Re: Minimum number of cases for a correlation?

Glucoboost wrote:
I am reviewing a piece of research in which a correlation of .6 has
been found between two variables. The scatterplot is okay. However, the
N is 12. I remember reading somewhere that an N of 50 was the minimum
required for a correlation.

Is an N of 12 just too small for a correlation?

Thanks


As one of the other respondents suggested, a correlation can be used both descriptively and inferentially. I would be more concerned about sample for the latter use. Rules of thumb about how many observations are needed are usually stated in terms of regression though. Here is a comment from Dave Howell's book (Statistical Methods for Psychology) on those rules of thumb.

www.angelfire.com/wv/bwhomedir/notes/linreg_rule_of_thumb.txt

One of the suggested rules (Harris, 1985) is that N should exceed p by at least 50 (where p = the number of predictors). Perhaps this is what you were thinking of.

--
Bruce Weaver
bweaver@xxxxxxxxxxxx
www.angelfire.com/wv/bwhomedir
.

Relevant Pages

  • Re: Minimum number of cases for a correlation?
    ... The scatterplot is okay. ... I remember reading somewhere that an N of 50 was the minimum ... Is an N of 12 just too small for a correlation? ... If the scatterplot is presented as well, then I wouldn't worry: people can judge for themselves what's going on. ...
    (sci.stat.consult)
  • Re: 2-year-olds reading?
    ... speech is "highly correlated" with reading, ... English is not considered a phonetic language -- one of the reasons ... level correlation between written words and spoken words. ... Organising and interpreting visual input is, as I said, quite distinct from ...
    (misc.kids)
  • Re: 2-year-olds reading?
    ... at *any* age is extending speech ability to the written word. ... There is the lower-level correlation of letters (or ... There would appear to be more connections between sight and reading, ...
    (misc.kids)
  • Re: OT- Least Shocking News of the Day
    ... not a suggestion of a relationship. ... And, as you know, "relationship" is correlation, not necessarily cause ... Jeffrey Dahmer's home made the argument that reading pornography ... If he couldn't pick a fight with somebody, he wouldn't be able to say ...
    (rec.gambling.poker)
  • Re: Contributor to the Flynn effect
    ... and lead) has been used to account for race differences in IQ. ... I remember reading that children in old houses were in danger of eating flakes of lead based paint, but I hadn't been aware that this had been used in an racial arguments! ... I know that correlation isn't causation and this is only supposition, not even correlation, but I wonder if anybody has investigated it. ...
    (uk.philosophy.humanism)
Quantcast