# How to Use jamovi: A Free Statistics Guide for t-Tests, ANOVA and Regression jamovi is a free, open-source statistics program that gives you point-and-click menus for t-tests, ANOVA and regression without a license fee. You download it from the jamovi project, install it like any desktop application, open a spreadsheet, and run your first analysis from the Analyses tab within minutes. The appeal is straightforward. Most students and bench researchers need a small set of procedures: compare two group means, compare three or more, and model a continuous outcome against one or more predictors. jamovi puts all three in a single window, updates the output live as you change options, and writes results into tables you can paste into a manuscript. Under the hood it runs on R, the open-source statistical language, so the numbers match what a scripted analysis would produce, but you never have to type code. That mix of zero cost, a graphical interface and a credible statistical engine is why jamovi now appears in peer-reviewed work. A forensic anthropology study measuring mandibular angles on 500 orthopantomograms (panoramic dental X-rays) ran its statistics in jamovi version 2.6, including regression models and discriminant function analysis [1]. ## What jamovi Is and How It Thinks jamovi is best understood as a front end. The program you see is a spreadsheet-style data editor plus a menu ribbon. The calculations are performed by R packages running behind the scenes. Each time you click an option, jamovi re-runs the underlying R function and refreshes the table on the right side of the screen. This live-updating design is the single most useful feature for learning statistics, because you can watch a p-value change as you tick a checkbox for equal variances or switch from a Student's t-test to a Welch's t-test. The program is modular. The core install ships with the common analyses: Descriptives, T-Tests, ANOVA, Regression, Frequencies, Contingency Tables and Correlation. Anything beyond that arrives as a module you install from the jamovi library, a built-in catalog. Modules cover mixed models, survival analysis, factor analysis, meta-analysis and more. Modules are free, and they install with a few clicks from inside the program. Three design choices matter for research practice: 1. **Reproducibility.** Every analysis you run is stored inside the save file, so reopening the file restores the full results, not just the data. 2. **Auditability.** jamovi has a syntax mode and an Rj Editor module that expose the underlying R commands, so a reviewer can see exactly what was computed. 3. **Citation hygiene.** The program displays its own version number, which you report in your methods section the same way laboratories report the version of any statistical package they used [1][2][3]. ## How to Download and Install jamovi The jamovi download is a single installer from the jamovi project's official website. There is no account, no license key and no registration form. Follow these steps: 1. Open the official jamovi website in a browser and go to the download page. 2. Pick the build for your operating system. Installers exist for Windows, macOS and Linux, and there is a ChromeOS route as well. 3. Choose between the current release and the slower-moving "solid" release. The solid build changes less often, which suits a lab that wants a fixed version for a long project. 4. Run the installer and accept the defaults. The download is a few hundred megabytes. 5. Launch jamovi. On first start you get an empty spreadsheet with a sample data set available from the menu. On a locked-down institutional machine you may need administrator rights to install. If you cannot install software, run jamovi from a portable install on a USB drive, which the project also supports. ## A Tour of the jamovi Window The window has three tabs across the top left and a results panel on the right. **Data tab.** A spreadsheet. Rows are observations (one participant, one sample, one tooth), columns are variables. You can type directly into cells, paste from Excel, or import a file. The bottom of the sheet can hold a filter row and a split-by row, both of which apply to every analysis downstream. **Variables tab.** A metadata view. For each column you set the variable type (continuous, ordinal, nominal or text) and the data type (integer, decimal, text). You can also add a description, set missing value codes, and define the measurement level. Getting this tab right prevents most downstream errors. **Analyses tab.** The ribbon. Buttons are grouped by family: Exploration, T-Tests, ANOVA, Regression, Frequencies, Factor. Clicking a button opens a dialog with a variable list on the left, a slot area on the right, and an options panel below. The results panel on the right holds one output block per analysis you run. Each block has a small arrow menu in its top-right corner for copying, duplicating or deleting the analysis. ## Preparing Your Data Before Any Analysis jamovi expects long-format data: one row per observation, one column per variable. A wide layout, where each participant occupies several rows or each group gets its own column, will break most analyses. Four preparation habits save time later: - **Set measure types deliberately.** A column of group labels should be nominal. A column of Likert responses is usually ordinal, though many analysts treat it as continuous with caveats. A column of reaction times is continuous. jamovi uses these settings to decide which analyses are offered and how tables are labeled. - **Code missing values explicitly.** Blank cells are read as missing. If your source file uses 999 or -99 for missing, declare that in the Variables tab so jamovi does not average those codes into your results. - **Use computed variables for derived measures.** The Data tab lets you create a new column from a formula, for example a difference score or a log transform. Computed columns recalculate automatically if the raw data change. - **Record group sizes and check them.** A t-test or ANOVA run on the wrong grouping variable produces a clean-looking table with the wrong answer. For importing, jamovi reads CSV, Excel, SPSS (.sav), Stata, SAS and plain text files. Files saved in jamovi's own format carry the .omv extension. ## Running a jamovi t Test The t-test family answers one question: is the difference between two means larger than the noise you would expect by chance. jamovi offers three variants under the T-Tests button. ### Independent Samples t Test Use this when two separate groups of people or samples are compared on one continuous outcome. 1. Click **T-Tests**, then **Independent Samples T-Test**. 2. Drag the outcome variable into **Dependent Variables**. 3. Drag the grouping variable into **Grouping Variable**. 4. In the panel that appears, select which two group levels to compare. 5. Open **Options** to choose the test variant and the extra output. The options panel is where the statistical decisions happen. Student's t-test assumes the two groups have equal variances. Welch's t-test does not and is the safer default when group sizes differ. The [Mann-Whitney U test](/blog/data-analysis/mann-whitney-u-test-definition-formula-example) is the nonparametric alternative, appropriate when the outcome is ordinal or badly skewed. You can also request a Cohen's d effect size, descriptive statistics, a [normality test](/blog/research-skills/how-to-test-for-normality) and an equality-of-variances test. Published work in exercise science shows the reporting standard jamovi supports. A biomechanical comparison of free-throw shooting in 25 athletes and 25 novices used independent t-tests or Mann-Whitney U tests with Bonferroni correction and Cohen's d effect sizes [4]. jamovi produces all of those elements in one dialog, which is exactly why it suits this kind of analysis. ### Paired Samples t Test Use this when the same units are measured twice, for example before and after a treatment, or left versus right side of the same subject. 1. Click **T-Tests**, then **Paired Samples T-Test**. 2. Drag both paired columns into **Paired Variables**. jamovi pairs them in the order you add them. 3. Request effect size and descriptives in **Options**. The paired test works on the within-subject differences, so it removes between-person variability from the error term. That is why it is almost always more powerful than an independent test applied to the same measurements. ### One Sample t Test Use this to compare a single mean against a fixed reference value, such as a manufacturer's specification or a published population norm. Drag the column into **Dependent Variables** and type the test value. ### What the t-Test Output Tells You The output block contains a descriptives table, an assumption check and the test table. Read them in that order. The assumption check reports Levene's test for equality of variances. A significant Levene's test means the variances differ, so read the Welch row rather than the Student row. The normality check (Shapiro-Wilk) matters most when group sizes are small. With large samples the t-test is robust to moderate departures from normality, and a significant Shapiro-Wilk result alone is not a reason to abandon the test. The test table gives the mean difference, its confidence interval, the t statistic, the degrees of freedom and the p-value. Report the mean difference and its confidence interval, not just p. A confidence interval that excludes zero carries the same information as p < 0.05 but also tells the reader the plausible size of the effect. ## Running a jamovi ANOVA Analysis of variance extends the t-test to three or more group means, and to designs with more than one factor. It does this by comparing the variability between group means to the variability within groups, summarized as an F ratio. ### One-Way ANOVA 1. Click **ANOVA**, then **One-Way ANOVA**. 2. Drag the outcome into **Dependent Variables** and the grouping factor into **Grouping Variable**. 3. Open **Post Hoc Tests** and add the comparisons you need. 4. Open **Assumption Checks** for Levene's test and normality. 5. Open **Effect Size** to add eta squared or omega squared. A one-way ANOVA tells you that at least one group mean differs from the others. It does not tell you which pair. That is the job of post hoc tests. Tukey's HSD is the standard choice when group sizes are equal and variances are similar. Games-Howell is the better choice when variances differ or group sizes are unequal. Bonferroni and Holm adjustments are also available. If variances are clearly unequal, jamovi also offers Welch's ANOVA, which does not assume homogeneity. ### Factorial ANOVA A factorial design crosses two or more factors, and the factorial ANOVA tests each factor plus their interaction. A dental study comparing microtensile bond strength across four groups illustrates the design: bur-cut versus laser-cut dentin crossed with two adhesive systems, ten specimens per cell, analyzed with two-way ANOVA models at a significance level of p < 0.05 [5]. In jamovi: 1. Click **ANOVA**, then **Factorial ANOVA**. 2. Put the outcome in **Dependent Variables**. 3. Put both factors in **Fixed Factors**. 4. Add the interaction terms in the **Model** builder. The default includes main effects and interactions. The output gives one F test per effect. The main effect of cutting method averages across both adhesives. The main effect of adhesive averages across both cutting methods. The interaction tests whether the effect of one factor depends on the level of the other, which is usually the scientifically interesting question. When an interaction is significant, main effects can be misleading. Request **Estimated Marginal Means** and plot the interaction to see the pattern. jamovi can draw the profile plot for you. ### Repeated Measures ANOVA For designs where the same subjects are measured at several time points or under several conditions, use **Repeated Measures ANOVA**. You define the within-subject factor, set the number of levels, and assign the corresponding columns. jamovi reports Mauchly's test of sphericity and offers the Greenhouse-Geisser correction when sphericity is violated. ### Assumption Checks and Post Hoc Tests ANOVA rests on three assumptions: independent observations, roughly normal residuals within groups, and similar variances across groups. jamovi checks the last two. Independence cannot be tested from the data and depends entirely on your design. Reaction time, gene expression ratios and colony counts often violate normality because they are skewed. A log or square-root transformation, created as a computed variable, frequently fixes this. Nonparametric alternatives such as the Kruskal-Wallis test are available under **ANOVA > Non-Parametric**. ## Running a jamovi Regression Regression models a continuous outcome as a function of one or more predictors. It answers two questions at once: how much of the variation in the outcome the predictors explain, and how much the outcome changes per unit change in each predictor while holding the others constant. ### Linear Regression 1. Click **Regression**, then **Linear Regression**. 2. Drag the outcome into **Dependent Variable**. 3. Drag predictors into **Covariates**. Categorical predictors go into **Factors**. 4. Add blocks if you want hierarchical entry, for example demographics first, then experimental variables. 5. Open **Model Fit**, **Model Coefficients**, **Collinearity Statistics** and **[Confidence Intervals](/blog/data-analysis/confidence-interval-formula-examples)** in the options. Published examples show how regression is used across very different fields. A retrospective study of guided implant surgery built linear regression models to trace the sources of positional deviation across 75 implant cases [6]. The forensic anthropology study of mandibular angles used regression to confirm sexual dimorphism after adjusting for age [1]. Both are ordinary applications of the same dialog. ### Reading the Output Tables jamovi produces [three tables](/blog/data-analysis/join-three-tables-sql) for a linear regression, and each answers a different question. **Model fit.** Reports R squared, adjusted R squared, the F statistic and its p-value. R squared is the proportion of variance in the outcome explained by the model. Adjusted R squared penalizes the addition of predictors that do not earn their place. A significant F test means the model as a whole explains more variance than an intercept-only model. **Coefficients.** One row per predictor, with an unstandardized estimate (B), a standard error, a t value, a p-value and a confidence interval. B is in the units of the outcome, so it reads as "a one-unit increase in this predictor is associated with a B-unit change in the outcome." Ticking **Standardized estimate** adds beta, which puts all predictors on a common scale and lets you compare their relative strength. **Collinearity statistics.** Reports tolerance and the variance inflation factor (VIF) for each predictor. VIF above roughly 5 signals that a predictor is largely redundant with the others, which inflates standard errors and makes coefficients unstable. ### Logistic Regression When the outcome is binary, such as success or failure, use **Regression > [Logistic Regression](/blog/data-analysis/logistic-regression-definition-formula)**. The output switches to odds ratios and a classification table. Odds ratios above 1 mean the predictor increases the odds of the outcome. Confidence intervals that include 1 correspond to non-significant effects. ## jamovi vs SPSS The practical question most people ask is whether jamovi can replace SPSS. For the core analyses covered here, the answer is yes. Both programs are menu-driven, both produce tables in the same general layout, and both let you save an analysis file that reopens with the results intact. The differences are cost, engine and ecosystem. | Feature | jamovi | SPSS | |--|--|--| | Cost | Free and open source | Commercial license, typically institutional | | Statistical engine | R packages running underneath | Proprietary | | Interface | Point-and-click with live-updating output | Point-and-click plus a separate syntax window | | File format | .omv native, opens .sav, .dta, .csv, .xlsx | .sav native, opens many formats | | Extensibility | Free modules from the jamovi library | Commercial add-on modules and macros | | Scripting | Syntax mode and the Rj Editor module | SPSS Syntax, Python and R integration | | Typical methods citation | Version number reported, e.g. version 2.6 [1] | Version number reported, e.g. version 26 [3] | SPSS remains deeply embedded in laboratory science teaching and clinical research. A randomized trial of blended learning in immunohematology analyzed its outcomes in SPSS version 26 [3], and a neonatal pain study used SPSS as well [2]. That pattern reflects institutional licensing and habit as much as capability. jamovi's edge is accessibility. A student can install it on a personal laptop the same evening, a small lab can equip every desk for nothing, and a reviewer can reproduce the analysis from the saved .omv file. Free and low-cost tools of this kind are increasingly used for quantitative work in resource-limited settings [7]. SPSS's edge is breadth. Large institutions with site licenses, long-running longitudinal datasets and specialized commercial modules have little reason to switch. For a new project built around t-tests, ANOVA and regression, jamovi covers the ground. ## Saving, Exporting and Reporting Results Save your work as a .omv file. It stores the data, the variable definitions and every analysis you ran, so the file is the complete record of your analysis. To move results into a manuscript, click the arrow in the corner of any output table and choose copy. Pasting into Word or Google Docs gives you an editable table. You can also export the whole results panel to PDF or HTML from the file menu. For reporting, follow the convention of your field. Report the test name, the statistic, the degrees of freedom and the p-value, then the effect size and its confidence interval. A t-test report reads: t(df) = value, p = value, d = value. An ANOVA report reads: F(df1, df2) = value, p = value, partial eta squared = value. A regression report gives R squared, F(df1, df2), the p-value for the model, then each coefficient with its standard error and confidence interval. Always include the jamovi version number in your methods section [1]. ## Common Mistakes and Limitations **Ignoring the measure type settings.** If a grouping column is left as continuous, jamovi will not offer it as a grouping variable. Set measure types before you start, not after an analysis fails. **Skipping the assumption checks.** Levene's test and the normality checks appear in the options for a reason. Reading a Student's t-test when variances are clearly unequal, or an ANOVA when group variances differ several-fold, produces a p-value you cannot defend. **Running many t-tests instead of one ANOVA.** Comparing four groups pairwise means six separate tests. At an alpha of 0.05 the chance of at least one false positive climbs well above 5 percent. A single ANOVA with a Tukey correction keeps the family-wise error rate in check. **Treating p < 0.05 as proof of a real effect.** A p-value is the probability of data this extreme if the null hypothesis were true. It says nothing about the size or the practical importance of the effect. Report effect sizes and confidence intervals. **Reading R squared as predictive accuracy.** A high R squared in the sample does not guarantee good prediction in a new sample, particularly with many predictors and few observations. Check adjusted R squared and, where possible, validate on held-out data. **Overlooking collinearity.** Two predictors that measure nearly the same thing will each show a non-significant p-value while the model as a whole is strongly significant. The collinearity table explains why. **Forgetting that modules change.** The core analyses are stable. Third-party modules update on their own schedule and an option can be renamed or moved between versions. Record the jamovi version and the module version in your lab notebook. **Assuming the software validates the design.** jamovi will happily compute a t-test on 3 observations per group, or a regression with 8 predictors and 12 cases. The program does not warn you about underpowered designs or confounding. Those judgments belong to you and, for human or animal research, to your institutional review board or ethics committee. ## Frequently Asked Questions ### Is jamovi free to use? Yes. jamovi is free and open source, with no license, subscription or account required, including for commercial and academic work. ### How do I download and install jamovi? Go to the official jamovi project website, choose the installer for Windows, macOS or Linux, and run it. The install takes a few minutes and needs no registration. ### How do I run a t-test in jamovi? Click the T-Tests button on the Analyses tab, choose Independent, Paired or One Sample, then drag your outcome and grouping variables into the slots. Options let you add effect sizes, descriptives and assumption checks. ### How do I run an ANOVA in jamovi? Click ANOVA and pick One-Way, Factorial or Repeated Measures. Assign the outcome to Dependent Variables and the grouping factors to the factor slots, then open Post Hoc Tests and Assumption Checks. ### How do I run a regression in jamovi? Click Regression, then Linear or Logistic. Put the outcome in Dependent Variable and the predictors in Covariates or Factors, then tick Model Fit, Model Coefficients and Collinearity Statistics in the options. ### Do I need to know R to use jamovi? No. Every analysis can be run entirely through menus. R knowledge only becomes useful if you want to write custom analyses through the Rj Editor module. ### Can jamovi open SPSS or Excel files? Yes. jamovi imports .sav, .dta, .csv, .xlsx and plain text files directly, and saves its own work in the .omv format. ### Is jamovi suitable for a thesis or a publication? Yes, for standard analyses including t-tests, ANOVA and regression. Report the version number in your methods section and keep the .omv file so a reviewer or supervisor can reproduce every table. ## Sources 1. [Digital radiography and GIMP software in mandibular sex estimation: implications for forensic anthropology.](https://pubmed.ncbi.nlm.nih.gov/42227786/) 2. [Comparison the Effects of Stimulating the Sucking and Grasp Reflexs on Pain and Crying Duration Induced by Blood Sampling in Neonates.](https://pubmed.ncbi.nlm.nih.gov/42909970/) 3. [Blended learning in laboratory science education: comparison of flipped classroom and traditional lectures in immunohematology training.](https://pubmed.ncbi.nlm.nih.gov/42897555/) 4. [Decoding free-throw accuracy: a 2D biomechanical comparison of shooting mechanics in athletes and novice basketball players.](https://pubmed.ncbi.nlm.nih.gov/42358510/) 5. [Comparison of Microtensile Bond Strengths of Giomers Using HEMA-Based and HEMA-Free Silane Adhesive With Laser-Cut and Bur-Cut Dentin.](https://pubmed.ncbi.nlm.nih.gov/42689153/) 6. [Comparison of the accuracy of guided implant surgery between two implant-planning software: a retrospective cohort study.](https://pubmed.ncbi.nlm.nih.gov/41657546/) 7. [Assessing some low- and no-cost image processing software for gunshot residue pattern analysis.](https://pubmed.ncbi.nlm.nih.gov/42425613/) ## Related Articles - [JASP Tutorial: Free Statistics Software for t-Tests, ANOVA and Bayesian Analysis](/blog/data-analysis/jasp-statistics-tutorial) - [GraphPad Prism Tutorial: t-Tests, ANOVA, Curve Fitting and Graphs for Biologists](/blog/data-analysis/graphpad-prism-tutorial) - [Two Sample t-Test: Formula, Calculation and Example](/blog/data-analysis/two-sample-t-test-formula-example) - [ANOVA Table Explained: Components, Formulas and Example](/blog/data-analysis/anova-table-explained) - [What Is Hypothesis Testing? Steps, Errors and Examples](/blog/data-analysis/what-is-hypothesis-testing)