Research

Plan Your Data Analysis Pipeline with AI Mind Maps

Use AI mind maps to design analysis workflows, select statistical methods, and anticipate edge cases before touching your data. Branch into alternatives.

Plan your analysis See example ↓

Start from a template

Clone this skeleton and draft a defensible, pre-registered analysis plan before you touch your data.

Clone this template

How it works

Analysis planning is where studies succeed or fail. A pre-registered analysis plan prevents p-hacking, but writing one requires thinking through every branch of your data pipeline before you see the data. Mind maps make this planning process visual and exhaustive.

Describe your study design and data. Enter your research question, study type, and the variables you’re collecting. The AI outlines the major stages of an appropriate analysis pipeline.
Branch into methodological choices. Each analysis stage has decision points. Should you use listwise deletion or multiple imputation for missing data? Mixed-effects or GEE for repeated measures? Each choice becomes a branch with its assumptions and tradeoffs explained.
Map contingencies. Real data surprises you. Branch into “what if” scenarios: What if the normality assumption fails? What if there’s an unexpected interaction? What if dropout is non-random? Each contingency gets its own sub-tree with a planned response.
Export your decision tree. The completed map is a visual pre-registration. It documents not just what you planned to do, but what alternatives you considered and why you chose each path.

Why branching matters for data analysis planning

Linear analysis plans read like recipes: “First do X, then do Y.” But real analysis is a decision tree. The right method at step three depends on what you found at step two. If your residuals are non-normal, your entire downstream plan changes. A branching structure captures this conditional logic.

Branching also exposes the “garden of forking paths” — the many defensible choices a researcher makes during analysis. By mapping all the forks before touching the data, you can identify which choices actually matter for your conclusions and which are inconsequential. This makes your eventual analysis more transparent and your results more credible.

Example

You’re planning the analysis for a randomized controlled trial of a mindfulness intervention on chronic pain. The root branches into primary outcome analysis, secondary outcomes, and subgroup analyses. Under primary outcome, you fork into intention-to-treat versus per-protocol. Under intention-to-treat, you branch into how to handle participants who never started the intervention. You realize you need to pre-specify a complier-average-causal-effect analysis — something you would have missed in a linear plan.

Pair this workflow with research methodology to ensure your analysis plan aligns with your study design, or use hypothesis exploration to clarify which hypotheses your analysis needs to distinguish. For structuring your written analysis section, try essay outlining.

Now try it yourself

Plan a data analysis pipeline for a longitudinal study of cognitive decline in aging adults

The pipeline has four stages: data cleaning and missingness assessment, descriptive trajectory characterization, mixed-effects modeling of decline rates, and sensitivity analyses for dropout and practice effects.

What approaches handle informative dropout in this design?

Pattern-mixture models stratify by dropout timing. Shared-parameter models jointly model the outcome and dropout process. Multiple imputation under missing-not-at-random assumptions is a third option, though it requires explicit sensitivity parameters.