I have a dichotomous dependent variable as 0 and 1 and a continuous covariate. I went for a binary logistic regression and it came up with a positive B but negative constant. So is it a problem? How can i interpret this B efficiently and accurately? Thanks a lot

.enter image description here

  • $\begingroup$ What is the range of the continuous variable, what values does it take in the sample? $\endgroup$ – Alecos Papadopoulos Jan 7 '15 at 21:40
  • $\begingroup$ The constant (aka "intercept") is utterly meaningless by itself. How it is interpreted depends on the units of measurement of the covariate (in particular, on what its zero value means). That is why @Alecos is requesting more information about it. BTW, unless you can get your software to display numbers with greater precision, you need to change those units so that you can see more than one significant digit in the coefficient of fixationduration! $\endgroup$ – whuber Jan 7 '15 at 21:42
  • $\begingroup$ @alecos my covariate mean is 470, it varies from 250 to 1000. It is about how many miliseconds a word is fixated by the participants. $\endgroup$ – Emrah Dolgunsöz Jan 7 '15 at 22:20
  • $\begingroup$ @whuber actually i want to know if negative constant is something that damage my output. I am not a statistician thats why i could not figure out completely what you mean, i am learning :) $\endgroup$ – Emrah Dolgunsöz Jan 7 '15 at 22:23

Browse other questions tagged or ask your own question.