---
title: Exploratory data analysis (EDA)
description: "A statistical approach, methodology, inspired by John Tukey. Tukey was something of a polymath and I recommend the [Wikipedia article about him](https://en.wikipedia.org/wiki/John_Tukey) not least for the glorious first hand description of one of his lectures! I think I am pretty…"
url: "https://www.psyctc.org/psyctc/glossary2/exploratory-data-analysis-eda/"
updated: "2024-04-29"
category: All OM book glossary entries
---

# Exploratory data analysis (EDA)

A statistical approach, methodology, inspired by John Tukey. Tukey was something of a polymath and I recommend the [Wikipedia article about him](https://en.wikipedia.org/wiki/John_Tukey) not least for the glorious first hand description of one of his lectures! I think I am pretty much a follower of the EDA tradition these days but perhaps compromised a bit from his pure form of EDA by the pressures to publish in journals that continue to expect inferential statistics and "confirmatory" analyses.

#### Details

EDA overlaps with descriptive statistics and into the realm of statistical estimation, i.e. confidence intervals but it is apt that Tukey's term was "Data Analysis" not "statistics". I can't do better than quote from the Wikipedia article about EDA:

> The objectives of EDA are to:
>
>
>
> - Enable unexpected discoveries in the data
> - Suggest hypotheses about the causes of observed phenomena
> - Assess assumptions on which statistical inference will be based
> - Support the selection of appropriate statistical tools and techniques
> - Provide a basis for further data collection through surveys or experiments
>
>  [https://en.wikipedia.org/wiki/Exploratory_data_analysis#Development](https://en.wikipedia.org/wiki/Exploratory_data_analysis#Development)

Those strike me as exactly the objectives we should have for most of our quantitative research in the fields of MH, W-B and therapies, we shouldn't be just trying to come up with pseudo-certainties from systematic reviews of randomised controlled trials.

#### Try also

Confidence intervals
[Descriptive statistics](https://www.psyctc.org/psyctc/glossary2/descriptive-statistics/)
Estimation
[Null hypothesis significance testing (NHST) paradigm](https://www.psyctc.org/psyctc/glossary2/null-hypothesis-significance-testing-nhst-paradigm/)

#### Chapters

This approach runs through the entire OMbook but we didn't mention it explicitly partly to avoid getting into culture wars!

#### Online resources

Not yet.

#### Dates

First created 29.iv.24.