Getting started with JASP

JASP has been developed and maintained by several research groups in Europe and it is available under a GNU Affero GPL v3 open access license [Reference Wagenmakers, Kucharský, van Doorn and van den Bergh12]. Key elements of JASP are: providing a user-friendly interface for basic and advanced statistical analyses, facilitating reproducibility of analyses (by being an interface to R code), aiming for clarity and easy use and promoting quality of figures and tables [Reference Wagenmakers, Kucharský, van Doorn and van den Bergh12]. It is used by researchers in hundreds of institutions in dozens of countries [Reference Wagenmakers, Kucharský, van Doorn and van den Bergh12] and users can download it from jasp-stats.org and it has options for Windows, macOS and Linux (at the time of writing, a recent stable version of JASP -0.98.1- was available).

JASP is extremely user-friendly, as it has a very easy-to-use menu for the main options for analyses, such as descriptive statistics, t-tests, ANOVA, among others (Figure 1A). The underlying engines for statistical analyses are more than 400 well known R packages (such as stats) [Reference Wagenmakers, Kucharský, van Doorn and van den Bergh12], which have been created, tested and maintained by the global scientific community [Reference Wagenmakers, Kucharský, van Doorn and van den Bergh12,Reference Giorgi, Ceraolo and Mercatelli16].

Figure 1.

A. A schematic overview of the upper menu of JASP (the arrows highlight the test covered in this tutorial). B. A schematic overview of the main menu of JASP. C. An overview of an example of a dataset. D. Datasets available in JASP and used in this tutorial.

Figure 1. Long description

Panel A (Upper Menu): A horizontal toolbar where blue arrows point to five key analysis modules covered in the tutorial: Descriptives, T-Tests, ANOVA, Regression, and Frequencies. Unhighlighted items include Edit data, Mixed Models, Factor, Machine Learning, and Meta-Analysis. Panel B (Main Menu): A two-column schematic layout. The left column lists navigation options under Open and Save (e.g., Computer, Data Library). The right column lists numbered analysis sections corresponding to the tutorial: 1. Descriptives, 2. T-Tests, 3. ANOVA, 4. Regression, 5. Frequencies, and 6. Factor. Panel C (Example of Dataset): A data table displaying sample rows with columns for ParticipantNumber, Condition (e.g., Horizontal, Fixation), and CriticalRecall values. Panel D (Datasets Used): A two-column table mapping available JASP datasets (e.g., “Scatter plot”, “Pain thresholds”) to their specific tutorial applications and figure cross-references.

For training purposes, JASP includes multiple files with examples for the different types of tests (Figure 1B), such as the ones covered in this tutorial. In addition, there are many manuals and videos online, in multiple languages (details can be found at jasp-stats.org/teaching)

For the analysis of their own data, users can load .csv or .xlsx files (commonly used by researchers) and easily select the analyses to be run. The software can also read data from several standard statistical software packages. Supported formats include IBM SPSS, SAS, and Stata. Figure 1C shows an example of a dataset, which has two categorical (nominal) variables and one quantitative (scale) variable. The datasets available in JASP and employed in this tutorial are described in Figure 1D.

In terms of output, it has the option of generating tables and figures, of publication-ready quality, that can be edited or exported to be used in articles or presentations, such as quantile-quantile (Q-Q), distribution, box and correlations plots (Figures 2A-2D). An interesting advantage of figures generated by JASP is that it can use options from the advanced ggplot2 package [Reference Wickham17] from the R computing environment. In addition, JASP shows the underlying R packages and codes used for the analyses, facilitating transparency and reproducibility in code and methods[Reference Streiber, Hoepel and Blok18]: Figures 3B and 3C, show examples of R packages employed and the underlying R code for a descriptive analysis, respectively.

Figure 2.

A. Example of a Q-Q plot. B. Example of a distribution plot. C. Example of a box plot. D. Example of a correlation plot.

Figure 2. Long description

Panel A (Q-Q plot): Shows Theoretical Quantiles vs. Sample Quantiles for Neuroticism scores (ranging from 1 to 5) aligned along a red diagonal reference line, assessing normality. Panel B (Distribution plot): A bell-shaped histogram showing frequency Counts on the y-axis (0 to 100) against Neuroticism scores on the x-axis (1.5 to 4.5), centered around 2.8. Panel C (Box plot): Compares Pain Tolerance (y-axis: 30 to 80) across four Hair Color categories (x-axis). Light Blond exhibits the highest median pain tolerance (~60), followed by Dark Blond (~52), Light Brunette (~42), and Dark Brunette (~35 with an outlier at 51). Panel D (Correlation plot): A scatter plot showing Anxiety (y-axis: 0 to 100) versus Exam scores (x-axis: 0 to 100) with a downward-sloping regression line, indicating negative correlations.

Figure 3.

A. Menu for contingency tables. B. Example of R packages employed. C. Example of R code.

Figure 3. Long description

Figure 3 consists of three panels (A–C) demonstrating the interface, backend packages, and generated R code for contingency table analysis in JASP: Panel A (Menu, Contingency Tables): Displays the JASP options setup window. Fields show “Revise” assigned to Rows and “Fruit Consumption” assigned to Columns. Selected statistical checkboxes include “X2” (Chi-square) under Statistics, “Contingency coefficient” under Nominal, and “Log Odds Ratio” under Odds Ratio (2x2 only). Panel B (Example of R packages employed): A text box listing core R packages used internally by JASP, including ggplot2, ggrepel, grid, and stats. Panel C (Example of R code): A syntax box showing equivalent R code executed by JASP: jaspDescriptives::Descriptives(data = NULL, version = “0.95”, formula = ~ ‘Weight After’ + ‘Weight Before’, median = TRUE, mode = TRUE, shapiroWilkTest = TRUE, skewness = TRUE)

Reproducibility and transparent science with JASP

An additional strength of JASP is its contribution to reproducible and transparent biomedical research workflows [Reference Love, Selker and Marsman8]. The software allows users to inspect and export the underlying analytical syntax linked to the R statistical environment, facilitating verification and reproducibility of statistical analyses. By integrating graphical user interfaces with open-source computational backends [Reference Shepherd and Richardson6], JASP promotes greater transparency in analytic pipelines and supports reproducible workflows that can be independently reviewed and replicated by other researchers. These features are increasingly important in the context of open science initiatives and current efforts to improve transparency and reproducibility in biomedical research [Reference Streiber, Hoepel and Blok18,Reference Wilkinson, Dumontier and Aalbersberg19]. Furthermore, the availability of exportable outputs and underlying code may facilitate teaching activities, collaborative research, and methodological standardization across institutions.

For developers, or advanced users, the underlying code and annotation for JASP are available at the GitHub repository (github.com/jasp-stats).

Descriptive statistics

Users can select multiple statistics for central tendency, such as the mean for variables with normal distribution and the median for variables with non-normal distribution (Figure 4A), in addition to statistics for dispersion, such as the standard deviation for variables with normal distribution and the interquartile range for variables with non-normal distribution [Reference Mishra, Pandey, Singh, Gupta, Sahu and Keshri20,Reference Vetter21] (Figure 5A). For the example shown in Figure 5A, the variable Neuroticism has a mean of 2.826 and a standard deviation of 0.453, with a minimum of 1.648 and a maximum of 4.375.

Figure 4.

A. Menu for descriptive statistics. B. Menu for independent samples t-test. C. Menu for independent ANOVA test. D. Menu for linear correlation.

Figure 4. Long description

Figure 4 is a composite diagram with four panels (A–D) showing statistical configuration windows in JASP: Panel A (Descriptive Statistics): Variables include “Moon” and “Other”. Selected checkboxes include Valid (Sample Size), Quartiles, Mode (Central tendency), Skewness (Distribution), and Std. deviation and Maximum (Dispersion). Panel B (Independent samples t-test): “CriticalRecall” is set under Dependent Variables. Selected options include Student test, Location parameter, Cohen’s d effect size, Normality assumption check, and Descriptives plots. Panel C (Independent samples ANOVA): “Revise” and “Exam” are listed under Dependent Variables, with “Hair Color” and “Exam” under Fixed Factors. Descriptive statistics display is checked. Panel D (Linear Correlation): “Revise” and “Exam” are under Dependent Variables. Options checked include Pearson’s r, Correlated hypothesis, Scatter plots, Display pairwise, Report significance, and Confidence intervals.

Figure 5.

A. Example of descriptive statistics. B. Example of results for an independent samples t-test C. Example of results for a paired samples t-test.

Figure 5. Long description

Figure 5 consists of three output tables (A–C) generated in JASP: Panel A (Descriptive Statistics): A table summarizing Neuroticism and Extraversion for 500 valid cases. Key metrics include means (2.826 and 3.487), standard deviations (0.453 and 0.355), Shapiro-Wilk p-values (0.054 and < .001), and percentiles (25th, 50th, 75th). Panel B (Independent two samples): An Independent Samples T-Test table comparing groups for variable “drp”. Results show Student’s t: t(42)=-2.267,p=.029, Cohen’s d=-0.684; Mann-Whitney U: W=135.000,p=.013, rank biserial correlation =0.441. Panel C (Paired two samples): A Paired Samples T-Test table comparing “Weight Before” vs. “Weight After”. Results show Student’s t: t(15)=-10.841,p<.001, Cohen’s d=-2.710; Wilcoxon signed-rank test: W=0.000,z=-3.516,p<.001, matched rank biserial correlation =-1.000.

In terms of figures, users can generate bar plots and pie charts [Reference Vetter21], in addition to box plots (including the options for Violin elements) [Reference Hu22] (Figure 2C).

Normality testing

The researchers can use the Shapiro–Wilk test [Reference Mishra, Pandey, Singh, Gupta, Sahu and Keshri20], in addition to skewness and kurtosis [Reference Kim23] (Figure 4A), to assess normality in selected continuous variables (Figure 5A). As the Shapiro–Wilk test is oversensitive in large samples (it can yield significant p values for trivial deviations from normality), it is important to carry out a visual inspection of the distribution plots and Q-Q plots [Reference Kim24,Reference Habibzadeh25], which can be easily generated in JASP (Figure 2A and 2B). For the example shown in Figure 5A, the Shapiro–Wilk test provides evidence against the null hypothesis of normality for Extraversion. However, as noted above, this result should be interpreted together with visual assessment of the distribution and Q–Q plots. In the case of the example shown in Figure 2A, the variable analyzed seems, visually, to have a normal distribution.

To facilitate the selection of appropriate tests by novice users [Reference Mishra, Pandey, Singh, Gupta, Sahu and Keshri20], we created a schematic flow diagram for the statistical tests covered here (Figure 6). In addition, it is important to take into account the respective assumptions underlying each test [Reference Bewick, Cheek and Ball26–Reference Whitley and Ball29]. In addition to p values, researchers should consider effect size estimates, which provide information about the magnitude and practical relevance of observed differences [Reference Dwivedi3]. Effect size measures complement statistical significance testing and may facilitate interpretation of the practical importance of study findings [Reference Dwivedi3].

Figure 6.

Decision-support flow diagram for selecting the basic statistical tests covered in this tutorial.

Figure 6. Long description

Figure 6 is a decision matrix guide mapping variable types to suggested statistical tests under normal and non-normal distributions: Row 1 (Continuous vs. Categorical, 2 groups): For a ruler icon (continuous) and Venn diagram icon (categorical, 2 groups), suggests Student’s t-test (independent/paired) for normal distributions, and Mann-Whitney or Wilcoxon signed-rank test for non-normal distributions. Row 2 (Continuous vs. Categorical, 3+ groups): For continuous and categorical (3+ groups) variables, suggests one-way ANOVA or repeated measures ANOVA for normal distributions, and Kruskal-Wallis or Friedman test for non-normal distributions. Row 3 (Categorical vs. Categorical): For two categorical variables, suggests Chi-square test for normal distributions (no non-parametric alternative listed). Row 4 (Continuous vs. Continuous): For two continuous variables, suggests Pearson’s r correlation for normal distributions, and Spearman’s rho correlation for non-normal distributions.

Comparison of means between two groups or repeated measures

Researchers are able to calculate, for continuous variables with normal distributions, the Student’s t-test for independent samples (Figure 4B) or for paired samples [Reference Whitley and Ball29] and their corresponding p values and effect sizes (Figures 5B and 5C). In addition to normality, homogeneity of variances is a key assumption in t-tests and the Levene’s test is commonly used to verify it (a p value < 0.05 means that the condition of equal variance is not met) [Reference Kim30]. In cases of lack of homogeneity of variances, for independent samples, it is possible to calculate the t-test for unequal population variances. The test is now popularly known as the Welch t-test [Reference Kim30], and has been proposed to be used more frequently in certain research fields [Reference Delacre, Lakens and Leys31].

For continuous variables with non-normal distributions, the researchers can estimate the following non-parametric tests: Mann–Whitney test for independent samples and Wilcoxon signed-rank test for paired samples [Reference Whitley and Ball32] (Figures 5B and 5C). Non-parametric tests have the advantage of requiring fewer assumptions on the data, in comparison with parametric tests and they are useful in cases of samples with very skewed distributions, presence of heteroscedasticity (unequal variances) and for analysis of ordinal variables [Reference Whitley and Ball32].

For the example shown in Figure 5B, both Student’s t-test for independent samples and Mann–Whitney test identify significant differences between the two groups. Moreover, for the example shown in Figure 5C, both Student’s t-test for paired samples and the Wilcoxon signed-rank test find highly significant differences between the two groups.

In terms of figures, users can generate bar plots, box plots or raincloud plots (Figure 2C), among others.

Comparison of means between three or more groups

For variables with normal distributions, users can calculate the one-way ANOVA test for independent samples (Figure 4C) or a repeated measures ANOVA test, including the corresponding p values [Reference Bewick, Cheek and Ball28,Reference Kim33] (Figures 7A and 7B) and descriptive statistics. Similarly to t-tests, homogeneity of variances is a key assumption for ANOVA and the Levene´s test can be used to verify it (JASP offers the option of running a Welch’s ANOVA when there is no homogeneity of variances). In order to identify the specific significant contrasts and to correct for multiple comparisons, users can adjust p-values with different types of tests, such as Tukey, Bonferroni and Scheffe (Figures 7A and 7B). In the context of ANOVA tests for independent sample, Tukey is commonly used and it tests all possible pairs, Bonferroni can be used for specific planned pair comparisons (as it is very conservative and reduces statistical power) and Scheffe is useful for testing both simple and complex contrasts [Reference McHugh34]

Figure 7.

A. Example of results for an independent samples ANOVA. B. Example of results for a repeated measures ANOVA.

Figure 7. Long description

Figure 7 displays ANOVA and non-parametric test results across two panels (A–B) in JASP: Panel A (Independent three or more samples): ANOVA (Pain Tolerance by Hair Color): Significant main effect, F(3,15)=6.791, p=.004. Kruskal-Wallis Test: Significant effect, H(3)=10.59, p=.014. Post Hoc Tests (Tukey HSD): Significant mean differences between Dark Brunette and Light Blond (-21.800, p=.004), and Light Blond and Light Brunette (16.700, p=.037). Panel B (Paired three or more samples): Repeated Measures ANOVA (Food): Significant effect, F(3,21)=3.794, p=.026,?^2=0.239 (sphericity assumption violated, p<.05). Friedman Test: Significant effect, X^2 (3)=11.53, p=.009, “Kendall’s “ W=0.480. Post Hoc Tests: Significant differences for Level 1 vs. Level 2 (p_“bonf” =.012) and Level 1 vs. Level 3 (p_“bonf” =.006).

When the assumptions of parametric ANOVA are not adequately met, users can consider nonparametric alternatives, including the Kruskal–Wallis test for independent samples and the Friedman test for repeated measures [Reference Bewick, Cheek and Ball35] (Figures 7A and 7B).

For the example shown in Figure 7A, one-way ANOVA test for independent samples and Kruskal–Wallis test identify significant differences between the multiple groups. Moreover, for the example shown in Figure 7B, both repeated measures ANOVA test and the Friedman test find significant differences between the multiple groups.

In terms of figures, users can generate bar plots or box plots (Figure 2C).

Contingency tables

For the analysis of categorical variables, users can select the option of Frequencies/Contingency tables, in order to calculate the χ2 (chi-square) test of independence (Figure 3A) and the corresponding p values [Reference Bewick, Cheek and Ball27,Reference McHugh36] (Figure 8A). According to the example shown in Figure 8A, there is statistical significance (p value of 0.007) for an association between the two categorical variables (each one with three categories). For 2X2 tables, JASP can estimate the odds ratio (OR), together with its confidence interval. An OR below 1 indicates lower odds of the outcome in the exposed group, whereas an OR above 1 indicates higher odds, with interpretation depending on the coding and reference categories [Reference Bewick, Cheek and Ball27].

Figure 8.

A. Example of results for a contingency table. B. Example of results for a linear correlation.

Figure 8. Long description

Figure 8 presents statistical analysis results generated in JASP across two panels (A–B): Panel A (Contingency tables): Crosstabulation Table: A 3×3 table comparing levels of Physical Activity (Low, Moderate, Vigorous) against Fruit Consumption (Low, Medium, High) across N=1,184 total observations. Chi-Squared Tests Table: Shows a statistically significant association between Physical Activity and Fruit Consumption, X^2 (4)=14.15, p=.007. Panel B (Linear correlation): A correlation table comparing Big Five personality traits using both Pearson (r) and Spearman (?) coefficients: Neuroticism vs. Extraversion: Significant negative correlation (r=-0.350, p<.001;?=-0.325,p<.001). Neuroticism vs. Agreeableness: Weak negative correlation (r=-0.134, p=.003). Extraversion vs. Openness: Significant positive correlation (r=0.267, p<.001). Openness vs. Agreeableness: Significant positive correlation (r=0.159, p<.001).

Linear correlation

For correlation coefficient analyses, the researchers can estimate the Pearson’s r correlation coefficients (Figure 4D) and their corresponding p values (Figure 8B), [Reference Schober, Boer and Schwarte37,Reference Akoglu38] in addition to visualizing the correlation plots [Reference Bewick, Cheek and Ball26] (Figure 2D). For the example shown in Figure 8B, there are significant correlations between several quantitative variables, such as a positive correlation between Extraversion and Openness (r: 0.267) and a negative correlation between Neuroticism and Extraversion (r: –0.350). For variables with a non-normal distribution, the users can calculate the non-parametric Spearman’s rho correlation coefficients and their corresponding p values [Reference Schober, Boer and Schwarte37,Reference Akoglu38] (Figure 8B).

Further considerations

This is one of the first peer-reviewed and user-friendly tutorials for basic statistical tests using the open JASP software. The basic statistical tests selected to be covered here are quite frequently used in biomedical research and are part of the basic competencies of health sciences researchers [Reference Cressman and Sharp5,Reference Enders, Lindsell and Welty13,Reference Kim, Kim and Kwak14]. In this context, we provided, for the novice user, a suggestion of peer-reviewed articles on the statistical tests covered here [Reference Bewick, Cheek and Ball26–Reference Bewick, Cheek and Ball28], which might be useful for the strengthening of key concepts in applied biostatistics, in addition to existing textbooks. This tutorial is intended as an introductory educational resource and should complement, rather than replace, formal training in biostatistics and research methods [Reference Enders, Lindsell and Welty13]. A key element in our tutorial paper, in comparison to short online materials, is that our suggested analytical steps are strongly based on peer-reviewed papers in the biostatistics field [Reference Enders, Lindsell and Welty13,Reference Habibzadeh25,Reference Bewick, Cheek and Ball27–Reference Whitley and Ball29,Reference Kim33]. Our article can be complemented with several short videos freely available on YouTube (https://jasp-stats.github.io/jasp-video-library) and for the interested reader, these are some popular YouTube channels with multiple videos on JASP software: @JASPStatistics; @ResearchByDesign; @StatisticsofDOOM; @TomFaulkenberry; @ChenShawJiun; @lindenloot2372. This tutorial has an intentionally introductory scope and does not cover more advanced analytical approaches available in JASP, such as multivariable regression, survival analysis, factor analysis, Bayesian methods, or machine-learning techniques. These methods require consideration of additional assumptions, model-building strategies, and interpretation issues and are therefore better addressed in dedicated methodological tutorials.

The strengthening of competencies in the use of basic statistical tests [Reference Enders, Lindsell and Welty13] is a major need in the health sciences workforce around the world [Reference Ocaña-Riola39], particularly in certain fields and regions [Reference Gosselin40]. The statistical methods used in medical research also reflect the importance of introducing statistics in medical education [Reference Nowacki, Brearley and Oster41]. To make the training meaningful for students, it is advisable to use user-friendly software that does not place an excessive financial burden on the training unit. Promoting and facilitating, through detailed tutorials, the adequate implementation of user-friendly and open statistical software might contribute to the training of scientists around the world [Reference Shepherd and Richardson6,Reference Weissgerber, Garovic and Milin-Lazovic42–Reference Balduzzi, Rucker and Schwarzer44].

Successful statistical data processing in medicine requires knowledge of medical research problems, statistics, and computational analysis software [Reference Shepherd and Richardson6,Reference Enders, Lindsell and Welty13]. Researchers in health sciences, including medicine, should not relinquish decision-making authority to consultants or biostatisticians in areas in which they have the most expertise. Scientists must understand how statistical software derives conclusions from their data regarding their research questions [Reference Enders, Lindsell and Welty13]. Researchers must be able to use statistical software as a tool when collecting and analyzing observational data and when reporting research results [Reference Shepherd and Richardson6]. It appears that researchers’ statistical data analysis is accelerating research projects and that processing statistical data by themselves appears to accelerate and potentiate projects. JASP is well suited for this purpose as its computational algorithms are comprehensive and reliable [Reference Love, Selker and Marsman8]. Another advantage is that it is easy to use, and there are no financial barriers to using the program [Reference Shepherd and Richardson6].

Implications for low- and middle-income countries

Open-source statistical platforms such as JASP may have particularly important implications for low- and middle-income countries (LMICs), where access to proprietary statistical software can represent a substantial economic barrier for students, researchers, and academic institutions. Strengthening statistical and data analysis competencies is increasingly recognized as a key component of research capacity building and digital transformation in health systems globally [45].

In addition, statistical literacy and data analysis competencies are increasingly recognized as core components of digital health capacity building, data-driven decision-making, and evidence-informed digital transformation initiatives within health systems [Reference Enders, Lindsell and Welty13].

In many LMIC settings, limitations in infrastructure, funding, and specialized training opportunities continue to affect the development of local research ecosystems [Reference Bates, Akoto and Ansong46,Reference Franzen, Chandler and Lang47]. In this context, freely available and user-friendly analytical tools may contribute to democratizing access to quantitative research training, improving reproducibility practices, and supporting evidence-based decision-making in biomedical and public health research. Furthermore, the integration of open statistical software into educational and research environments may facilitate broader participation in global scientific collaboration and open science initiatives [Reference Peršić and Straza48].

The use of open software, in addition to facilitating access to researchers in environments with limited economic resources [Reference Shepherd and Richardson6], has advantages in terms of reproducibility and transparency, facilitating the sharing of statistical code [Reference Haven and Ioannidis7,Reference Glasziou, Altman and Bossuyt49]. Computational reproducibility has become an increasingly important component of high-quality biomedical research, particularly in the context of growing concerns regarding the reproducibility crisis across scientific disciplines. Recent studies have highlighted persistent challenges in reproducing computational workflows, including incomplete documentation of analytical pipelines, unavailable code, missing dependencies, and insufficient reporting of statistical procedures [Reference Samuel and Mietchen50,Reference Antunes and Hill51]. In this context, open-source platforms such as JASP may contribute to improving transparency and reproducibility by facilitating access to underlying statistical code, standardized analytical workflows, and exportable outputs. In addition, integrating reproducible computational practices into statistical education may strengthen open science competencies among students and researchers in the biomedical and public health sciences [Reference Streiber, Hoepel and Blok18]. Promoting user-friendly tools that combine graphical interfaces with transparent computational backends may therefore represent an important strategy for advancing reproducible and trustworthy biomedical research.

As biomedical datasets continue to increase in volume and complexity, strengthening foundational competencies in statistical reasoning becomes essential before researchers can appropriately adopt more advanced analytical approaches, including machine learning and artificial intelligence [Reference Hao, Xu, Li and Evans52,Reference Bi, Goodman, Kaminsky and Lessler53], in addition to other methods, such as multivariable regression models [Reference Sperandei54] and factor analysis [Reference Rogers10], among others. In this context, JASP brings an interesting opportunity as it has user-friendly tools for multiple machine learning methods, both supervised and unsupervised, such as random forests, neural networks, support vector machines, among others [Reference Bi, Goodman, Kaminsky and Lessler53]. Multivariable regression models are fundamental and widely used tools in medical and epidemiological research for quantifying associations between risk factors, treatments, and patient health outcomes while accounting for multiple covariates. JASP offers tools for implementing several regression approaches in clinical medicine, epidemiology, and health care research. The selection, specification, and interpretation of these models should be guided by the research question, study design, characteristics of the outcome variable, and underlying model assumptions [Reference Nieminen, Toljamo and Vahanikkila55].

By lowering technical and financial barriers to statistical analysis while promoting reproducible research practices, open-source platforms such as JASP may contribute to strengthening global biomedical research capacity and advancing more transparent and equitable science [Reference Shepherd and Richardson6,Reference Haven and Ioannidis7].