EHypMidP {CooccurrenceAffinity} | R Documentation |
This function does the analogous calculation to that of EHypQuInt, but with the Extended Hypergeometric distribution function F(x) = F(x,mA,mB,N, exp(alpha)) replaced by (F(x) + F(x-1))/2.
EHypMidP(x, marg, lev)
x |
integer co-occurrence count that should properly fall within the closed interval [max(0,mA+mB-N), min(mA,mB)] |
marg |
a 3-entry integer vector (mA,mB,N) consisting of the first row and column totals and the table total for a 2x2 contingency table |
lev |
a confidence level, generally somewhere from 0.8 to 0.95 (default 0.95) |
This function does the analogous calculation to that of CI.CP, but with the Extended Hypergeometric distribution function F(z, alpha) = F(z,mA,mB,N, exp(alpha)) replaced by (F(z,alpha) + F(z-1,alpha))/2.
This function returns the interval of alpha values with endpoints (F(x,alpha)+F(x-1,alpha))/2 = (1+lev)/2 and (F(x,alpha)+F(x+1,alpha))/2 = (1-lev)/2.
The idea of calculating a Confidence Interval this way is analogous to the midP CI used for unknown binomial proportions (Agresti 2013, p.605).
Eric Slud
Agresti, A. (2013) Categorical Data Analysis, 3rd edition, Wiley.
EHypMidP(30,c(50,80,120), 0.9)
AlphInts(30,c(50,80,120), lev=0.9)$CI.midP
EHypMidP(20, c(204,269,2016), 0.9)