Home / Statistical Tools / Analysis Tools / Confidence Interval / Preparing Your Data
Preparing Your Data¶
Confidence Interval analyzes numeric data columns to calculate interval estimates. This page explains data requirements for each CI type.
When to Use Each Type¶
Four CI types are available from the Excel ribbon dropdown:
| Type | Data Required | When to Use |
|---|---|---|
| CI for the Mean | Numeric (Continuous, Integer, Count) | Estimate where the true population mean lies |
| CI for Standard Deviation | Numeric (Continuous, Integer, Count) | Estimate the range of true process variability |
| CI for Binomial Proportion | Binary categorical (Nominal with exactly 2 values, or numeric with 2 values) | Estimate proportions for binary outcomes (Pass/Fail, Yes/No) |
| CI for Poisson Rate | Count data (Integer, Count) | Estimate the rate of occurrence for event counts |
Data Type Restrictions by CI Type
- Nominal (text) columns are only allowed for CI for Binomial Proportion. The data must have exactly 2 unique values.
- Continuous columns are only allowed for CI for the Mean and CI for Standard Deviation. Poisson Rate requires Integer or Count data.
How Data Types Work¶
The table below shows what happens when you assign each column type in the Confidence Interval dialog. Click a numbered badge to see an example using that configuration.
| Data Column | ||||||
|---|---|---|---|---|---|---|
| Nominal | Continuous | Integer | Count | DateTime | ||
| Frequency Column | None | Proportion only. Requires exactly 2 unique values. Each value gets its own CI bounds. 3 | Mean and StdDev only. Each column is one CI calculation. Multiple columns produce multiple whiskers on the same chart. 12 | Each column is one CI calculation. Multiple columns produce multiple whiskers on the same chart. 4 | Each column is one CI calculation. Multiple columns produce multiple whiskers on the same chart. | Not Allowed |
| Nominal | — | |||||
| Continuous | — | |||||
| Integer | — | |||||
| Count | Proportion only. Expand dataset by count. | Mean and StdDev only. Expand dataset by count. | Expand dataset by count | Expand dataset by count | Not Allowed | |
| DateTime | — | |||||
— indicates this frequency type is not available for selection
Examples¶
1 Quick Start — CI for the Mean — Confidence interval for population mean
2 CI for Standard Deviation — Confidence interval for population standard deviation
3 CI for Binomial Proportion — Confidence interval for binary outcome proportions
4 CI for Poisson Rate — Confidence interval for rate of occurrence
Common Mistakes¶
Avoid These Issues
- Proportion requires exactly 2 unique values — If your data column has fewer than 2 or more than 2 unique values, the Binomial Proportion CI will report "requires 2 unique values" instead of producing an interval.
- Minimum 2 data points for Mean and StdDev — CI for the Mean and CI for Standard Deviation require at least 2 observations.
- Forgetting that frequency expands the dataset — When using a Count frequency column, the data is expanded before the CI is calculated.
- Missing data — Rows with empty or non-numeric values are excluded from the calculation.