.02 
## AIC: 727.02


## R INPUT
exp(cbind(OR = coef(m), ci))

## R OUTPUT
##            OR  2.5 % 97.5 %
## pared  2.8511 1.6958  4.817
## public 0.9429 0.5209  1.681
## gpa    1.8514 1.1136  3.098

What can I do to enhance/improve my OLR model in R? What output values should I be focusing on?

Does improving my OLR model mean looking at my P values and coefficients and removing the ones that are not statistically significant?

I am stuck as I don't really know what to do after I run the model.

Thank you.

AdamO
  • 52,330
  • 5
  • 104
  • 209
conender3
  • 11
  • 1
  • 2
    What do you want to do with this model? – dimitriy Jan 22 '18 at 17:54
  • 2
    "looking at my P values and coefficients and removing the ones that are not statistically significant" is almost never a good approach in any regression. – EdM Jan 22 '18 at 19:26

0 Answers0