Analysing data from a past experiment I found that size of animals and a level of serum biomarker were associated with time to death when considered separately. Though when combining them in a cox model both the variable lost significance. How should I interpret and report this? Is it correct (and also relevant in scientific terms) to affirm that univariable association is lost when analysing both variables in a multivariable cox model?
Asked
Active
Viewed 27 times
0
-
Is it possible that size and market are collinear? – Yuval Spiegler Jan 18 '17 at 00:00
-
The correlation is 0.65 with a p of 0.004 using the spearman test, could this be the reason? – GGA Jan 18 '17 at 17:44
-
Possible dups: https://stats.stackexchange.com/questions/57423/why-is-anova-interaction-no-longer-significant-after-including-a-covariate, https://stats.stackexchange.com/questions/27257/significant-predictors-become-non-significant-in-multiple-logistic-regression, https://stats.stackexchange.com/questions/269976/is-it-possible-to-lose-significance-for-categorical-variables-when-changing-the, https://stats.stackexchange.com/questions/157475/model-change-after-switching-reference-level-in-r-logistic-regression-model-with – kjetil b halvorsen Sep 30 '19 at 22:55