Search results

  1. noetsi

    Identifying potential confounders

    I think its a confounder if it shares a relationship wit the DV and even one other predictor. I would not use univariate analysis to find confounders in any case. I would add it to the regression model and see what occurs.
  2. noetsi

    RESOLVED but still open for discussion: When to transform and when to use non-parametric tests?

    The problem is that can eat up your cases very quickly. I lost 7 percent of my cases with just 3 variables. And some argue if results are missing your results will be invalid, the MNAR issue.
  3. noetsi

    Need Help with Regression Analysis ?

    How is your dependent variable measured? These type of issues get a lot of attention in the academic and finance literature so you might want to see if such analysis has been run already. I think you will find there is no simple answer (or it would be built into existing models and the market)...
  4. noetsi

    RESOLVED but still open for discussion: When to transform and when to use non-parametric tests?

    Yes. Multiple imputations is the best way to address such.
  5. noetsi

    Comments on residuals

    I have about 30 variables. When you say the top 2.5 percent of values, what defines a value?
  6. noetsi

    Comments on residuals

    thanks Jake a lot. One problem is that I can tell from the literature when violations occur. But not when the violations are serious enough to matter. I am not sure with thousands of data points if normality and hetero make a difference because of the asymptotic nature of regression. I use...
  7. noetsi

    RESOLVED but still open for discussion: When to transform and when to use non-parametric tests?

    I don't know which is more technically correct, but in economics its well accepted to use say logs with skewed data. I assume if its that common in a highly advanced field methodologically there must be a good reason to do so... Its not only violations of assumptions that are involved...
  8. noetsi

    Comments on residuals

    For a major major project and I want to be sure I get this right. I only care about the slope coefficients. I am testing the assumptions with a large population (about 4500 cases). The dependent variable is logged income at closure. I don't see any obvious heteroskedasticty....although lots of...
  9. noetsi

    What do infinite t values mean?

    I found the error. I accidently included weekly earnings as a predictor and weekly earnings end up being the DV. Sorry for the confusion.
  10. noetsi

    What do infinite t values mean?

    The r square is 1 so I assume I am getting perfect predictions. But I can't figure our how. I have thousands of cases and maybe 30 predictors. None looks problematic I can see.
  11. noetsi

    What do infinite t values mean?

    Why is the standard error zero :) Seriously I did not notice that until a few moments ago. But I still don't know what is causing it or why logging the DV gets rid of the problem. Many of the variables in the model have tolerances beyond .97 - I have never read this can make the SE zero.
  12. noetsi

    What do infinite t values mean?

    Other than something is wrong with your model. I was running a series of variables today with income at close as the dependent variable and a series of predictors some interval, but most categorical. There are thousands of cases (about 400 of which are missing). I got the t values and the f...
  13. noetsi

    test for linearity in linear regression

    I have decided, after rereading Fox to use the partial regression plots. He, and others recommend it, I just have concerns about my judgement which is why I look for formal test.
  14. noetsi

    test for linearity in linear regression

    If I knew the general linear model I might do that although I don't know any formal tests for non-linearity in that.
  15. noetsi

    Other remedial measures for multicollinearity?

    How do you know Multicolinearity is a problem? It is almost never a good idea to get rid of a variable that makes theoretical sense because of mulitcolinearity. John Fox wrote a sage monograph called Regression Diagnostics you may want to look at. One way to deal with it if you think it is a...
  16. noetsi

    test for linearity in linear regression

    The partial slope coefficients are one way to test for this but I wondered if there was a formal test for linearity in linear regression (for a given predictor).
  17. noetsi

    competing risk analysis in SAS

    I lack the expertise to answer this but you might want to look at "Survival Analysis Using SAS" by Paul Alison
  18. noetsi

    Relative impact

    Miner this is why I am reluctant to use beta weights. It comes from "Regression Modeling Strategies" by Harrell He argues that the use of beta weights is questionable when using dummy variables. Or for non-normally distributed variables. We have many of both in our analysis. His suggested...
  19. noetsi

    Droping variables that are not statistically significant

    That makes sense although I thought the goal was to develop a parsimonious model. One thing I am doing, and yes this is strange, is trying to find confounds, that is control variables that should be in a model. We work for a federal agency who is developing a control model - that is a model...
  20. noetsi

    Detrending data

    That is what I thought to. But it appears that is no longer best practice. The link I posted above goes into that in some detail.