jamovi and JASP: free point-and-click statistics

Install, open a CSV and set variable types

Install

1. jamovi: download from jamovi.org/download. On Windows, run the installer; jamovi's installation guide says a .zip version exists for computers where IT policy blocks installers. On macOS, open the .dmg and drag jamovi to Applications. On Linux and Chromebooks it comes from Flathub. To skip installing, use jamovi Cloud in a browser.
2. JASP: download from jasp-stats.org/download. Windows users can choose the Microsoft Store (which updates itself) or an installer; macOS has separate downloads for Apple Silicon and Intel; Linux and ChromeOS use Flatpak. No internet connection is needed to run JASP once it is installed.

Open the course file

Download households.csv (240 households, 13 columns). Illustrative data, invented for teaching The district names are real places; every number is made up.

1. jamovi: click the file menu (☰) at the top left, choose Open, then This PC, and pick households.csv. The data appear as a spreadsheet on the left; results will appear on the right.
2. JASP: click the main menu (the three blue stripes), choose File, then Open, and browse to the file. JASP needs a header row naming each column, which this file has.

Check the variable types

Both programs guess a type for each column, and a wrong guess changes which analyses accept it. JASP's getting-started page lists three types (Nominal, Ordinal, Scale) and its rules: a numeric column with 3 to 10 distinct whole numbers becomes Ordinal. In our file hh_size runs from 1 to 8, so check whether it arrived as Ordinal and set it to Scale if you want its mean. In jamovi, double-click a column header (or press F3) to open the variable editor and set the measure type.

• district, caste, area and the Yes/No columns should be Nominal.
• monthly_pc_exp, land_acres and head_edu_years should be continuous (Scale).
• hh_id is an identifier. Leave it out of analyses.

Descriptives, split by group

1. jamovi: Analyses > Exploration > Descriptives. Drag monthly_pc_exp into Variables and area into Split by. Open the Statistics section for more statistics and Plots for box plots and histograms.
2. JASP: choose Descriptives on the ribbon, move monthly_pc_exp into the variables box and area into the split box.

The table should give N, mean, median, standard deviation, minimum and maximum for rural and urban households separately, with the two N values adding to 240. The R cell computes the same numbers:

In both groups the mean is above the median, which is the usual sign of a few large values pulling the mean up. A box plot from the Plots section will show them as points above the whisker.

Compare means: t-tests and one-way ANOVA

Independent samples t-test

1. jamovi: Analyses > T-Tests > Independent Samples T-Test. Put monthly_pc_exp in Dependent Variables and area in Grouping Variable. The grouping variable must have exactly two levels.
2. JASP: T-Tests on the ribbon, then the classical Independent Samples T-Test, with the same two variables.

Read the t statistic, its degrees of freedom and the p-value. Tick Welch's as well: it does not assume the two groups have equal variances, and urban and rural expenditure rarely do. The R cell runs both:

On this file R gives t = −9.84 on 238 degrees of freedom for Student's test. jamovi should show the same size of t; its sign depends on which group jamovi lists first.

One-way ANOVA

1. jamovi: Analyses > ANOVA > One-way ANOVA. Put monthly_pc_exp in Dependent Variable and caste in the grouping box, then choose equal or unequal variances. Tick Equality of variances for Levene's test.
2. JASP: ANOVA on the ribbon, then the classical ANOVA.

R gives Welch's F = 7.84 (3 and 89.2 degrees of freedom) and Fisher's F = 4.63 (3 and 236). Check that jamovi's One-Way ANOVA table matches the version you chose.

The ANOVA asks whether mean expenditure differs across the four caste groups at all. It does not say which groups differ; for that, ask for post-hoc tests in the same dialog.

Contingency tables, correlation and regression

Contingency tables

1. jamovi: Analyses > Frequencies > Independent Samples (χ² test of association). Put caste in Rows and has_toilet in Columns. Under Cells, tick row percentages.
2. JASP: Frequencies > Contingency Tables, with the same rows and columns.

Correlation

1. jamovi: Analyses > Regression > Correlation Matrix. Move monthly_pc_exp, head_edu_years, hh_size and land_acres into the box. jamovi shows each coefficient once, below the diagonal, with stars for significance.
2. JASP: Regression > Correlation.

Linear regression

1. jamovi: Analyses > Regression > Linear Regression. Put monthly_pc_exp in Dependent Variable, head_edu_years in Covariates (continuous) and caste and area in Factors (categorical). jamovi makes the dummy variables for you.
2. JASP: Regression > Linear Regression.

In the Model Coefficients table, each caste row is the difference from the reference level (General, if you set it first in module 2) for households with the same schooling and area. The R cell fits the same model; your estimates should match its Estimate column, and R² should match too.

jamovi syntax mode, Rj, and JASP's R button

Both programs run R to produce their tables, and both will show you the R. This is the bridge from clicking to coding.

jamovi syntax mode

1. Click the application menu (⋮) at the top right of jamovi and tick Syntax mode.
2. Every analysis now shows the R call that produces it, using the jmv R package. Right-click the syntax to copy it.
3. Untick it to leave syntax mode.

The calls look like the lines below, which follow the examples in jamovi's jmv reference. They are code to read, written for our households file; the exact options jamovi writes depend on what you ticked.

jmv::descriptives(data = data, vars = vars(monthly_pc_exp), splitBy = "area")
jmv::ttestIS(formula = monthly_pc_exp ~ area, data = data)
jmv::anovaOneW(formula = monthly_pc_exp ~ caste, data = data)
jmv::contTables(data = data, rows = "caste", cols = "has_toilet", pcRow = TRUE)
jmv::corrMatrix(data = data, vars = vars(monthly_pc_exp, head_edu_years, hh_size))
jmv::linReg(data = data, dep = monthly_pc_exp, covs = vars(head_edu_years),
            factors = vars(caste, area),
            blocks = list(list("head_edu_years", "caste", "area")))

The syntax does not include the step that reads the data. In R you would read the CSV first, install jmv with install.packages("jmv"), and the same calls give the same tables.

Rj: write R inside jamovi

1. Click Modules at the top right, choose jamovi library and install Rj.
2. Choose Rj Editor from the new R icon in the Analyses ribbon.
3. Your open dataset is available as a data frame called data. Type R code and run it with the green triangle, or Ctrl+Shift+Enter (⌘+Shift+Enter on a Mac). The results appear in the results panel like any other analysis.

JASP's R button

From JASP 0.17, core analyses carry an R button. Click it after setting your options and JASP shows the R function call behind the analysis, with its settings. You can paste that code into another JASP to reproduce the analysis, or send it to a colleague so they can see exactly which options you used.

JASP's Bayesian analyses

JASP's features page lists Bayesian versions of the t-tests, ANOVA (including repeated measures and ANCOVA), correlation, linear and logistic regression, binomial and multinomial tests, contingency tables and log-linear regression. In each ribbon menu the Bayesian analyses sit beside the classical ones.

What a Bayes factor says

A p-value asks how surprising the data would be if there were no difference. A Bayes factor compares two hypotheses directly: BF₁₀ is how many times better the data are predicted by the alternative (a difference) than by the null (no difference). BF₁₀ = 5 means the data are five times as likely under the alternative; BF₁₀ = 0.2 means five times as likely under the null. That second reading is something a p-value cannot give you: evidence for no difference, which matters when a programme report needs to say two districts did not differ.

1. In JASP, open T-Tests and choose the Bayesian Independent Samples T-Test.
2. Put monthly_pc_exp in the variable box and area as the grouping variable.
3. Read BF₁₀ in the results table, and note the prior shown in the options panel; the Bayes factor depends on it.
4. Repeat with head_gender as the grouping variable and compare the two Bayes factors.

Save, export, and when to move to R

Save and share

• jamovi saves data, analyses and results together in one .omv file that a colleague can reopen. Right-click any table or plot to copy it, APA-formatted, into Word or an email. Each analysis has an annotation space for your interpretation, so the output can double as a draft write-up. jamovi's release notes record export to Excel (.xlsx) in 2.7.12, LibreOffice (.ods) in 2.7.13, LaTeX in 2.7.14, and Word (.docx) and .odt in 28.4.
• JASP saves a .jasp file holding the data, the analyses and your notes. Its How to use JASP page shows how to export results to HTML, copy tables straight into a word processor, and copy tables as LaTeX.
• Templates for routine data. jamovi can import a new data file into an existing .omv (from the ☰ menu): the old rows are replaced, columns are matched by name, and every filter, computed variable and analysis updates. For a monthly monitoring dataset with the same columns each month, that turns your analysis into a reusable template.

When to move to R

jamovi and JASP are enough for a lot of evaluation work. Move to R (or Stata) when you meet one of these:

• Joining files. jamovi's documentation says that "at present, there is no way to combine files horizontally": you cannot merge a household file with a district file. R does it in one line.
• Survey designs. NFHS, PLFS and most large household surveys need weights, strata and clusters for correct standard errors. R's survey package, Stata's svy and SPSS Complex Samples handle that.
• Repeating the same work across many files or rounds, where a script beats clicking.
• Analyses with no menu, such as difference-in-differences with fixed effects or small-area estimation. Rj can run some of these inside jamovi, but at that point you are writing R anyway.

Where next