Significance of lod scores (2)

Here is a list of lod scores derived from different tests. Each can be converted to a corresponding chi-squared statistic using the formula T=2ln(10)Z and the table shows the number of associated degrees of freedom and whether the test is one- or two-sided. (If the test is one-sided then the p value associated with the chi-squared statistic should be halved.) In fact, for theoretical reasons most of the tests which are said to have two degrees of freedom will be somewhat conservative, as the free parameters are confounded with each other. The asymptotic p values work well for large samples but may be inaccurate in certain situations.

HLOD - maximised over theta and alpha, 2 df, one-sided

MLOD - maximised over f1, 1 df, two-sided

MALOD - maximised over f1 and alpha, 2 df, one-sided

MFLOD - maximised over alpha, 1 df, one-sided

MLS (sib pairs) - maximised over z1 and z2, 2 df (fewer if possible triangle constraint used), one-sided

MLS with z1 fixed to 0.5 - maximised over z2, 1 df, one-sided

lodsig.html, lodsig3.html