What Is Relative Frequency?
Relative frequency is the proportion of all observations that fall into a particular value, category or class. Calculate it by dividing the frequency by the total number of observations: RF = f / n. For example, if 8 of 20 students choose the bus, the relative frequency is 8/20 = 0.40, or 40%.
The key word is proportion. While frequency tells you how many times something occurred, relative frequency tells you what share of the total that count represents. A frequency of 40 means nothing in isolation — 40 out of 50 is very different from 40 out of 4,000. Relative frequency provides that context by normalizing the count against the total.
Relative frequency can be expressed three ways — all correct, all describing the same share:
0.40 and 40% represent the same relative frequency. Avoid writing 0.40% — that equals 0.004 as a proportion, which is a completely different value. Always be clear about which form you are using.
Relative frequency is a core concept in descriptive statistics. It appears in frequency distributions, histograms, cumulative plots, two-way tables, and forms the empirical foundation for probability estimation.
Relative Frequency Formula
RF = relative frequency
f = frequency (count) of the category
n = total number of observations
To convert relative frequency to a percentage, multiply by 100:
Two useful reverse calculations follow directly from the same formula:
| If you know… | Rearrange to find… | Formula |
|---|---|---|
| RF and n | Frequency f | f = RF × n |
| f and RF | Total n | n = f / RF |
Reverse example: If the relative frequency of a category is 0.35 and the sample size is 200, the count is 0.35 × 200 = 70. If 15 observations have a relative frequency of 0.30, the total is 15 / 0.30 = 50.
How to Find Relative Frequency
The calculation takes three steps for any category:
Count the observations in that category (f)
Tally how many times the value, category or class appears in your dataset. This is the raw frequency.
Count the total number of observations (n)
Add up all observations across every category. Make sure n matches the number of data points relevant to your question — not a different group, not including excluded observations.
Divide f by n
RF = f / n. Convert to a percentage by multiplying by 100 if needed. Check your work: for mutually exclusive, exhaustive categories, all relative frequencies should sum to 1 (allowing for minor rounding differences).
12 red marbles in a bag of 30 marbles. What is the relative frequency of red?
f = 12 (count of red marbles)
n = 30 (total marbles)
RF = 12 / 30 = 0.40 = 40%
✅ The relative frequency of red marbles is 0.40 (fraction: 12/30, decimal: 0.40, percentage: 40%).
Relative Frequency Example — Categorical Data
A survey asks 20 students how they travel to school. The results are recorded below. This is a straightforward categorical dataset — a good starting point for building a relative frequency table from scratch.
| Travel Method | Frequency (f) |
|---|---|
| Walk | 5 |
| Bus | 8 |
| Car | 4 |
| Bicycle | 3 |
| Total (n) | 20 |
Applying RF = f / n to each row:
- Walk: 5 / 20 = 0.25 = 25%
- Bus: 8 / 20 = 0.40 = 40%
- Car: 4 / 20 = 0.20 = 20%
- Bicycle: 3 / 20 = 0.15 = 15%
Verification: 0.25 + 0.40 + 0.20 + 0.15 = 1.00. Sum checks out — every student belongs to exactly one travel category.
How to Make a Relative Frequency Table
A relative frequency table adds a column (or two) beside the raw frequency count to show each category's share. The completed table for the student travel survey:
| Travel Method | Frequency (f) | Relative Frequency (RF) | Percent (%) |
|---|---|---|---|
| Walk | 5 | 0.25 | 25% |
| Bus | 8 | 0.40 | 40% |
| Car | 4 | 0.20 | 20% |
| Bicycle | 3 | 0.15 | 15% |
| Total | 20 | 1.00 | 100% |
When one category has a relative frequency of exactly 1/3, the decimal value is 0.3333… Rounded to two decimal places in three equal groups, you may see 33.3% + 33.3% + 33.3% = 99.9%. This is a rounding artifact, not a calculation error. Keep more decimal places internally if precision matters.
Frequency vs Relative Frequency
The distinction sounds simple but matters in practice. Frequency is an absolute count; relative frequency is a normalized proportion. When two groups have different sizes, raw counts can mislead.
Example: School A has 30 students who ride the bus. School B has 40 students who ride the bus. Raw count alone puts School B ahead. But School A has 50 students total (bus RF = 60%) while School B has 100 students total (bus RF = 40%). School A's bus ridership rate is actually 50% higher.
| Feature | Frequency | Relative Frequency |
|---|---|---|
| What it measures | Raw count | Share of total |
| Formula | Count directly | f / n |
| Units | Number of observations | Proportion or percent |
| Example | 8 | 0.40 or 40% |
| All categories sum to | n | 1 or 100% |
| Best for | Actual counts needed | Comparing different-sized groups |
Relative frequency normalizes counts by n, but it does not make small and large samples equally reliable. A proportion of 0.40 from a sample of 10 carries far more uncertainty than 0.40 from a sample of 10,000. For inference, you still need standard errors and confidence intervals.
Relative Frequency Distribution
A relative frequency distribution shows the proportion of observations in every category or class. It is simply a frequency distribution with proportions in place of counts — same categories, different scale.
The two distributions carry the same basic pattern. For identical categories, one is a rescaled version of the other. The difference shows up in the y-axis: a frequency distribution lists counts; a relative frequency distribution lists proportions or percentages.
Numerical Example
Dataset: 2, 2, 3, 3, 3, 4, 4, 5, 5, 5 (n = 10)
| Value | Frequency (f) | Relative Frequency (RF) | Percent |
|---|---|---|---|
| 2 | 2 | 0.20 | 20% |
| 3 | 3 | 0.30 | 30% |
| 4 | 2 | 0.20 | 20% |
| 5 | 3 | 0.30 | 30% |
| Total | 10 | 1.00 | 100% |
Grouped Relative Frequency
When data are continuous or span a wide range of values, they are organized into class intervals. The same RF = f / n formula applies — divide each class count by the total.
Example: 20 exam scores grouped into five-point intervals.
| Score Interval | Frequency (f) | Relative Frequency | Percent |
|---|---|---|---|
| 50 ≤ x < 60 | 2 | 0.10 | 10% |
| 60 ≤ x < 70 | 5 | 0.25 | 25% |
| 70 ≤ x < 80 | 6 | 0.30 | 30% |
| 80 ≤ x < 90 | 5 | 0.25 | 25% |
| 90 ≤ x ≤ 100 | 2 | 0.10 | 10% |
| Total | 20 | 1.00 | 100% |
Class intervals use non-overlapping notation (50 ≤ x < 60) to ensure each observation belongs to exactly one class. The Frequency Distribution Table Maker can build these grouped distributions from raw numerical data automatically.
Relative Frequency as a Percentage
A percentage is relative frequency on a 0–100 scale. The two forms are numerically equivalent: RF = 0.73 equals 73%. Converting between them:
- Decimal to percent: multiply by 100 → 0.42 × 100 = 42%
- Percent to decimal: divide by 100 → 42% / 100 = 0.42
- Fraction to percent: divide then multiply → (7/28) × 100 = 25%
Some textbooks use the term "percentage frequency" or "percent relative frequency." These all refer to the same quantity expressed in the 0–100 range. The choice of form depends on context — proportions work better in formulas; percentages work better in tables and charts aimed at general audiences.
Relative Frequency Histogram
A relative frequency histogram is a bar chart for quantitative (numerical) data where the x-axis shows class intervals and the y-axis shows relative frequency (proportion or percentage) rather than raw count. Everything else about the construction matches a standard histogram — the bars touch, the width of each bar corresponds to the class width, and there are no gaps for continuous data.
When the bins (class intervals) are identical, the relative frequency histogram and the corresponding frequency histogram have exactly the same shape. Only the vertical scale changes. This means:
- The tallest bar in both charts represents the same class.
- The overall distribution pattern (symmetric, skewed, bimodal) looks identical.
- The y-axis units differ — counts on one, proportions on the other.
| Feature | Frequency Histogram | Relative Frequency Histogram |
|---|---|---|
| Y-axis | Count (raw frequency) | Proportion or percentage |
| Depends on n | Directly — larger n → taller bars | Normalized — scale is always 0 to 1 |
| Good for | Absolute count comparisons | Comparing distributions of different sizes |
| Shape (same bins) | Identical to RF histogram | Identical to frequency histogram |
If the shapes look different, check that the bins are the same and that neither axis has been distorted.
Relative Frequency Calculator
Calculate Relative Frequency
Relative Frequency vs Cumulative Relative Frequency
Relative frequency answers "what share falls in this category?" Cumulative relative frequency answers "what share falls in this category or any category before it?" It is the running total of relative frequencies.
| Feature | Relative Frequency | Cumulative Relative Frequency |
|---|---|---|
| What it shows | Share in one category | Share up to that category |
| Formula | f / n | Cumulative f / n |
| Values | Rise and fall across categories | Never decreases (for ordered data) |
| Final value | Relative frequency of last class | Always ends at 1 (or 100%) |
| Requires ordering | No | Yes — ordering must be meaningful |
How to Calculate Cumulative Relative Frequency
For each row, add its relative frequency to the running total from all previous rows. Using the exam score data from earlier:
| Score Interval | Frequency (f) | Relative Frequency | Cumulative Relative Frequency |
|---|---|---|---|
| 50 ≤ x < 60 | 2 | 0.10 | 0.10 |
| 60 ≤ x < 70 | 5 | 0.25 | 0.35 |
| 70 ≤ x < 80 | 6 | 0.30 | 0.65 |
| 80 ≤ x < 90 | 5 | 0.25 | 0.90 |
| 90 ≤ x ≤ 100 | 2 | 0.10 | 1.00 |
| Total | 20 | 1.00 | — |
The cumulative value at 80 ≤ x < 90 is 0.90, meaning 90% of students scored below 90. The final cumulative value is always 1.00 for a complete single-membership distribution.
Cumulative relative frequency is only meaningful when categories have a logical order — ages, scores, income brackets, ordered ratings. Applying it to unordered nominal categories like "red, dog, pizza" produces a number that depends entirely on which arbitrary order you chose, and has no useful interpretation.
An ogive plots cumulative frequency or cumulative relative frequency against class boundaries. The Ogive Graph Maker creates these plots from your data directly.
Joint, Marginal and Conditional Relative Frequencies
Two-way tables (also called contingency tables) display frequencies across two categorical variables simultaneously. Three types of relative frequency appear in this context, each using a different denominator. Getting the denominator right is the most important step.
A cell count of 8 can produce 8/30, 8/18 or 8/12 depending on whether the question asks for joint, row-conditional or column-conditional relative frequency. These are three different numbers with three different interpretations. Read the question before dividing.
Example Two-Way Table
30 students, cross-classified by grade level (6 or 7) and travel method (Bus or Walk):
| Bus | Walk | Row Total | |
|---|---|---|---|
| Grade 6 | 8 | 10 | 18 |
| Grade 7 | 4 | 8 | 12 |
| Column Total | 12 | 18 | 30 |
Joint Relative Frequency
Joint relative frequency uses the grand total (30) as the denominator. It answers: "What proportion of all students are in both this grade AND this travel group?"
Grade 6 + Bus: 8 / 30 ≈ 0.267 (26.7% of all 30 students are Grade 6 bus riders)
Marginal Relative Frequency
Marginal relative frequency also uses the grand total, but divides a row or column total rather than a single cell. It answers: "What proportion of all students belong to this row or column group?"
- Marginal RF of Bus riders: 12 / 30 = 0.40 (40% of all students take the bus)
- Marginal RF of Grade 6: 18 / 30 = 0.60 (60% of all students are in Grade 6)
Conditional Relative Frequency
Conditional relative frequency changes the denominator to a specific subgroup — either a row total or a column total. It answers: "Among those in this row (or column), what proportion belong to this cell?"
- Row-conditional: Among Grade 6 students (n=18), the proportion who take the bus: 8 / 18 ≈ 0.444 (44.4%)
- Column-conditional: Among bus riders (n=12), the proportion in Grade 6: 8 / 12 ≈ 0.667 (66.7%)
| Type | Denominator | Example (using cell = 8) |
|---|---|---|
| Joint RF | Grand total (30) | 8/30 = 0.267 |
| Marginal RF (row) | Grand total (30) | Row total 18/30 = 0.60 |
| Marginal RF (column) | Grand total (30) | Col total 12/30 = 0.40 |
| Row conditional RF | Row total (18) | 8/18 = 0.444 |
| Column conditional RF | Column total (12) | 8/12 = 0.667 |
For a deeper look at two-way tables and how to read them, see the guide on cross-tabulation.
Relative Frequency and Probability
Relative frequency and probability are related but not identical. Relative frequency describes what actually occurred in a set of observed data. Probability describes a theoretical or model-based likelihood.
Concrete example: A fair coin is flipped 20 times. Heads comes up 12 times. The observed relative frequency of heads is 12/20 = 0.60. The theoretical probability of heads for a fair coin remains 0.50. These two numbers differ — and that is expected. A small sample will rarely match the theoretical probability exactly.
Observed relative frequency can be used as an estimate of probability (called empirical probability), but it is not the same as the true theoretical probability. With more trials under stable conditions, relative frequency tends to get closer to probability — this is the Law of Large Numbers.
The Law of Large Numbers formalizes this: as the number of independent trials grows, empirical relative frequency converges toward the true probability. With 20 flips, seeing 60% heads is unremarkable. With 20,000 flips, seeing 60% heads would be very strong evidence against a fair coin.
How to Calculate Relative Frequency in Excel
If you already have frequency counts in a column, the formula is straightforward:
Scenario: Frequencies are in B2:B5 and the total is in B6.
In cell C2, enter: =B2/$B$6
The $ signs lock B6 as an absolute reference so the denominator doesn't shift when you copy the formula down.
Copy C2 down to C3:C5 to get relative frequencies for every row.
To show percentages: select C2:C5, then apply the percentage number format (Home → Number → %).
✅ Each cell in column C now shows that row's share of the total — fraction 0.40, or formatted as 40%.
Starting from raw data with COUNTIF: If your raw categorical data is in column A (e.g., "Bus", "Walk", "Car"), you can count each category with COUNTIF and then divide by the total:
Excel COUNTIF Approach
Assume raw travel-method data is in A2:A21 (20 observations). Category list is in D2:D5.
E2: =COUNTIF($A$2:$A$21,D2)/20
Copy E2 down to E5 for all categories. Replace 20 with COUNTA($A$2:$A$21) to make n dynamic.
For numeric data that needs to be grouped into class intervals, use Excel's Data Analysis Toolpak histogram feature, then divide each bin count by n to get relative frequency. For categorical data, use a bar chart rather than a histogram — the two chart types are not interchangeable.
Common Relative Frequency Mistakes
| Mistake | Why It's Wrong | Correction |
|---|---|---|
| RF = 15 (no denominator) | 15 is a frequency, not a relative frequency. Without n, no proportion exists. | Divide by n: 15/50 = 0.30 |
| Dividing by number of categories | Wrong denominator. 4 categories ≠ 4 total observations. | Divide by total number of observations (n) |
| Writing 0.40 as "0.40%" | 0.40% = 0.004 as a proportion — 100× too small. | Write 0.40 (proportion) or 40% (percentage) |
| Treating 40 as the relative frequency | 40 is a count unless you explicitly state 40%. | State RF = 40% or RF = 0.40 |
| Confused joint and conditional RF in two-way tables | Wrong denominator changes the interpretation completely. | Joint: divide by grand total. Conditional: divide by row or column total. |
| Expecting RF to always sum to exactly 1.00 | Rounding can give 0.999 or 1.001; multiple-response categories can sum above 1. | Allow ±0.01 for rounding. Explain if >1 due to multiple-response design. |
| Using cumulative RF for unordered categories | A cumulative total over "red, dog, pizza" in arbitrary order has no meaning. | Cumulative RF requires a logical ordering (scores, ages, intervals). |
Practice Questions
7 of 28 survey respondents prefer Brand A. Find the relative frequency.
f = 7, n = 28
RF = 7/28 = 0.25 = 25%
✅ The relative frequency of Brand A preference is 0.25 (25%).
A relative frequency is 0.15 and the sample size is 200. What is the frequency?
f = RF × n = 0.15 × 200 = 30
✅ Frequency = 30.
Four categories have relative frequencies: 0.20, 0.35, 0.30, 0.25. Is there an error?
Sum: 0.20 + 0.35 + 0.30 + 0.25 = 1.10
Sum exceeds 1.00, which is unexpected for mutually exclusive, exhaustive categories.
⚠️ Likely error: wrong denominator, an overlapping-category design, or a calculation mistake. If it's a multiple-response question (respondents could choose more than one), summing above 100% is expected.
Relative frequencies for four ordered classes are: 0.10, 0.25, 0.40, 0.25. Find the cumulative relative frequency at the third class.
Cumulative at class 1: 0.10
Cumulative at class 2: 0.10 + 0.25 = 0.35
Cumulative at class 3: 0.35 + 0.40 = 0.75
✅ Cumulative relative frequency at class 3 is 0.75 (75% of observations fall in the first three classes).
Frequently Asked Questions
Relative frequency tells you what fraction of the total a particular count represents. If 8 out of 20 students take the bus, the relative frequency of bus riders is 8/20 = 0.40 — meaning 40% of students take the bus. It converts a raw count into a share so you can compare groups of different sizes.
Not exactly. Relative frequency describes observed data — what actually happened in your sample. Probability describes theoretical or model-based likelihood. The observed relative frequency of heads in 20 coin flips might be 0.60, while the true probability for a fair coin is 0.50. As sample size grows, observed relative frequency tends to get closer to probability (Law of Large Numbers), but the two are conceptually distinct.
For standard single-membership categories, relative frequency must be between 0 and 1 (0% and 100%). A value above 1 usually means the wrong denominator was used or there is a calculation error. The exception is multiple-response data, where respondents can select more than one category — category percentages can then sum above 100% by design.
Rounding. When individual relative frequencies are rounded to, say, one decimal place, small rounding errors accumulate. Three categories each with an exact frequency of 1/3 will each round to 33.3%, summing to 99.9%. This is not an error — it is expected when rounding. To minimize it, keep more decimal places internally and round only for display.
Marginal relative frequency divides a row total or column total by the grand total. It gives the proportion of all observations that belong to one category of a single variable, ignoring the other variable. For example, if 12 of 30 students take the bus (regardless of grade level), the marginal RF of Bus is 12/30 = 0.40.
If frequencies are in column B (B2:B5) and the total is in B6, enter =B2/$B$6 in C2 and copy down to C5. The dollar signs lock B6 as an absolute reference. Apply percentage formatting to display as 40% instead of 0.4. For raw data, use COUNTIF to count each category first, then divide by n.
The formula RF = f / n works regardless of whether categories overlap. But the interpretation changes: if categories are mutually exclusive and exhaustive, relative frequencies sum to 1. If they overlap (a respondent belongs to multiple categories), or if not all observations are included, the total may not equal 1 — and that is fine as long as you explain the denominator used.
Key Takeaways
- Definition: The proportion of observations in a given category. RF = f / n.
- Three equivalent forms: 8/20 = 0.40 = 40% — choose based on context.
- Not the same as frequency: Frequency is a raw count; relative frequency is a normalized share.
- Why it matters: Comparing groups with different totals requires relative frequency — raw counts favor larger groups.
- Sum check: For mutually exclusive, exhaustive categories the relative frequencies sum to 1 (±rounding).
- Cumulative RF: Running total of relative frequencies; requires ordered categories; ends at 1.
- Denominator is everything: In two-way tables, joint RF uses grand total; marginal RF uses grand total; conditional RF uses row or column total.
- Not theoretical probability: Observed relative frequency estimates probability but equals it only in the limit (Law of Large Numbers).
- Bounds: 0 ≤ RF ≤ 1 for standard single-membership data. Above 1 likely means a denominator error or multiple-response design.