Changes in version 1.1.0 (2025-09-17)                  

  - Added functions repeat_power_marginaleffect() and
    repeat_power_linear() along with plotting methods for the resulting
    objects to easily create data and plot power curves from a range of
    sample sizes and models (#72).
  - Fixed print method for rctglm() and rctglm_with_prognosticscore()
    correctly say "active" for the counterfactual mean estimate for
    group A=1 (#67).
  - Added a predict method for objects of class rctglm.
  - Added attributes with information about assumptions for the power
    calculation to output of power_xx functions.

                 Changes in version 1.0.1 (2025-07-01)                  

Updated ALL power_xx functions to use 1-alpha/2 as the quantile of the
critical value when testing (changed from 1-alpha) (#65).

                 Changes in version 1.0.0 (2025-04-08)                  

Major overhaul of package. Still focuses on analysing data with the use
of prognostic scores, but takes a more general approach that allows any
distribution of response and covariates within the scope of generalised
linear models (GLMs) and does not necessarily run on a number of data
sets created by simulation.

The package provides novel methods for:

  - rctglm: Finding any marginal effect estimand and estimating the
    standard error using influence functions to avoid inflation of
    type 1 error
  - rctglm_with_prognosticscore: Do the above, but leveraging historical
    data to increase precision with prognostic scores.

Additionally, the package includes functionalities for

  - fitting a discrete super learner in fit_best_learner, which is
    leveraged in rctglm_with_prognosticscore
  - approximating power using
      - standard methods for ANCOVA models (see help topic power_linear)
      - a novel method for any model estimating marginal effects
        (power_marginaleffect)
  - generating data from a GLM (glm_data)

                        Changes in version 0.2.1                        

Features

  - Added function simulate_collection that takes function arguments for
    how to simulate covariates and model the outcome in the historical
    and "current" data to give the user full flexibility (previously a
    multivariate normal distribution was assumed)
    
      - sim.lm which simulates data from a multivariate normal
        distribution and models the outcome with a linear model is now a
        wrapper of the new - more general - simulate_collection.

                        Changes in version 0.2.0                        

Features

  - Added option to use sandwich HC estimators for the covariance matrix
    in sim.lm

  - Updated default value of ATE_shift in sim.lm

Code cleanup

  - Modularised code. Fx. split lm.hist into lm.procova and lm.psm

  - Renamed some functionalities

  - Correcting errors in documentation

Best practices package

  - Updated DESCRIPTION

  - Created README

  - Added explicit package imports in form of foo::xx

  - Added a few tests

                        Changes in version 0.1.0                        

Initial package created from local files. Package contains
functionalities to create simulation study for a specific purpose
related to an article. Functionalities include generation of a
collection of data sets and a way to analyse these data sets assuming a
special case of multivariate normal distribution of covariates with a
linear model of the response. In addition, functionalities to estimate
the power of certain parameter tests based on the results.