Statsly
Tutorial library

Learn by doing

Short, reproducible workflows built around research decisions rather than software features.

From import to a defensible result

Import a dataset, review quality flags, explore variables, and run a recommended analysis.

12 minStart
  1. 1Load the inference study from the example gallery.
  2. 2Review Data and create a version only when changes are staged.
  3. 3Explore the outcome and group variables, then continue to Analyze.
  4. 4State the research question, confirm the recommended method, and inspect assumptions.

Auditable data cleaning

Stage replacements and preparation rules, preview the diff, and read dataset lineage.

10 minStart
  1. 1Open a column filter and stage a replacement.
  2. 2Inspect the staged-change summary and before/after preview.
  3. 3Create a named version and inspect its parent and operations.

Reliability and exploratory factors

Use scale items for Cronbach's alpha and exploratory factor analysis.

15 minIntermediate
  1. 1Load the Scale reliability example.
  2. 2Inspect item distributions and correlations in Explore.
  3. 3Run reliability first, then EFA as a separate analysis.

Time-to-event analysis

Compare survival curves and adjust hazard ratios with Cox regression.

18 minAdvanced
  1. 1Load the Survival follow-up example.
  2. 2Assign time, event, and treatment variables.
  3. 3Compare Kaplan-Meier curves before adding covariates in Cox regression.

Check whether cleaning changes conclusions

Re-run one specification across versions and compare effect, uncertainty, and sample.

14 minIntermediate
  1. 1Create at least two dataset versions.
  2. 2Open a completed result and choose version comparison.
  3. 3Document whether direction, magnitude, or conclusion changes.

Agreement between raters and measurements

Choose Cohen's kappa, Fleiss' kappa, or ICC for paired ratings and repeated measurements.

12 minIntermediate
  1. 1Load the Scale reliability example and inspect the rater columns.
  2. 2Use Cohen's kappa for two categorical raters or Fleiss' kappa for three or more.
  3. 3Use ICC for numeric measurements and interpret the interval with the measurement scale.

Test practical equivalence

Use TOST to distinguish evidence of equivalence from a non-significant difference.

10 minAdvanced
  1. 1Define the smallest meaningful difference before examining group results.
  2. 2Select a numeric outcome, a two-level group, and enter the margin in outcome units.
  3. 3Conclude equivalence only when both one-sided tests reject the equivalence bounds.

Interactions and indirect paths

Separate moderation from mediation and state the assumptions needed for each interpretation.

16 minAdvanced
  1. 1Use moderation when the association of one predictor may vary across another.
  2. 2Use mediation only when predictor, mediator, and outcome have a defensible temporal order.
  3. 3Interpret interaction or bootstrap indirect-effect tables before using causal language.