0

This is a binary model that my colleagues and I developed that regards a connection between the distance (dependent variable) and overall satisfaction with public transport (covariate). I don't know how to interpret the results shown below.

R glm output as an image

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Katarina
  • 55
  • 7
  • Hello. So, we are trying to research how different aspects of the public transport system impact the overall satisfaction with the former. In this particular case (that you can see in the screenshot), we did a logit model, where we tried to see the correlation between distance of homes of the people surveyed and overall satisfaction with the PTS. We are not sure how to interpret the results, meaning the coefficients and residuals. Hopefully, this makes it a bit more clear. Thank you! – Katarina Dec 29 '18 at 10:41
  • 2
    Welcome to the cite Katarina. I find your question unclear because (1) you claim to do a binary model - does that mean your dependent variable is binary? If so how is that consistent with the dependent variable being distance? It appears from the output - although I am guessing since output is incomplete - that you are running a glm model with binomial as link, in which case the dependent should be binary. I suppose distance is continuous and hence again I am confused. Please edit the question. You can include R output simply by using code tags. – Jesper for President Dec 29 '18 at 15:05
  • Hello Jesper. Thank you for your reply! We have translated all variables to be binary, since we are working with qualitative data. Perhaps there are mistakes, since as I said, we are very confused, and maybe it cannot even be done with distance as a dependent variable. Yes, we did try to run a glm model with binomial as a link. I have other codes, but I do not know how to put it here as another screenshot. Perhaps you can give me your e-mail, and we can communicate directly? Once again, thank you for your response and help! – Katarina Dec 29 '18 at 15:26
  • "We have translated all variables to be binary, since we are working with qualitative data". I find it hard to believe that "distance" is qualitative, what scale is the variable measured on? Are you sure you transformed this to be binary? – Jesper for President Dec 29 '18 at 15:41
  • We divided it into levels (since distance was in classes of km) and then transformed it into binary. There were no errors and the results that we got are in the screenshot. Do you believe that to be a mistake? – Katarina Dec 29 '18 at 15:43
  • 2
    Hello Katarina - ideally you would provide a [MWE](https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) when posting a model like this. It's also a little unclear exactly what you want someone to answer here, would they go through every part of the output and explain that? It seems like it would be a lot of work, perhaps there's something more specific you're thinking of which hasn't come across with the way the post is written. – baxx Dec 29 '18 at 15:48
  • I strikes me as a very strange approach. You have data with an ordinal variable "overall satisfaction" and "distance" which it is probably safe to treat as a quantitative variable (depending on the nature of the sample you may have a lot of people with 0 distance, which may require special attention). My first choice of model for analyzing the relation between two such variables is certainly not a binary logit model. I would reconsider the model choice. This is as close as I get to saying what you are doing is a mistake, without actually saying it. – Jesper for President Dec 29 '18 at 15:59
  • Hello. Basically what I am asking is if somebody can help me and interpret the results AKA the coefficients? I hope it is clear now. Thank you for your answer! – Katarina Dec 29 '18 at 16:06
  • Perhpas it is. As I said, we are very confused and lost at this point. :D Perhaps you can help me with another model? I will share the link here: https://stats.stackexchange.com/questions/384878/how-to-interpret-results-of-a-binary-model-in-r?noredirect=1#comment722630_384878 – Katarina Dec 29 '18 at 16:07
  • Perhaps you could update your post with something along the lines of *how should I interpret the coefficient in the output of this Logistic Regression* to make it clearer for others (information in the comments can be easily missed). Also try to keep in mind that people try to ensure posts on stackexchange have a fairly long life to them, and are able to help other users in future (which is why MWE's are encouraged). It might be helpful to copy paste the text from your model into your post as well ( format it as code so that the formatting doesn't get messed up). – baxx Dec 29 '18 at 16:15
  • Is the independent variable also binary? – Jesper for President Dec 29 '18 at 16:15
  • Hi Jasper. Yes, everything in this case is binary. Could you try and interpret the coefficients that were given? Just to see if anything has been done right. Thank you! – Katarina Dec 29 '18 at 16:30

1 Answers1

-1

The question lacks a clear description of the problem. It seems that distance is the dependent variable and it is transformed into binary. Logistic regression with satisfaction (a categorical variable) as the independent variable is run here.

Anyway, the way to interpret the results obtained from a logistic regression model by R is well described in this ucla page. Interpretations are available for both categorical and continuous covariates.

Hasan Mahmud Sujan
  • 1
  • 3