Significance in Correlation table but not coefficients (Multiple Regression)

Home Forums Methodspace discussion Significance in Correlation table but not coefficients (Multiple Regression)

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
    Posts
  • #1665

    Hi All, 

    Just wondered if anyone would be able to provide me with a bit of help. I have conducted a Multiple Regression using the Enter method and this includes six predictor variables (one of which is dummy coded) against the criterion.

    Anyway, in my output from SPSS in the Correlations table, one of my predictors appears to be significant at the 5% level, but with a -19% correlation with the criterion. In my coefficients table that I am reporting in my results section this predictor doesn’t appear to be significant.

    Can anybody give me an idea as to why this might have happened? If it is not significant in my coefficients table but it is in my correlations table, am I still able to report this in the results section of my research?

    Thank you so much in advance and because I am new to this site, I don’t know how I will receive replies so if you’d like feel free to email me at: K0052613@live.tees.ac.uk

    #1666

    As I understand it, you run a regression with six predictor and additionally report the bivairtae correlations between each predictor and the criterion. If this is true, it would not be surprising that a covariate with a significant bivariate corelation turns out non-significant in the multivariate analysis. Predictors are likely to be correlated and controlling for other covariates might introduce collinearity. Adding a predictor B might also screen-off the effect of another predictor A, which happens when B is a cause of A and the criterion. There is nothing that speaks against reporting the bivariate and multivariate results.

Viewing 2 posts - 1 through 2 (of 2 total)
  • You must be logged in to reply to this topic.