Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: Validity of correlation coefficient for binary data
Posted:
Oct 24, 1996 11:31 AM


Hi Ronan, Drake, and others who may be interested in STATISTICAL HISTORY or those who have to take a STATISTICAL HISTORY TEST!
Below is a summary of some of the names given to a few indicators of relationships.
 Joe *********************************************************************** * Joe Ward 167 East Arrowhead Dr. * * Health Careers High School San Antonio, TX 782282402 * * Phone: 2104336575 * * joeward@tenet.edu * ***********************************************************************
SUMMARY OF SOME "INDICATORS OF RELATIONSHIP" FROM CHAPTER 9 OF GLASS AND STANLEY "STATISTICAL METHODS IN EDUCATION AND PSYCHOLOGY" (PRENTICEHALL, 1970)
 MEASUREMENT OF VARIABLE X _____________________________________________________________________________    MEASUREMENT OF   Dichotomous, with  Interval VARIABLE Y  Dichotomous  underlying normal  or Ratio ___________________________________________________________________________    Dichotomous  phi   ___________________________________________________________________________    Dichotomous, with   tetrachoric  underlying normal   r(tet)  ___________________________________________________________________________    Interval or  point  biserial  Pearson Ratio  biserial   productmoment  r(pb)  r(bis)  r(xy) ___________________________________________________________________________
On Thu, 24 Oct 1996, Ronan M Conroy wrote:
> Date: Thu, 24 Oct 1996 05:01:57 0400 > From: Ronan M Conroy <rconroy@rcsi.ie> > To: Multiple recipients of list <edstatl@jse.stat.ncsu.edu> > Subject: Re: Validity of correlation coefficient for binary data > > At 3:07 pm 23/10/96, Drake R. Bradley wrote: > >Actually, I think the correlation between two binary variables would be > >a phi coefficient. The correlation between a binary (dichotomous) variable > >and a "true" quantitative variavle (nonbinary) is a point biserial > >correlation. The formulae for rphi and rpb are just simplifications of > >the standard Pearson productmoment r. > > > > The phi coefficient is tested using the chisquare test (2x2),[snip] > > in fact, phi squared is chi squared divided by N. Although I have known > about phi since my youth, I have never once fired it in anger, or even > pointed it at anyone in fun. Does anyone out there use it? > > _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ > > _/_/_/ _/_/ _/_/_/ _/ Ronan M Conroy > _/ _/ _/ _/ _/ _/ Lecturer in Biostatistics > _/_/_/ _/ _/_/_/ _/ Royal College of Surgeons > _/ _/ _/ _/ _/ Dublin 2, Ireland > _/ _/ _/_/ _/_/_/ _/ +353 1 402 2431 fax 402 2329 > > _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ > Why does noone whistle anymore? > > >



