NEWS
zcurve 2.5.0 (2026-11-06)
- Fixes the weighted EM algorithm in
zcurve_clustered(method = "w"), which down-weighted the component densities instead of the observations and overestimated ERR and EDR. Its bootstrap now resamples articles instead of individual test statistics. Contributed by Shaheed Azaad (#33, #37).
- Uses a cluster bootstrap in
zcurve_clustered(method = "b"), so the confidence intervals include the between-article sampling variation. Contributed by Shaheed Azaad (#36, #38).
- Fixes the EM algorithm reporting the log-likelihood of the previous iteration, which affected the convergence check, the choice of starting values, and the reported fit, and fixes the normalizing constant of the truncated normal used when the component means are estimated. Identified by Richard D. Morey (#30, #31, #32, #40).
- Fixes
zcurve_data() misreading rounded censored test statistics (e.g., "F(2,23) > 10") and treating values in scientific notation as rounded (#41).
- Fixes
zcurve_clustered() dropping censored results above the fitting range (#41).
- Fixes the parallel bootstrap failing when the number of bootstrap samples is smaller than the number of cores (#41).
- Counts significant results (ODR, expected and missing number of studies) at
sig_level rather than at the start of the fitting range (#41).
zcurve 2.4.7 (2026-10-19)
- Replaces the evmix dependency with an internal implementation of reflected Gaussian kernel density estimation, preserving the density_dbc option.
- Documents control settings as option lists to satisfy R-devel Rd checks.
zcurve 2.4.6 (2026-01-21)
- Adds warnings for small sample sizes and homogeneity as suggested by E. Zwet.
zcurve 2.4.5 (2025-09-05)
- Fixes incorrect computation of ERR identified by Richard D. Morey
zcurve 2.4.4 (2025-08-20)
- Fixes incorrect computation of EDR/ERR when the start of the fitting range does not match the significance level.
zcurve 2.4.3 (2025-05-16)
- Adds conf.level argument to the summary function to change the confidence level of the confidence intervals.
- Updates documentation and references.
zcurve 2.4.2 (2024-05-28)
- Fixes https://github.com/FBartos/zcurve/issues/22
zcurve 2.4.1 (2024-05-23)
- Fixes ODR computation with censored data
zcurve 2.4.0 (2023-10-27)
- Implementation of ggploting function.
zcurve 2.3.0 (2023-01-05)
- Implementation of parallel bootstrap.
zcurve 2.2.0 (2022-11-29)
- Implementation of z-curve version that deals with clustered estimates.
- Fixing bootstrap of censored data -- now both the censored and uncensored data are bootstrapped.
zcurve 2.1.2 (2022-02-07)
- Fixing number of statistically significant studies displayed in the annotation of z-curve plot.
zcurve 2.1.1
- Allowing to pass the xlim and ylim arguments to the z-curve plot function.
zcurve 2.1.0
- Adding 'zcurve_data' function to transform different tests statistics into a z-curve data object.
zcurve 2.0.0
- Adding the possibility to fit z-curve to censored/rounded input.
zcurve 1.0.9 (2021-05-07)
- Changing the ERR estimates to be take the directionality of the expected replication into account.
zcurve 1.0.8 (2021-01-17)
- Fixing an input test of z-scores (checking whether p-values weren't used by a mistake) being triggered when all z-scores were negative.
zcurve 1.0.7 (2020-12-15)
- Fixing incorrect passing of the significance level when the lower fitting range is not equal to alpha.
zcurve 1.0.6 (2020-09-26)
- Fixing the formatting of the NEWS.md file.
zcurve 1.0.5
- Adding convenience functions for extracting the z-curve estimates directly from the fitted object. (issue #3)
zcurve 1.0.4
- Adding observed discovery rate, number of included studies, and number of significant studies to the summary output. (issue #3)
zcurve 1.0.3
- Fixing a help file entry and making sure of proper control specification '[[]]' vs '$' list accessing.
zcurve 1.0.2
- Fixing a bug when producing figure with no non-significant z-scores.
zcurve 1.0.1
- Added check for supplying p-values as z-scores. (issue #2)
zcurve 1.0.0 (2020-03-31)
- The release version of the package.