Possible additional developments in no particular order:
---------------------------------------------------------

 [done?] options to handle quantiles in rounded continuous data like SUDAAN
   does (if I can work out what it does).

 - Score-based confidence intervals for glms

 [mostly done] Rao-Scott-type tests for glms.

 [done] More general HT estimator for sampling without replacement
 	- using joint inclusion probabilities
	- using population first-order inclusion probabilities and Hartley-Rao approximation
	- using Overton's approximation and sample inclusion probabilities
        - using Berger's approximation and sample inclusion probabilities.

 - GEE (you can do the independence working model just as another level of clustering).

 - an interface to the Auckland semiparametric approach to two-phase designs.

 - Parametric survival models (you can do these with svymle)

 - linear mixed models?
     - simpler case: multilevel model following sampling design
       - nested model exactly matching design
       - design is simpler than model
     - interesting case: separate model and sampling designs.
       - cluster sampling, so that higher-order sampling probabilities are available
       - more general sampling. 

 [done] standard errors for survival curves, particularly in two-phase studies

 - an interface to twophase and calibrate for IPTW estimation.

 - Replicate weights for two-phase designs and for multistage designs
   with large sampling fraction (Fukuoka's BBE?)

 [experimental] parallel processing for replicate weights, svyby(), svyrecvar()?

 - Gini index and other summaries of concentration and inequality.

 - Krista Giles' respondent-driven sampling estimators?

 [done] database-backed designs for replicate weights.

 - Multivariate statistics
   [done] principal components
   [experimental] factor analysis 
    - SEMs? 

