This functions evaluates many qpadm models simultaneously by keeping the target population
and the rightfix populations fixed, and distributing the leftright populations by keeping some
in the set of left population and adding the remaining populations to the right populations.
(See details for an example of how models are generated)
qpadm_rotate(
f2_blocks,
leftright,
target,
rightfix = NULL,
full_results = FALSE,
verbose = TRUE
)3d array of blocked f2 statistics, output of f2_from_precomp.
Populations which will be distributed between left and right
Target population
Populations which will be on the right side in all models
Return all output items which are returned by qpadm. By default (full_results = FALSE), weights and several other statistics will not be computed for each model, making it faster and the output more readable. If full_results = TRUE, the output will be a nested data frame where each row is one qpadm model, and each column has one data frame item from the regular qpadm output (weights, f4, rankdrop, popdrop).
Print progress updates
A data frame with Chi-squared statistics and p-values for each population combination
If leftright consists of the populations L1, L2, L3, L4; rightfix is the population R; and target is T,
the following models will be genrated:
(left), (right), (target)
(L1), (L2, L3, L4, R), (T)
(L2), (L1, L3, L4, R), (T)
(L3), (L1, L2, L4, R), (T)
(L4), (L1, L2, L3, R), (T)
(L1, L2), (L3, L4, R), (T)
(L1, L3), (L2, L4, R), (T)
(L1, L4), (L2, L3, R), (T)
(L2, L3), (L1, L4, R), (T)
(L2, L4), (L1, L3, R), (T)
(L3, L4), (L1, L2, R), (T)
if (FALSE) {
pops = dimnames(example_f2_blocks)[[1]]
qpadm_rotate(example_f2_blocks, leftright = pops[1:4],
target = pops[5], rightfix = pops[6:7])
}