Ben

I found that the `model.frame.svyglm()` I wrote while fixing this issue does less than I had hoped, not reliably creating a model frame to which `model.response()` can be applied. See...

1. Re `prepareMatching()`, I set out in this branch to replace it with `edgelist()`. Pretty sure its mention in the docs is a vestige of docs I must have created...

NB: We (I) should maybe attend to #161 before we merge the proj1-nodeprices branch into master, or have it supercede the master branch. But merging (what we want from) the...

@josherrickson, I suspect this issue can be closed -- whatever outstanding problems there were would have been addressed during the hasty merge we made to restore optmatch to cran. Would...

What a thicket! Two thoughts: 1. I'd also be happy with functionality to make `fill.NAs` warn or stop if the user tries to give it a 2-column response, coupled with...

This approach looks very very promising. Perhaps it will help us knock out a number of the `fill.NAs` related issues in our issue queue. In terms of coherence w/ the...

Speaking of gaps in testing, in optmatch 0.9-5 the below fails, as discussed in comments to #103: ``` r > glm(fill.NAs(pr~cap+factor(ne), data=nuclearplants), family=binomial) Error in model.matrix.default(mt, mf, contrasts) : model...

Another something to include in testing suite: check that `sandwich:::estfun.glm` generates estimating functions corresponding to generated .NA columns as well as columns present in original data frame, and in general...

Checking in with this after a little hiatus. @markmfredrickson would it seem like a little thing or a big thing to take up this thread from here? (Perhaps the answer...

@vkartha and @renyc432, I have the two of you in mind with this issue. I don't see starting on this soon myself, but we (package developers) would be happy to...