                        Changes in version 5.1.0                        

Maintenance release. Corrects the seed handling in the cluster
randomized design AIPE sample size planners.

  - The ss.aipe.crd.es.* planners (ss.aipe.crd.es.nclus.fixedwidth,
    ss.aipe.crd.es.nindiv.fixedwidth, ss.aipe.crd.es.nclus.fixedbudget,
    ss.aipe.crd.es.nindiv.fixedbudget, ss.aipe.crd.es.both.fixedbudget,
    and ss.aipe.crd.es.both.fixedwidth) now default seed to NULL instead
    of the fixed value 123321. With the previous default, every call
    silently reused the same seed, which made the a priori Monte Carlo
    results look perfectly reproducible and misled users about the
    sampling variability of the planned sample size. With seed = NULL
    the simulation now draws from the current state of the random number
    generator (honest variability), and the planner restores the
    caller's RNG state on exit so that calling it does not perturb the
    user's random number stream. Supplying an explicit integer seed
    still produces reproducible results exactly as before.

                 Changes in version 5.0.1 (2026-06-04)                  

Maintenance release. Corrects the sampling-variance scaling in the
polynomial change model sample size planner ss.aipe.pcm().

  - ss.aipe.pcm(): the single-subject sampling variance of the change
    coefficient now includes the factor frequency^(2m) (where m is the
    order of the polynomial trend), matching Raudenbush and Liu (2001)
    and the companion power function ss.power.pcm(). Previously this
    factor was omitted, so for designs measured more than once per unit
    of time (frequency greater than 1) the planned sample size was too
    small and the resulting confidence interval was wider than
    requested. Results for frequency = 1 are unchanged, including the
    published Kelley and Rausch (2011) examples, which still return 278
    and 316.
  - ss.aipe.pcm(): corrected the constant for a cubic trend from
    1/1000800 to the correct 1/100800 (that is, (3!)^2 / (6! 7!)). The
    earlier value, a transcription error with an extra digit, inflated
    the implied error variance for trend = "cubic" by a factor of about
    9.93 and so overstated the planned sample size. Linear and quadratic
    trends were unaffected.
  - ss.aipe.pcm(): in the assurance branch, the t critical value is now
    recomputed as the sample size iterates rather than reusing the value
    from the expected-width search. The numerical effect is small, but
    the previous code used a critical value carrying the wrong degrees
    of freedom.

                 Changes in version 5.0.0 (2026-06-01)                  

Maintenance release. MBESS is mature, stable, and widely used, and it
remains fully supported. Maintenance mode does not mean the package is
finished or abandoned: MBESS will continue to be maintained so that it
keeps working with current versions of R and its dependencies, and it
will remain on CRAN. New methodological development now takes place in
the successor package, DMAR (Design, Measurement, and Analysis in R),
which is available on CRAN. MBESS and DMAR are designed to coexist;
existing code that relies on MBESS will keep working, so there is no
need to migrate.

  - Documentation and references brought up to date. Citations that had
    previously appeared as "submitted", "in press", or "under review"
    now cite the published articles (for example, Kelley, 2008; Kelley &
    Lai, 2011; Lai & Kelley, 2012; Chattopadhyay & Kelley, 2016, 2017;
    Lachowicz, Preacher, & Kelley, 2018; Li, McLouth, & Delaney, 2020).
  - Numerous spelling and typographical corrections throughout the help
    pages.
  - ci.reliability(): fixed a misnamed variable that caused the
    lower/upper logit confidence limit swap to be skipped in an edge
    case.
  - Internal calls into Suggested packages were brought up to their
    current, fully namespaced forms. ss.aipe.rmsea.sensitivity() now
    calls sem::sem() directly, so it no longer fails when the sem
    package is installed but not attached. ci.reliability() now uses
    lavaan::lavInspect() in place of the older lavaan::inspect(), with
    identical numerical results.
  - Added an inst/CITATION file, this NEWS file, and an expanded README.
  - Contributor acknowledgments now ship with the package (see the
    CONTRIBUTORS file in the installed package) rather than depending on
    an external web page.

                 Changes in version 4.0.0 (2016-02-17)                  

Major release.

                 Changes in version 3.0.0 (2010-05-01)                  

Major release. Versions 3.0.0 through 3.3.3 included Keke (Brian) Lai as
coauthor and maintainer. Lai, Ken Kelley's doctoral advisee at Indiana
University and the University of Notre Dame, has since stepped away from
package maintenance and development, but his contributions to individual
functions remain and he is credited as their coauthor.

                 Changes in version 2.0.0 (2008-11-23)                  

Major release.

                 Changes in version 1.0.0 (2007-12-06)                  

Major release.

                 Changes in version 0.0.1 (2006-05-06)                  

First public release.
