4
$\begingroup$

I have carried out a number of Pearson Chi Square tests on the relationship between tail lesions and abscessation in pigs. I have found a significant (p = 0.005) but weak (Cramers V = 0.288) association. How do I interpret this?

For example, does this mean that there is a clear relationship between the two but that it is just on a small scale?

$\endgroup$
  • $\begingroup$ If it's a Pearson correlation, why do you cite Cramér's [NB] $V$? Alternatively, if it's not a Pearson correlation, please edit accordingly. $\endgroup$ – Nick Cox Sep 5 '14 at 10:38
  • $\begingroup$ I meant to say Pearson Chi square, I've edited the post. $\endgroup$ – Grace Carroll Sep 5 '14 at 10:41
  • $\begingroup$ Thanks for the edit. For your thesis/paper/report, note accent and apostrophe as at en.wikipedia.org/wiki/Cram%C3%A9r%27s_V Some software is very lax on either or both. However, "correlation" really should now be "association" in your title and question. $\endgroup$ – Nick Cox Sep 5 '14 at 10:47
3
$\begingroup$

More meaningful in this case is the $\text{R}^2$ which explains the proportion of variation in your observations accounted by the association. For example if your $R$ was 0.1 (p= 0.005) due to the large sample size, it means 1% of the variation in tail lesions in pigs is accounted for by abscesses. In a multifactorial situation such associations though informative may not be very meaningful. Again be cautious since correlation does not imply causation.

$\endgroup$
  • 1
    $\begingroup$ There may be some typos in your post, by "...if your $R$ was 0.1... it means 1% of the variation...", did you mean '...if your $R^2$ was 0.1... it means 10% of the variation...'? $\endgroup$ – gung - Reinstate Monica Sep 5 '14 at 15:17
  • $\begingroup$ No, he's correct. R here is the correlation coefficient and R^2 is, as its name implies the square of the correlation coefficient. It's also the share of the variation in one variable that is explained by the other. $\endgroup$ – RoyalTS Sep 5 '14 at 16:56
7
$\begingroup$

You could say it like this:

The association is small, but not zero.

However, I don't know that I'd call a V of 0.288 "small".

Don't confuse "statistically significant" with important. Statistical significance is very different from practical importance.

$\endgroup$
  • $\begingroup$ Thanks, can you explain what the actual meaning of this finding it? Why is it significant but weak? $\endgroup$ – Grace Carroll Sep 5 '14 at 10:43
  • $\begingroup$ It's not big, but the sample is large enough to make it clear it's different from zero. $\endgroup$ – Glen_b -Reinstate Monica Sep 5 '14 at 10:58
3
$\begingroup$

Adding on to @Glen_b 's excellent answer and your comments there.

As he said, it is significant but weak because the sample size is large enough to make a small effect significant. But now you need a way of showing the effect size. V is one such way, but it isn't intuitive to many and is not as well known as some other measures.

From your question, it seems like you have a 2x2 table of lesions (yes/no) and abscess (yes/no). You could then show that table; you could give the odds ratio or any of a number of other measures for 2x2 tables.

$\endgroup$
  • $\begingroup$ n-thing the Odds ratio. I know nothing about pig's tails, but the most informative question here seems to be: how much more likely is a pig to have an abscessation, given it has a tail lesion? (Include a CI, too.) $\endgroup$ – jona Sep 5 '14 at 13:00
1
$\begingroup$

I like to comment on the perceived usefulness of the relationship. For instance:

The association is statistically significant but not practically relevant.

That being said, your association doesn't seem to be that weak.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.