iSixSigma

statistic

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
    Posts
  • #40728

    McNabb
    Participant

    I would be most grateful if you could help me with the following question.
     
    I am using SPSS (version 13) to do my research. I have 3oo companies(each company has 15 financial ratio). I would like to check the normality assumption for my data. SO, I would like to check the normality for each of the financial ratio variable for 300 companies. I don’t know how to check the normality for my data. Do I just simply plot the graph and examine the Kolmogorov-Smirnov value? Some of the books said that I have to check the normality of the residual for my variables(expected value vs actual value)?I don’t know what is this?how do I do it in SPSS?? What is the difference between simply checking the normality for my variable and checking the normality of residuals?
     
    Furthermore,if my data has kurtosis problem, I don’t know how to transform my data. The usual data transofmration (e.g. log, ln, 1/x, etc..) doesn’t work for kurtosis problem. Do u know how to solve it out in SPSS?
     
    Actually, I would like to use the K means cluster analysis to classify my 300 samples into different group. Some notes said that I need to check for normality, linearity , homoscedasticity multicollinearity and outliers for my data. However, some notes said that I need to check for multicolinearity and outliers only.
     
    I feel confused about the data assumption for cluster analysis? Do I need to check for all data assumptions before using cluster analysis? If I didn’t check for normality, linearity and homoscedasticity for my data, I can’t use pearson correlation to check for multicollinearity (as each of the parametric test has specific data assumption). If I skip some “checking data assumption process”, I may have trouble for using praametric tests in the future.
     
    Although I would like to for check all the data assumptions( normality, linearlity, homoscedasticity,multicollinearity–pearson correlation and outliers) in order to use parametric test, I have problem about the first step of checking data– transforming data that have kurtosis problem in SPSS.
     
    Thank you very indeed for all your great help and useful suggestion.

    0
    #127082

    HF Chris
    Participant

    Mary,
    Keep in mind that a cluster analysis does not determine significance it just groups the data into a specified number of groups and ranks them. As far as using SPSS look into the tutorial library and run and SPSS search on goggle. I have found many walk through examples. As far as normality you may want to run a Levine’s test. You need to check for heterogeneity and heterogeneity which looks at pooled and individual differences. Most of your descriptives can be obtained in SPSS by clicking options in the variable enter box.
    Chris
     

    0
Viewing 2 posts - 1 through 2 (of 2 total)

The forum ‘General’ is closed to new topics and replies.