I have a sample dateset consists two variables as "Score"(independent continuous) consists score of students in maths and school(dependent categorical)having 3 categories as "S1","S2","S3". I'd perform ANOVA test so null hypothesis would be that means of all samples are equal(as I've studied). The means in above mentioned samples comes as:

S1=80.5; S2=28.5; S3=60.2.

As, means for all samples are not equal so null hypothesis got rejected. Now, my question is:-

a) what does it mean by rejecting null hypothesis? what would be the conclusion?

b) Does rejecting a null hypothesis also decides that the independent variable is not a important feature for modelling and need to be removed?

  • $\begingroup$ see stats.stackexchange.com/questions/163957/… $\endgroup$ – user83346 Aug 22 '17 at 19:32
  • $\begingroup$ @fcop points to an exact duplicate of question (a); visit that thread for the answers. Question (b) may be slightly different, though, because it is tantamount to asking about the role of using null hypothesis tests (that is, p-values) in model building. $\endgroup$ – whuber Aug 22 '17 at 21:35
  • $\begingroup$ @fcop do we need to perform the statistics test in between predictors or tests in between dependent variable(target) and independents variables is sufficient? $\endgroup$ – Bits Aug 23 '17 at 2:00

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.