iSixSigma

ANOVA appropriate here

Six Sigma – iSixSigma Forums General Forums Methodology ANOVA appropriate here

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #53398

    Rider
    Participant

    Sirs,

    I have in front of me a stats book, but I am uncertain whether I’m using the correct tool. I have attached an Excel file with my data. I am attempting to ascertain whether there is a significant difference in turnover based on average number of employees. I have my 5 sites listed in Groups and then the average number and terminations in. Running the ANOVA in Excel, I get F=3.175; F crit=5.31. I know to fail to reject the null in this case. I’m wondering if I should use Single Factor ANOVA, or Two Factor in Excel, or some other tool all together. I keep confusing myself so any help would be appreciated. I’ve done a Chi-Square too, but I may be off there as well.

    I will gladly accept any ridicule you may have if only you’ll provide some guidance as I’m flying in the dark.

    Ghost Rider

    0
    #189981

    Robert Butler
    Participant

    If your groups consist of nothing more than the 5 sites and if all you are interested in comparing is the average turnover then a one way ANOVA would be appropriate. ANOVA is robust with respect to non-normality, however, it is sensitive to unequal population variances. If the variances are unequal what will most likely happen is ANOVA will tell you there are no significant differences in the means when in fact they may be different.

    You can test for variance equality using Levene’s test. If the site variances are unequal you could use Welch’s test in place of ANOVA, however, if the variances are unequal it would probably be worthwhile thinking about aspects of the sites that would result in such a difference in the first place.

    0
    #189982

    Rider
    Participant

    Thanks. Your response is much appreciated. I’ve saved your posting as you have said a great deal in just a few words. Thank you again.

    GR

    0
Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic.