Changes in version 0.2.0                        

  - Added an option to choose the model selection criterion. Previously
    the criterion was fixed; users can now select among the available
    criteria (e.g. Cochran's Q test for heterogeneity or the MSE
    criterion).
  - Empirical Bayes shrinkage is now applied to standard errors of
    cohort-time estimates, not just to the point estimates.
  - Cohort-time confidence intervals shown by plot() are now widened
    using a Bonferroni correction for the number of cohort-time cells
    displayed, controlling the familywise error rate at 0.05.
  - plot() now draws a reference line at the ATT, along with an
    explanatory note in the caption.

                        Changes in version 0.1.0                        

  - Initial CRAN submission.