One way to analyse multiallelic sibling case-control results is to look at each allele in turn to see if the number of times it is seen in cases but not controls occurs more often than the number of times it occurs in controls but not cases:
Paired results for individual alleles:
1 2 3 4 5
AAvXX 1 1 0 1 0
AvX 6 7 5 7 9
XvA 11 6 1 5 7
XXvAA 2 1 0 1 1
p = 0.1937 0.8273 0.6547 0.5000
The entries for allele 1 mean that in 6 pairs the case had one more copy of allele 1 than the control, in 11 pairs it was the other way round, in 1 pair the case was homozygous for allele 1 and the control had no copy of allele 1, and in two pairs the control was homozygous for allele 1 and the case had no copy.
One way to analyse these results is to test whether the AAvXX+AvX totals are about equal to the XXvAA+XvA totals for each allele (as one would expect by chance). A chi-squared statistic is calculated according to SUM(O-E)2/E, where the expected number of pairs in which more A alleles are observed in the case than the control is half the total number of pairs informative for that allele. E.g. for allele 1, chi-squared = (7-20)2/20+(13-20)2/20
This chi-squared statistic has one degree of freedom, but since one may be testing multiple alleles in turn, the p value must be corrected.