Descriptive Statistics
This is the comprehensive descriptive analysis tool. It computes a full descriptive analysis on your raw data — central tendency, dispersion, distribution shape, normality tests, charts, frequency tables — all from one input. It supports both single-variable analysis and a multi-variable comparison table.
Choose Analysis Type
When you open the test, you see a Choose Analysis Type screen with two cards:
Single Variable Analysis "Calculate descriptive statistics for one variable at a time"
Features:
- Detailed statistical measures
- Data type-specific analysis
- Normality tests
- AI-powered interpretation
- Interactive chat assistant
Multi-Variable Table "Create comparative tables for multiple variables"
Features:
- Side-by-side comparison
- Summary statistics table
- Multiple data types support
- Export functionality
- Batch processing
Single Variable Analysis
Single Variable Analysis is the standard mode — one variable at a time, with full descriptive output.
Provide your data
You can supply data in two ways:
| Method | How it works |
|---|---|
| Type or paste values | Free-text input where you paste numbers separated by commas, spaces, or new lines |
| Excel import | Upload an .xlsx / .xls / .csv file (this is the MyData capability). The module reads the sheets, lets you pick one sheet and one column, and converts the values |
When you import from Excel, the system automatically suggests a data type by analysing the column's contents (see Automatic Data-Type Recommendation below).
Choose the data type
A Select Data Type panel asks you to declare what kind of variable you have. There are six options, each with a short description and example:
| Type | Description | Example |
|---|---|---|
| Continuous | Measurable values (height, weight, temperature) | 1.75, 2.34, 5.67 |
| Discrete | Countable integers (number of children, number of cars) | 1, 2, 3, 4, 5 |
| Nominal (Categorical) | Unordered categories (gender, color, brand) | Male=1, Female=2 |
| Ordinal (Ranked categorical) | Ordered categories (education level, satisfaction) | Low=1, Medium=2, High=3 |
| Interval | Measurements with equal intervals (temperature °C, IQ) | 20°C, 25°C, 30°C |
| Binary | Data with two options (yes/no, success/failure) | 0, 1 or No=0, Yes=1 |
The data type choice matters: the test applies different statistics depending on what you pick. Continuous and interval data get mean / SD / skewness / kurtosis / normality tests; nominal gets a frequency distribution; binary gets the proportion of each value; ordinal gets median / mode and an optional mean (displayed with the note "Interpret with caution"); discrete gets the standard summary measures.
Calculate Statistics
Click Calculate Statistics. The system parses the values, validates them, and produces the Statistical Results panel.
Read the results
The output adapts to your declared data type. The full output may include any combination of:
Header section
- Column name (if imported from Excel) — "Data from: <column name>"
- Data type label
- Sample Size (n)
Central Tendency
- Mean (continuous, discrete, interval, ordinal-optional)
- Median (continuous, discrete, interval, ordinal)
- Mode — values that appear most often, or "No mode found" if every value is unique
For binary data, instead of these you see two cards:
- Proportion of "1" (Yes)
- Proportion of "2" (No)
Dispersion
- Standard Deviation
- Variance
- Range (max − min)
Distribution Shape (continuous-type data only)
- Skewness — labelled as Approximately symmetric / Positively skewed / Negatively skewed
- Kurtosis — labelled as Normal distribution / Heavy-tailed / Light-tailed
Normality Tests (continuous-type data only)
The module runs two normality tests and shows their statistics:
- Shapiro-Wilk Test — W statistic + p-value
- Kolmogorov-Smirnov Test — D statistic + p-value
A small interpretation tells you whether the data appears to follow a normal distribution.
Frequency Distribution (nominal, ordinal, binary, discrete)
A table listing every distinct value and how often it appears in the data — useful for understanding the spread of categorical data.
View the charts
A Descriptive Charts panel below the numerical results shows visualisations appropriate to the data type — for example, a histogram with a normal-distribution overlay for continuous data, or a bar chart of frequencies for categorical data.
Discuss the results with the chatbot
After the results, a UniversalChatBot is shown. It is configured with testType="DescriptiveStatistics" and the full results, your data type, and the imported column name. You can ask it questions like:
- "What does this skewness value mean for my analysis?"
- "Is my data normal enough to run a t-test?"
- "Should I transform this variable before regression?"
The chatbot's role is explanation and discussion of the results — the statistics themselves are already computed by the module.
Multi-Variable Table
Multi-Variable Table mode is built for producing a Table 1 — the comparative summary table that almost every research paper opens with, showing patient demographics or baseline characteristics side by side.
Two ways to load data
You can either:
- Generate a built-in sample dataset — pick from one of the Sample Scenarios (e.g. "Clinical Trial Demographics" with 50 patients including age, BMI, blood pressure, gender, and treatment group, or "Hospital Outcomes Study" with 40 patients including lab values, pain scores, length of stay, and discharge status). The system then generates the dataset for you and you can experiment immediately.
- Import your own Excel file (
.xlsx/.xls/.csv) — the importer reads all sheets, lets you pick which sheet to use, and then for each column lets you decide whether to include it and what data type it is. This is again the MyData capability.
Build your table
For each included column, the system computes the right summary statistics for that data type:
- Continuous / interval / discrete → mean, SD, median, range
- Binary / nominal / ordinal → counts and proportions per category
You can add columns one by one, remove columns you do not want, and rearrange them. The result is a clean summary table you can copy or export.
Export
The page includes an Export option built around SheetJS (xlsx library), so you can download the table as an Excel file ready to drop into your manuscript or report.
Universal Chat Bot
A UniversalChatBot is again available after the table is generated, so you can ask follow-up questions about what the numbers mean across groups, what to test next, or how to write up the table for a paper.
Automatic data-type recommendation (Excel imports only)
When you import a column from Excel, the system does not just take whatever you tell it — it actually inspects the values and recommends a data type for that column. A panel titled Recommended Data Type appears with:
| Element | What it shows |
|---|---|
| Recommended type | One of the six data types (Continuous, Discrete, Nominal, Ordinal, Interval, Binary) with its description |
| Confidence badge | Either High Confidence, Medium Confidence, Low Confidence, or Uncertain — indicating how sure the system is |
| Reason | A short explanation of why the system thinks this type fits |
| Warning (when applicable) | A yellow callout if the system spots something unusual that you should review |
| Accept This Recommendation button | One-click accept — sets the data type and moves you forward |
| Alternative Options | A list of other types that might also work, so you can override the recommendation |
This means you can rely on the system for fast typing while staying in control: if the recommendation does not fit your variable, just pick an alternative.