1

I have a dependent variable $Y \in \{0, 1\}$ and 4 independent variables $X_1$ through $X_4$. First, I run a total of 5 logistic regressions:

  1. $Y \sim X_1$
  2. $Y \sim X_2$
  3. $Y \sim X_3$
  4. $Y \sim X_4$
  5. $Y \sim X_1, X_2, X_3, X_4$

I find that the coefficient of $X_1$ is significant in (1), but is not in (5). I also find that the coefficient of $X_2$ is not significant in (2), but is now significant in (5).

There are no interaction in the model. I don’t have many missing values in my variables. Hosmer-Lemeshow statistic gives p-value > 0.05.

Why does this happen? What should I pay attention to?

Thanks.

Art
  • 428
  • 2
  • 7
Eva
  • 11
  • 1
  • 3
    This happens for the same reasons it does in ordinary multiple regression (many answers and comments on posts on site discuss this issue), and illustrates the futility of using univariate regressions to decide what will be important in a multivariate regression (whether ordinary multiple regression, logistic regression, or some other). – Glen_b Aug 31 '19 at 04:01
  • So I explain the results of multivariable logistic regression model and also mention what happened in a univariate logistic regression;; Or I don't need to mention it; – Eva Sep 01 '19 at 15:26
  • Hosmer-Lemeshow is considered obsolete: https://stats.stackexchange.com/questions/273966/logistic-regression-with-poor-goodness-of-fit-hosmer-lemeshow – kjetil b halvorsen May 14 '20 at 12:12

0 Answers0