Timeline for Can independent variables with low correlation with dependent variable be significant predictors?
Current License: CC BY-SA 3.0
16 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 3, 2014 at 8:04 | history | edited | Nick Stauner | CC BY-SA 3.0 |
grammar, punctuation, extra tags
|
| Mar 25, 2014 at 16:31 | comment | added | Scortchi♦ | @Nick: I agree entirely; it's that judicious examination I'm recommending - of the full model before the stepwise procedure has mutilated it for its own inscrutable reasons, & in ignorance of the scientific or other considerations that might militate for a simpler model in this particular case. | |
| Mar 20, 2014 at 18:59 | answer | added | Greg Snow | timeline score: 8 | |
| Mar 20, 2014 at 15:47 | comment | added | Nick Cox | @Scorchi Here as almost everywhere else I tend to agree with you. But a 8-predictor fit carefully examined usually allows judicious choice of a simpler model without invoking any of the stepwise machinery. I wouldn't delegate a choice that should be sensitive to the underlying science to a program. | |
| Mar 20, 2014 at 15:34 | comment | added | Scortchi♦ | @Nick: If people use step-wise after finding the full model over-fits & then confirm that it's improved matters, I'm not inclined to "shoot rockets", but it worries me when they seem to use it for no reason at all without checking anything. | |
| Mar 20, 2014 at 15:24 | comment | added | Scortchi♦ | The weaker your grasp of statistics, the less you ought to be messing about with variable selection procedures. If your goal's to examine how each IV relates to the DV after controlling for the others, that's exactly what the coefficient estimates (with their confidence intervals) from the full model are telling you. Looking at variance inflation factors alongside indicates how correlations between IVs are contributing to the uncertainty. Use a cross-validated or adjusted coefficient of determination, $R^2$, to assess the predictive capability of the whole model & to check for over-fitting. | |
| Mar 20, 2014 at 15:20 | comment | added | Nick Cox | If tempted to use stepwise, reach for Frank Harrell, Regression modeling strategies Springer, NY, 2001 as an antidote. He's active on this site and likely to shoot rockets if he hears the word "stepwise". | |
| Mar 20, 2014 at 15:09 | comment | added | Elle | Scortchi - my grasp on everything stats related is greatly lacking :( None of the IVs are strongly correlated either with each other, or with the DV. I did put all 8 IVs into the stepwise regression analysis, and it excluded all the but two of the DVs (which indicates that they are not sig predictors?). T | |
| Mar 20, 2014 at 15:01 | comment | added | Scortchi♦ | Saying a step-wise regression "showed that only two IV can predict the DV" suggests you don't understand how it works. If two IVs are strongly correlated, & either predicts the DV about equally well, a stepwise procedure can remove one quite arbitrarily. What's the problem with using the full 8-IV model? | |
| Mar 20, 2014 at 13:22 | answer | added | miura | timeline score: 10 | |
| Mar 20, 2014 at 13:19 | answer | added | Nick Cox | timeline score: 4 | |
| Mar 20, 2014 at 13:16 | review | First posts | |||
| Mar 20, 2014 at 13:28 | |||||
| Mar 20, 2014 at 13:14 | comment | added | Penguin_Knight | Yes, it's possible. One reason is high sample size. Another reason is confounding: the main independent variable may show a low correlation with the depedent because it is counfounded by another independent variable. Once that confounder is added to the model, it can make the original independent variable change from not predictive to predictive (or predictive to not predictive, depending on the types of confounding.) Regression will fully agree with all correlation tests only when all independent variables are uncorrelated, that nearly never happens. | |
| Mar 20, 2014 at 13:11 | comment | added | Nick Cox | Your title and contents show some confusion between the terms "dependent" and "independent". Please check that my edit preserves your intended meaning. The fact that people get confused about which is which strengthens the case for more evocative terminology, such as "response" or "outcome" rather than "dependent variable". Finally on abbreviations note that for many people "IV" means instrumental variable. | |
| Mar 20, 2014 at 13:08 | history | edited | Nick Cox | CC BY-SA 3.0 |
added 2 characters in body; edited title
|
| Mar 20, 2014 at 13:00 | history | asked | Elle | CC BY-SA 3.0 |