---
title: Understanding Percentage Types
category: Pivot Chart and Table
slug: percentage-types
blurb: "Learn about the three different percentage calculations in AddMaple tables: row percentages, column percentages, and sample percentages."
order: 8
---

# Understanding Percentage Types

AddMaple provides three different ways to calculate percentages in [pivot tables and charts](../frequently-asked-questions/how-to-pivot-your-data). Each percentage type answers a different question about your data and helps you understand relationships from different perspectives.

## Working with Percentages in AddMaple

### Percentage Base Selection (Charts)

When viewing a pivot chart, you can choose which percentage base to display using the **Percentage Base** control in the left menu:

- **Column A** - Shows row percentages (how each row distributes across columns)
- **Column B** - Shows column percentages (how each column is composed of rows)
- **All Rows** - Shows sample percentages (percentage of the entire dataset)

This control appears when you switch from **Count** to **Percentage** mode in the chart menu.

### Cell Tooltips (All Views)

When you hover over any cell in a chart or table, AddMaple shows a detailed tooltip that includes:

- The primary metric you've selected (count or percentage)
- **All three percentage types** (row, column, and sample percentages)
- The underlying counts and denominators
- [Statistical significance](../stats/significance-testing) indicators (when enabled)

This lets you quickly see the data from multiple perspectives without changing your view settings.

### Column Visibility (Tables)

In pivot tables, you can customize which columns to display:

- Show or hide individual percentage columns (row %, column %, sample %)
- Show or hide count columns (results, rows, weighted values)
- Toggle statistical columns ([z-scores, p-values](../stats/significance-testing))

Use the column visibility controls in the table header to customize your view.

## The Three Percentage Types

### Row Percentage (Column A Percent)
**What it shows:** The percentage of each row's total that falls into each column.

**Calculation:** `(cell count ÷ row total) × 100`

**When to use:** When you want to see how a specific group (row) distributes across different categories (columns).

**Example:** If you're looking at "Company vs Rating" data:
- Apple row: 36 people rated 1, 33 rated 2, etc.
- Row percentage shows: "Of all Apple respondents, 5.3% gave a rating of 1"
- All row percentages for Apple will add up to 100%

### Column Percentage (Column B Percent)  
**What it shows:** The percentage of each column's total that comes from each row.

**Calculation:** `(cell count ÷ column total) × 100`

**When to use:** When you want to see how each category (column) is composed across different groups (rows).

**Example:** In the same "Company vs Rating" data:
- Rating "1" column: 36 from Apple, 25 from Google
- Column percentage shows: "Of all people who rated 1, 59.0% were Apple users"
- All column percentages for rating "1" will add up to 100%

### Sample Percentage
**What it shows:** The percentage of the entire sample that falls into each cell.

**Calculation:** `(cell count ÷ total sample) × 100`

**When to use:** When you want to see the overall distribution across your entire dataset.

**Example:** In the same data:
- Apple rating "1" cell: 36 people out of 2,371 total
- Sample percentage shows: "1.5% of all respondents were Apple users who rated 1"
- All sample percentages will add up to 100%

## Visual Example

Here's how the same data looks with different percentage types:

| Company | Rating | Count | Row % | Column % | Sample % |
|---------|--------|-------|-------|----------|----------|
| Apple   | 1      | 36    | 5.3%  | 59.0%    | 1.5%     |
| Apple   | 2      | 33    | 4.8%  | 51.6%    | 1.4%     |
| Google  | 1      | 25    | 1.5%  | 41.0%    | 1.1%     |
| Google  | 2      | 31    | 1.8%  | 48.4%    | 1.3%     |

## When to Use Each Type

### Use Row Percentages When:
- **Comparing groups:** "How do Apple users rate compared to Google users?"
- **Understanding group behavior:** "What percentage of Apple users gave each rating?"
- **Identifying patterns within groups:** "Do different companies have different rating distributions?"

### Use Column Percentages When:
- **Understanding category composition:** "Who makes up each rating level?"
- **Market share analysis:** "What percentage of high ratings come from each company?"
- **Identifying dominance:** "Which company dominates the low ratings?"

### Use Sample Percentages When:
- **Overall distribution:** "What's the general pattern across all respondents?"
- **Market sizing:** "What percentage of the total market does each segment represent?"
- **Absolute impact:** "How significant is each cell in the overall dataset?"

## Row and Column Totals

### Row Totals
- **Count:** Sum of all counts in the row
- **Row %:** Always 100% (by definition)
- **Column %:** Shows sample percentage (since column percentages don't sum meaningfully)
- **Sample %:** Sum of sample percentages in the row

### Column Totals  
- **Count:** Sum of all counts in the column
- **Row %:** Shows sample percentage (since row percentages don't sum meaningfully)
- **Column %:** Always 100% (by definition)
- **Sample %:** Sum of sample percentages in the column

## Three-Way Tables

For three-way tables (e.g., Company × Rating × Gender), the percentage calculations work the same way, but with additional complexity. You can create three-way tables by adding multiple pivots or by [grouping columns](grouping-columns) together.

- **Row %:** Percentage within each company-gender combination
- **Column %:** Percentage within each rating level
- **Sample %:** Percentage of the entire sample

The totals follow the same logic as two-way tables, with row totals showing sample percentages for column percentages and vice versa.

## Practical Tips

- **Use row percentages** to compare groups
- **Use column percentages** to understand category composition  
- **Use sample percentages** to see overall distribution
- **Always show counts alongside percentages** for context

## Weighting and Percentages

When you apply a [weight column](../preparation/weighting) to your data, AddMaple automatically adjusts all percentage calculations to reflect the weighted sample.

### How Weighting Affects Calculations

**Weighted vs Unweighted:**
- **Unweighted percentages:** Based on raw response counts
- **Weighted percentages:** Based on weighted response counts that reflect your target population

**Example with Weighting:**
- Raw data: 100 Apple users, 200 Google users
- Weighted data: 150 Apple users, 150 Google users (after applying demographic weights)
- All percentages (row, column, sample) will be calculated using the weighted counts

### Weighted Totals

When weighting is active, you'll see both unweighted and weighted totals in the status line:
- **Unweighted:** "Showing 1,431 results from 870 rows"
- **Weighted:** "Weighted: 1,520.5 results from 905.2 rows"

The percentage calculations use the weighted totals, ensuring your analysis reflects the true population distribution.

## Multiselect Questions and Rows vs Results

[Multiselect questions](../data-types/multi-select) allow respondents to choose multiple options, which creates an important distinction between **rows** and **results**.

### Understanding Rows vs Results

**Rows (Respondents):**
- Number of people who answered the question
- Each person counts as one row, regardless of how many options they selected

**Results (Answers):**
- Total number of individual selections made
- Each selected option counts as one result

### Multiselect Example

If 100 people answer "Which brands do you use?" and can select multiple:
- **50 people** select only "Apple"
- **30 people** select both "Apple" and "Google"  
- **20 people** select only "Google"

**Counts:**
- **Rows:** 100 (number of respondents)
- **Results:** 150 (50 + 60 + 20 individual selections)

### How This Affects Percentages

**Row Percentages (Column A Percent):**
- Based on the number of respondents in each row
- "Of all Apple users, what percentage selected each option?"

**Column Percentages (Column B Percent):**
- Based on the total number of times each option was selected
- "Of all 'Apple' selections, what percentage came from each user group?"

**Sample Percentages (Sample Percent):**
- Based on the total number of selections across all respondents
- "What percentage of all selections were 'Apple' from each group?"

### Status Line with Multiselect

The status line shows both counts:
```
Showing 1,431 results from 870 rows
```

This tells you:
- 870 people answered the question (rows)
- They made 1,431 total selections (results)
- Results exceed rows because people could select multiple options

### Hiding Multiselect Data

When you [hide categories](filtering-vs-hiding) in multiselect data:
- **Empty-like categories** (EMPTY, ".") hide entire rows
- **Other categories** only hide the matching results, not the rows

Example status line:
```
Showing 1,411 results from 870 rows
20 results hidden
• None of the above — 20 results
```

This means 20 individual selections were hidden, but the 870 respondents still count toward the row base.

## Key Points

- Row percentages show how each group distributes across categories
- Column percentages show how each category is composed across groups
- Sample percentages show the overall distribution in your dataset
- Each percentage type answers a different research question
- In charts, use the Percentage Base control to select which percentage to display
- Hover over any cell to see all three percentage types in the tooltip
- In tables, customize which percentage columns are visible
- Row and column totals use sample percentages when the other percentage type doesn't sum meaningfully
- Choose your percentage type based on what you want to learn from your data
- Weighting adjusts all calculations to reflect your target population
- Multiselect questions create a distinction between respondents (rows) and individual selections (results)
- Understanding rows vs results is crucial for interpreting multiselect data correctly