I have got the following data: I have the amount of walking of the person in min as the dependent variable (continuous) and I have got a mix of categorical and continuous variables as independent variables (however mostly categorical variables) (e.g. age (continuous),gender, scorehealth, educationcategorie, incomecategorie, level of the work, rating giving in terms of safeness and pleasantness of the neighborhood, having or not having a dog and the distance to green (continuous) . I was wondering what type of statistal test I should apply to my case

My purpose is to use the output of the statistical test in my simulationmodel to simulate walking behavior, so if someone for example is female with certain characteristics, she will walk for ..... minutes, something like that.

  • 2
    $\begingroup$ You can't apply a statistical test to a research in general, however you can apply it to a question. I suggest you that instead of asking about techniques (like statistical test) you better describe which are the questions you want to ask. For some of them, maybe, you could answer with a statistical test, but for others maybe not. $\endgroup$ – rapaio Jul 17 '18 at 8:10
  • $\begingroup$ stats.meta.stackexchange.com/q/5338/1352 $\endgroup$ – S. Kolassa - Reinstate Monica Jul 17 '18 at 8:12
  • $\begingroup$ my main research question is "What is the effect of green on the walking behavior in elderly?" however I for now I would like to see from the varialbes the importance or the weight they contribute to walking in min. So I would like to know the importance of these variables and use these weights afterwards for my simulation model. Maybe I should have elaborated that in my text above. $\endgroup$ – Joy Jul 17 '18 at 8:19
  • 2
    $\begingroup$ I guess that "green" here means distance to the nearest green space. The general territory here is regression, but you need to watch out. Are there zeros in the outcome or response (a much better term than dependent variable), so that some people didn't do any walking? My first recommendation is Poisson regression here and don't be put off by a myth that the response must be discrete. See e.g. blog.stata.com/2011/08/22/… More generally, a generalized linear model sounds a good idea. $\endgroup$ – Nick Cox Jul 17 '18 at 8:41
  • $\begingroup$ @NickCox Wow thanks for the information. Indeed with green, I meant the distance to green and indeed as you have guessed there were people who didn't walk at all. I never heard of the Poisson distribution, but I will look it up and come back later. $\endgroup$ – Joy Jul 17 '18 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.