Re: anova and chi-square test

Bruce Weaver <bweaver@xxxxxxxxxxxx> wrote in news:1186512446.379964.168920
@d55g2000hsg.googlegroups.com:

On Aug 6, 9:37 pm, Mike <Sulfate...@xxxxxxxxx> wrote:
On Aug 6, 8:30 pm, Tim De Meyer <Tim.DeMe...@xxxxxxxx> wrote:

Hi

I usually confuse anova and chi-square test.
Is there any simple way to tell from?

Mike

Oversimplified (I know!):
anova: variable of interest (dependent) is continuous
chi-square: variable of interest (dependent) is factor

So if the dependent variable is discontinuous, then use chi-square?


The chi-square test of association is used to test the null hypothesis
that there is no association between two nominal scale (categorical)
variables. It requires no distinction between independent and
dependent variables.

No debate on your statement as far as it goes. It is unfortunate, however,
that the tests using the chi-square distribution as their reference have
acquired the label "chi-square" without qualification and without
understanding of the severe limitations of the "test of association". There
are other forms of "chi-square" tests that are be more valuable
(statistically powerful) in specific situations. The chi-square tests of
ordered relationships ("trend tests") are often overlooked in situations
where they would be perfectly valid and far more powerful than the test of
general association.

--
David Winsemius
.

Relevant Pages

  • Re: anova and chi-square test
    ... anova: variable of interest is continuous ... So if the dependent variable is discontinuous, then use chi-square? ... dependent variables. ... I suggest you find one of the many websites that list guidelines on ...
    (sci.stat.math)
  • Re: ANOVA or Chi Square?
    ... Neither non-random selection nor small sample size (which isn't that ... small) suggest chi-square is superior to ANOVA. ...
    (sci.stat.consult)
Quantcast