Releases: jasp-stats/jasp-desktop
Releases · jasp-stats/jasp-desktop
Release 0.17.1
Bug fixes
- Repeated Measures ANOVA does not run if the model terms are not explicitly set (#1987)
- JASP crashes sometimes when starting it with an internet proxy (#1986)
- Wrong name in T-Tests "Wilcoxon signed-rank" option should be labelled “Mann-Whitney” (#1990)
- Linear Regression: Casewise Diagnostics table shows wrong case ID when having missing data in dependent variable (#1989)
- No error bars displayed when running a 2-way ANOVA (#1983)
Release 0.17.0
New features and improvements
- Syntax mode
- Improved keyboard navigation (#4943)
- New module: Acceptance Sampling
- Structural Equation Modeling
- Network
- New analysis: Bayesian network analysis (#1864)
- Meta-Analyses
- New analysis: Bayesian Penalized Meta-Analysis (Utrecht project)
- JAGS
- Customizable Inference added (#83)
- Show tables and plots for specific variables or subsets of variables.
- Order the shown variable in the tables and plots by their posterior mean, posterior median, or a custom order.
- Compute and plot custom intervals for posterior distributions (CRI, HDI, or the probability of the area between two user-specified values).
- Save posterior samples generated to external file (#76)
- Improved for large data files (#1920)
- Customizable Inference added (#83)
- Regression
- Descriptives
- Added confidence intervals for mean, standard deviation, and variance (#123)
- Distributions
- Confidence intervals added to Q-Q and P-P plots (#82)
- Prophet
- Prediction is optional (#60)
- Engines and R-library separation behavior improved (#4726)
- Improved preference help file (#4881)
- Module selection in R commander (#1912)
- All plots can now be saved as .pptx, .pdf, and .eps (#113, #1910)
- Added documentation on how to connect to Snowflake warehouse (#1888)
Bug fixes
- Show strength in centrality table and plot in the network analysis module (#1692)
- Variance-covariance matrix could not always be calculated in SEM (#1845)
- Errors when using flexplot in visual modeling module (#1327, #1849)
- Temporarily disable no-bom-native-codepage workaround for CSVs on windows. Use utf-8 whenever you can to avoid this problem.
Release 0.16.4
New features and improvements
- SQL Database can be connected to JASP (#4798)
- ANOVA
- T-Tests
- Factor (#75, #93)
- Frequencies
- Cochrane Meta-Analyses (#12)
- Export selected datasets
- Improved error message for dataset with one estimate
- Improved Add Estimates
- Machine Learning
- New dataset: Palmerpenguins (#128)
- Mixed Models
- Network
- Improved Graphical Options (#51)
- Regression
- Reliability
- SEM (#90)
- Summary Statistics
- New analysis: Bayesian z-test (#77)
- New transformations added to Compute Column (#4799)
- Power transformations (Box-Cox, inverse Box-Cox, two parameter power transformation, Yeo-Johnson)
- Logit and inverse logit transformations
- JASP now uses R 4.2.1 (#4801) (Improves unicode handling on windows)
- Maximum amount of engines used by JASP can be forced by admin (#1693, #4838)
- Exponent notation available for Table Options (#4824)
- Userdefined label for missing values in data (#4833)
Bug fixes
- PCA results based on covariance instead of correlation did not change output (#1667)
- Custom contrasts in ANOVA did not update after filtering data (#627, #4741)
- Special character in username results crashes plots (#1758)
- Raincloud plot title fix (#181)
- Posterior model probabilities should update when the model prior changes (#195, #1816)
- Fix to Column Filter (#4736, #1661)
- (Partial) fix to results displaying ugly on windows (#1747)
- Solves a crash on macOS in linear regression for extremely small numbers (#1777)
- Fix Prophet to allow it to find timezone and date-format information (#1714)
Release 0.16.3
New Features and Improvements
- New Module: Quality Control.
- New Module: Bayesian State Space Models.
- New analysis added: Frequentist/ Classical Generalized Linear Model to the Regression module (jasp-stats/jasp-issues#1656).
- ANOVA:
- The Bayesian repeated-measures ANOVA now includes random slopes for all but the highest order repeated-measures interaction. This brings the results from the frequentist and Bayesian repeated-measures ANOVA closer together. The old behavior can be retrieved by checking Legacy results.
- Bayesian ANOVA, Bayesian ANCOVA, and Bayesian repeated-measures ANOVA (jasp-stats/jaspAnova#151)
- The value for the r-scale of the prior distribution on the coefficients can now be set individually for fixed effects.
- The principle of marginality can be violated individually for fixed effects and random slopes. By default it is respected for fixed effects and violated for random slopes.
- The prior on the model space can now be adjusted. Possible options are Uniform (default), Beta binomial, Wilson, Castillo, Bernoulli, and Custom. Custom allows for specifying individual inclusion probabilities for each effect. Note that the inclusion probabilities for interactions change if the principle of marginality is violated.
- Nuisance predictors and other predictors common to all models can be shown in the model by unchecking “Hide nuisance in model”.
- Warning message about changes (jasp-stats/jaspAnova#158).
- Message added when Welch correction fails (jasp-stats/jaspAnova#131).
- Improvements to Order restricted hypotheses (jasp-stats/jaspAnova#153)
- Regression:
- Bayesian linear regression now exports residuals and residual standard deviations (jasp-stats/jaspRegression#110).
- Add bootstrapped p-value to linear regression (jasp-stats/jaspRegression#120).
- Improved part and partial correlations (jasp-stats/jasp-issues#1638).
- T-Tests
- Coefficient of variation added to the descriptives tables of all T-Tests (jasp-stats/jaspTTests#122, jasp-stats/jasp-issues#1599).
- Bain:
- Draw correct posterior probability plot in all AN(C)OVA / regression / SEM situations (jasp-stats/jaspBain#80)
- Descriptives:
- Improved interface (jasp-stats/jasp-issues#1635)
- Frequencies:
- Bayesian Binomial Test (jasp-stats/jaspFrequencies#83, jasp-stats/jasp-issues#1605)
- Improved help files.
- Improved lay-out.
- Footnote added to display what kind of beta prior is used on the alternative hypothesis.
- Multinomial Test (#87)
- Increased number of decimals in the input table (from 1 to 3).
- Expected counts can be decimals (#1652).
- Bayesian Binomial Test (jasp-stats/jaspFrequencies#83, jasp-stats/jasp-issues#1605)
- Machine Learning:
- Matthews correlation coefficient added to evaluation metrics for classification (jasp-stats/jaspMachineLearning#134, jasp-stats/jasp-issues#750).
- Decision Trees able to only show relevant splits (jasp-stats/jaspMachineLearning#152).
- Improvement to decision boundary matrix (jasp-stats/jaspMachineLearning#152).
- Audit:
- UI improvements
- Reliability:
- Krippendorff's alpha added to Rater Agreement (#1665).
- Improvements to fit measure table for Bayesiansingle factor model (jasp-stats/jaspReliability#94).
Bug fixes
- Raincloud plot lines that connect data points within-subject (jasp-stats/jaspAnova#155).
- In RM ANOVA contrast couldn’t be negative (jasp-stats/jasp-issues#1704).
- ANOVA affected coding options in Regression (jasp-stats/jasp-issues#1677).
- Error in JAGS (jasp-stats/jasp-issues#1691).
- Help button not clickable in Bayesian Unidimensional Reliability (jasp-stats/jasp-issues#1621).
- Flatpak no longer crashes on certain actions (jasp-stats/jasp-issues#1424).
Release 0.16.2
New Features and Improvements
- JASP now runs natively on Apple Silicon!
- JASP uses the latest version of R 4.1.3
Bug Fixes
Release 0.16.1
New Features and Improvements
- New languages added: French (incomplete)
- Repeated Measures ANOVA post hoc tests
- Machine Learning
- New analyses decision tree regression and classification
- New analyses Support vector machine regression and classification
- K-medoids and K-medians clustering techniques added to Neighborhood-based (formerly K-means)
- Extra plot in k-nn regression and classification showing the relative weights as a function of the distance
- Extra evaluation metrics for all classification analyses
- New module: Cochrane Meta-Analysis
- Allows users to select and analyze data from Cochrane Database of Systematic Reviews
- Regression
- Bain
- An extra column showing PMPc has been added to the ANOVA, ANCOVA, SEM, and linear regression standard table outputs
- Descriptive Statistics
- Option to hide/ show Valid and Missing cases (#1480)
- Factor Analysis
- New option for parallel analysis: 1) can be based on FA or 2) PCA (#1319)
- Scree plot is overhauled to be more concise. New option to turn parallel analysis results on and off in the plot
- Reliability
- Audit
- Added an option to specify the maximum sample size
- Fixed a bug in the Bayesian workflow where the sample size started counting at 5, sometimes resulting in sample sizes that did not match the desired input criteria
- Improvements to “About”
- Improved layout of network graphs in Network module (#1366)
Bug fixes
- Path plots in Factor module fixed (#1548, #1570, #1573)
- Factor correlation matrix showed the correlations of the estimated factor scores. Now the proper factor intercorrelations are used
- Computing Z-scores does not give an error anymore (#1507)
- Pairwise plots able to handle ordinal variables (#1523)
Release 0.16.0
New Features and Improvements:
- New languages added: Spanish, Indonesian (incomplete)
- Meta-Analysis:
- WAAP-WLS
- PET-PEESE
- Prediction Model Performance
- Set seed option in Bayesian Meta-Analysis
- Machine Learning:
- New Neural Network regression and classification analyses
- Model saving for all regression and classification analyses
- Models can be saved locally as .jaspML files
- Prediction for all regression and classification models
- Locally saved .jaspML files can be loaded and applied to new data in the new prediction analysis
- New module: Circular
- Perform Circular Statistics
- Summary Statistics:
- General Bayesian Tests
- ANOVA:
- GORIC(A) in AN(C)OVA and RM ANOVA
- Regression:
- Row/column/total proportions parallel to counts for confusion matrix in Logistic Regression
- Marginal effects plots in Linear Regression
- Audit:
- UI revamp
- Prior and posterior predictive plots
- Ability to use base 10 for scientific notation in tables
Bug fixes:
Release 0.15.0
New Features and Improvements:
- Basic plot editing functionality
- New languages added: Japanese, Galician, Portuguese, Chinese (incomplete), Spanish (incomplete)
- Better support for non-western systems:
- Support installing jasp for user with username in native locale characters and make jasp not crash immediately
- Support reading “native” locale encoded csvs and utf-8 both, dependent on the BOM file format
- Allow custom fonts in plots
- SEM:
- T-Tests and ANOVAs:
- Raincloud plots
- One-Sample T-Test provides location and effect size estimates relative to the test value (#1158)
- Distributions:
- Major update including 23 new continuous and 5 discrete distributions
- Learn Bayes:
- Binary classification
- Game of chance
- Game of skill
- Buffon’s needle
- Some edge case scenario and esthetical fixes for binomial estimation
- Descriptives:
- Regression:
- Frequencies:
- Factor analysis
- Factor loadings can now be sorted in EFA and CFA (#1215)
- Mixed Models:
- Changing default links for gamma and negative binomial family
- Fixing plotting themes
- Improving warning messages
- Meta-Analyses:
- Expanded Robust Bayesian Meta-Analysis to account for publication bias (https://psyarxiv.com/kvsp7/)
- Reliability:
- Classical intraclass correlation coefficients
- Tables now report means and standard deviations of participants’ sum scores or mean scores
- Restructuring of R-code to improve computational time
- Reduced .jasp file sizes produced by sampling in reliability analysis
- Bain:
- Added Bayesian informative hypotheses evaluation for structural equation models
- New module “Prophet” for time series prediction (in beta)
Bug fixes:
- Several fixes to sav loading (spss files)
- Loading data repeated measures ANOVA (#1114)
- Fixed results table of Classical Binomial Test (#1281)
- Fixed error in density graph Bayesian meta-analysis (#1236)
- Fixed error in computing the total number of edges in Network analysis (#1339)
- Resized plots are kept after refreshing (#824)
Release 0.14.1.0
Improvements:
- Several changes to the audit module
-- Evaluate audit samples more efficiently #4411
-- More prior construction methods #700
-- UVB functionality #1149 - More computing speed for RoBMA #4408
- Previously the effect size estimates and confidence intervals in the one-sample t-test were defined in terms of differences with respect to the test point. Both the effect size estimates and confidence intervals are now defined independently of the test point #4407
- Updates to Data Library (new Datasaurus and World Happiness examples) #4393
Bug fixes
0.14.0.0
New Features and improvements:
- Adjust for publication bias in meta-analysis
- Robust Bayesian meta-analysis
- Selection models
- Learn Bayes module
- PDF export of results
- Choose your font
- German translation
- Frequentist partial correlations
- Confidence and prediction intervals in linear regression
- New distributions in the Distributions module (log-normal and logistic)
- Installing Jasp without admin rights (#861)
Bug fixes:
- Fixes to spreadsheet (#901, #914, #933, #935)
- Large dataset: JASP crash when changing variable type (#938)
- Analysis of Effects: Across Matched Models (#852)
- Corrections of bugs found during the JASP workshop (#4257)
- Summary Statistics Bayesian Independent Samples T-Test: Labels in prior and posterior plots switch upon changing from BF10 to BF01 (#796)
- Fixes to Quill (#694, #695, #696)
- Issues with display of icons and characters (#851, #929)
- ROC and PR curve (after classification) (#277)
- Plots on linux are only saved as png (#889)
- Inconsistency between frequentist and bayesian non parametric paired t-test (#868)
- Add methods median, ward.d, ward.d2 and mcquitty (#4222)
- Contingency tables properly reverse column labels (#4211)
- Posterior summaries in Bayesian linear regression and Bayesian ANOVAs show exclusion probabilities (#625)
- Log Odds Ratio value not updated (#915)
- Bayesian Mixed Effect Models with aggregated binomial family (#957)
- Reliability: Average interitem correlation 95%CI, does not include point estimate (#888)