0

My model has seven independent variables and one dependent variable. Each are subscales for a 5 point likert scale questionairre. I tested each variable data and they are all non normal. I did spearman correlation and identified several significant correlations. I tried testing the model using linear regression but a very small R square. I am considered a beginner in statistics hence i dont know what other technique i need to use in order to test my model. I have small sample size of 247.Hope anyone can help me.

Thanks

Ali
  • 1
  • 1
  • 2
    Data cannot be parametric or nonparametric; **models** are parametric or nonparametric. How did you "test each variable data" (I am not sure what that even means). Also, you can't use other techniques to test the overall model, you come up with a model and then test it. – Peter Flom Mar 01 '13 at 12:51
  • sorry i meant non normal, tested it through KS and SW. I have the model but how can i test it. The idependent variables are dimensions of a firm's reputation and the dependent is brand preference, and I am trying to prove that these 7 together impact the dependent variable.The dependent and independent variables are non normal. – Ali Mar 01 '13 at 13:02
  • 4
    Neither the X nor the Y variables need to be normally distributed (see here:[what-if-residuals-are-normally-distributed-but-y-is-not](http://stats.stackexchange.com/questions/12262/)). If your response variable is *ordinal* you should probably look into *ordinal logistic regression*; there are standard tests for this. – gung - Reinstate Monica Mar 01 '13 at 14:48
  • +1. Is your dependent variable also ordinal? Their are routines in R (for example polr in {MASS}) that already implement such procedures. – usεr11852 Mar 01 '13 at 16:03
  • All the variables are continuous ordinal, and non normal. I didnt get the part for the routines,what do u mean? When i tried linear regression Rsq was 0.09. Thanks alot for your help – Ali Mar 02 '13 at 11:19
  • Variables are continuous, which means that they can (theoretically) take on any value - like height, or weight; or they're ordinal, which means that they take on one of a series of values which are in order, like strongly agree, agree, don't know, disagree, strongly disagree (scored 1, 2, 3, 4, 5). A variable can't be continuous ordinal. – Jeremy Miles Apr 02 '13 at 23:26

0 Answers0