ANOVA POST HOC Testing - PowerPoint PPT Presentation

1 / 5
About This Presentation
Title:

ANOVA POST HOC Testing

Description:

STAT 3130 Statistical Methods I Session 3 ANOVA POST HOC Testing ANOVA Post Hoc Testing From the previous notes, we found an F-statistic of 24.33. – PowerPoint PPT presentation

Number of Views:96
Avg rating:3.0/5.0
Slides: 6
Provided by: Mauric110
Category:
Tags: anova | hoc | post | fisher | testing

less

Transcript and Presenter's Notes

Title: ANOVA POST HOC Testing


1
STAT 3130Statistical Methods I
  • Session 3
  • ANOVA POST HOC Testing

2
ANOVA Post Hoc Testing
  • From the previous notes, we found an F-statistic
    of 24.33. This equated to a very low p-value
    (lt.001), allowing us to reject the Null
    Hypothesis and conclude that a difference amongst
    the means was present
  • This does not really guide our decision making
    although we concluded that at least one of the
    plans is different, at this point we have no
    information regarding WHICH plan(s) is (are)
    different.
  • To determine this, we need to execute a Post Hoc
    test.

3
ANOVA Post Hoc Testing
  • Post Hoc tests are a critical step in the ANOVA
    process because they allow us to determine WHICH
    plans are different from the others.
  • There are several possible Post Hoc tests to
    choose from based upon the specifics of your
    analysis. We will look at the two most common
    Fishers LSD and Tukey.
  • ltOther post hoc tests which you may hear about
    include Duncan, Scheffe, Bonferroni, Dunnett,
    Sidak, Waller, Welchgt

4
ANOVA Post Hoc Testing
  • Both Fishers LSD test and Tukeys test are
    commonly accepted tests amongst statisticians.
    You really cannot go too far wrong with either.
    Here are a few general points of comparison
  • Fishers LSD (least significant difference) is
    less conservative than Tukeys HSD (honestly
    significant difference).
  • Fishers LSD executes a series of ttests amongst
    the pairs, resulting in a higher Type 1 error
    rate, while Tukeys HSD specifies a family error
    rate across all comparisons
  • Tukeys HSD calculates Tukeys Yardstick any
    mean differences which are at least as large as
    the Yardstick are considered to be different.

5
ANOVA Post Hoc Testing
  • Additional sources for detailed information
    regarding Post Hoc tests
  • http//support.sas.com/documentation/cdl/en/statug
    /63347/HTML/default/viewer.htmstatug_glm_a0000000
    867.htm
  • http//www.uwsp.edu/psych/cw/statmanual/posthocs.h
    tml
  • http//pages.uoregon.edu/stevensj/posthoc.pdf
  • http//practicalstats.labanca.net/index.php?title
    Selecting_a_Post_Hoc_test
Write a Comment
User Comments (0)
About PowerShow.com