Data Visualization Example: How to use a column-structured table to generate an estimation plot?

Feb 28, 2025

5 min read

Welcome to the AI and Statistics series! Let's dive into how AI can transform tabular data into various types of charts.

Today, we will be using a Column structure table to generate an Estimation Plot.

In statistical analysis, differences between two sets of data are often analyzed using repeated t-tests, where the mean of each test serves as the height of the bar in a bar graph, and the error bar height can be represented using "standard deviation (SD)" or "standard error of the mean (SEM)."

There are instances where inconsistencies arise, such as:

Scenario 1: A large mean difference between two groups, yet the t-test P-value is not significant (ns).

Scenario 2: A small mean difference between two groups, yet the t-test P-value is highly significant**.

These inconsistencies may occur because the "two sample sets" have differences, but the "mean difference" and "standard deviation" levels are quite similar.

We know that the Estimated Mean fluctuates around the true score (T), and fewer experimental repetitions can result in larger estimation deviations.

Typically, in Scenario 1, we increase the sample size.

In Scenario 2, we apply multiple hypothesis testing correction, calculating FDR⁺ or q-value⁺.

However, due to the limitations of P-values, the framework for P-value calculation revolves around "null-hypothesis significance testing (HNST)," focusing on "whether there is a difference." Most scientific conclusions cannot be crudely interpreted in a binary fashion.

An article published in 2019 in Nature Methods, titled "Moving beyond P values: data analysis with estimation graphics," provides a method for visualizing experimental data from the perspective of Estimation statistics: the Estimation Plot.

With Estimation Plot visualization:

The left part shows two groups of control and experimental data, transforming traditional bar graphs (height as mean) with error bars into scatter plots, allowing observation of the general distribution pattern of each data point.
The right part presents estimated values or statistics, such as means, medians, and their confidence intervals.

In this case, the effect size used is the mean difference, with two horizontal lines representing the average values of the two groups. The distance between these lines indicates the effect size.

For clarity, dual Y-axes are used, aligning the '0' scale on the right Y-axis with the mean of the control group on the left Y-axis.

The line segment on the right part represents the range of this effect size distribution, where a 95% confidence interval shows that its true value (T) falls within the segment range.

This approach allows us to incorporate effect size alongside traditional statistics like mean, P Value, SD, and SEM, thus mitigating the risks of relying solely on a single statistical measure.

Today, we are going to use non-paired t-test and Estimation Plot to compare the study time distribution between Male and Female groups from our data source.

Our aim is to visualize these differences accurately, providing a clear and concise comparison.

This approach is particularly useful for understanding group differences in study times.

Don't worry about using the AI Agent-driven Bayeslab, all you need is natural language to get the data analysis result.

All content will be explained in the most comprehensible natural language descriptions to help you get started with data analysis from scratch.

We’ll start with a data table featuring two columns: "Female" and "Male", each containing individual data points representing study hours for females and males, respectively.

This Estimation Plot will help us visualize differences between groups and assess their distributions, unlike conventional plots, it highlights estimation precision and group comparisons effectively.

We’ll delve into how these prompts influence the final charts and uncover techniques for effective data visualization. In just 2 minutes, you'll learn to visualize and compare group differences using an Estimation Plot. Let's start it right now.

Using different prompt inputs, we'll demonstrate how AI generates estimation plots.

Our steps will include:
Step 1 - Perform a Normality Test to check if the data follows a normal distribution.
Step 2 - Conduct an Unpaired t-test to compare the means of the two groups.
Step 3 - Generate a Conventional Scatter Plot to visualize the control and treatment groups.
Step 4 - Create an Initial Drawing of the Estimation Plot.
Step 5 - Optimize the Estimation Plot by making stylistic adjustments.
Step 6 - Finalize the Estimation Plot by adjusting the vertical line positions for clarity.