Features

Everything an empirical paper needs.

Modellum is one engine with two faces: a scripting language compatible with Stata's do-files, and a desktop app whose every menu builds the same commands.

Data management

  • use / save / import delimited / import spss / export delimited
  • generate, replace, egen, recode, rename, drop, keep, order
  • merge, append, collapse, contract, reshape long|wide
  • encode/decode, destring/tostring, duplicates
  • label variable/define/values/data
  • preserve/restore, by/bysort with _n and _N

Describe & tabulate

  • describe, list, count
  • summarize [, detail]
  • tabulate with χ², row/column/cell percentages
  • correlate, spearman, ktau

Linear & generalized models

  • regress with robust (HC1) and cluster-robust errors
  • logit/logistic, probit, poisson with exposure and IRRs
  • glm: gaussian, binomial, poisson, nbinomial, gamma, igaussian × 6 links
  • nbreg (NB2)
  • nl: nonlinear least squares with symbolic derivatives
  • Factor variables: i., ib#., c., #, ##

Inference after estimation

  • predict
  • test, testparm, lincom
  • margins: predictive margins, dydx(), at(), atmeans
  • marginsplot
  • contrast and pwcompare (Tukey, Bonferroni, Šidák, Scheffé)
  • estimates store/restore/table

ANOVA & multivariate

  • anova with Type III or sequential SS, nested and error terms
  • Repeated measures with Huynh–Feldt, Greenhouse–Geisser, Box
  • oneway with Bartlett's test
  • manova: Wilks, Pillai, Lawley–Hotelling, Roy
  • hotelling, mvtest means/covariances/correlations/normality

Survey data

  • svyset: strata, PSUs, fpc, singleunit
  • Replicate weights: bootstrap, BRR with Fay, jackknife
  • svy: mean, total, proportion, ratio, tabulate (Rao–Scott)
  • svy: regress, logit, probit, poisson
  • estat effects (design effects), margins after svy

Panel data & IV

  • xtset, xtreg fe/re/be, xtsum, hausman
  • ivregress 2sls, liml, gmm
  • estat firststage, overid, endog

Survival analysis

  • stset with delayed entry and multiple records
  • sts list/graph/generate/test, stsum
  • stcox with Breslow or Efron ties, strata, robust errors
  • estat phtest (Grambsch–Therneau)
  • streg: exponential, Weibull, Gompertz, lognormal, loglogistic

Measurement & scales

  • pca, factor (pf, ipf, pcf, ml)
  • rotate varimax/promax
  • estat kmo, estat smc, factor scores
  • alpha with item statistics

Nonparametric

  • ranksum with exact p-values
  • signrank, signtest, kwallis, median
  • Exact tests up to 200 observations

Programming

  • local/global macros, extended macro functions
  • foreach, forvalues, while, if/else
  • program define with syntax, args, marksample
  • r() and e() results, scalars, matrices
  • postfile/post, tempvar/tempname/tempfile
  • capture, quietly, noisily, return codes

Output

  • esttab to .docx, .tex, .md, .csv
  • summarize/tabulate/correlate export()
  • Graphs to SVG, PNG and PDF with embedded fonts
  • log using
Graphs

Publication graphics, one renderer.

The same drawing code makes the figure on screen and in the file, so exports match exactly on every machine. In the app, graphs follow the light or dark theme and show values on hover.

  • histogram, kdensity, graph box, violin
  • scatter with lfit/qfit/lowess and CI bands
  • line, twoway overlays, area, connected
  • graph bar/hbar with CIs, over() and nested over()
  • coefplot, marginsplot, corrplot, rvfplot, qnorm
  • over() to colour by group and by() for small multiples on every graph
How it compares

Modellum next to the incumbents

A broad-strokes comparison of approach, not a feature audit; both incumbents are mature products with capabilities Modellum doesn't have yet (see the roadmap).

ModellumStataSPSS
Point-and-click interface✓Dialogs✓
Scripting language✓✓✓
Explanations computed from your data✓——
Uses all CPU cores at no extra cost✓MP editionSome procedures
Open-source engine✓——
Roadmap

What's next

Next up

Bootstrap and jackknife, ml for custom likelihoods, SPSS writer, Excel import, more of the GPU.

For professionals

Multilevel models, dynamic panels, xtlogit/xtpoisson, ARIMA, multiple imputation, power analysis.

Depth

SEM, Bayesian estimation, quantile regression, a plugin API, and R/Python interop.