Relative Frequency Calculator

What relative frequency measures

A frequency counts how many times something happened. A relative frequency turns that count into a share by dividing it by the total number of observations. If 24 of 60 students play a sport after school, the frequency is 24 and the relative frequency is 24 / 60 = 0.40, which is 40 percent.

Relative frequencies are proportions, so they land between 0 and 1, and the entries of a complete table add to 1, or 100 percent. That property is what makes them useful for comparison. A school with 60 students and a school with 900 cannot be compared by raw counts, but their shares can. The NIST/SEMATECH e-Handbook uses the same rule for histograms, where relative counts are normalized by dividing each class count by the total number of observations so the heights sum to one.

The denominator is the group you are describing. Change the denominator and you change the question. A count of 12 ninth graders with a part-time job becomes 12/100 of all students, 12/30 of ninth graders, or 12/50 of students who hold a job, and each of those is a different, correct relative frequency.

Reading a frequency table into relative frequencies

Start with counts, then divide every row by the total. Table 1 shows 60 students asked what they do most after school.

Check the arithmetic by adding the shares: 0.40 + 0.25 + 0.20 + 0.15 = 1.00. A sum of 0.98 or 1.02 means a category is missing or each row was rounded too hard. Keep four decimals while you work and round only for display.

The largest relative frequency marks the mode, the most common category. Here that is sports at 0.40, and the smallest is video games at 0.15. Sorting the rows from largest share to smallest turns the table into a Pareto chart, which is a bar chart of shares in descending order.

Cumulative relative frequency

Cumulative relative frequency adds the shares as you move down the table in order. The first row is unchanged, the second row sums the first two, and the final row must land on 1.00. Table 2 extends the after-school example.

The cumulative column answers questions about "at or below this row." Of the 60 students, 65 percent chose sports or music, and 85 percent chose one of the first three activities. The row where the cumulative value first passes 0.50 holds the median category, which is why grouped tables pair a cumulative column with the counts.

OpenStax Introductory Statistics 2e runs the same column through a table of student work hours, where 3 of 20 students work 2 hours (0.15) and 5 work 3 hours (0.25), giving cumulative values of 0.15, 0.40, 0.55, 0.85, 0.95 and 1.00. Its table of 100 soccer player heights puts 40 players in the 65.95 to 67.95 inch bin, a relative frequency of 0.40, with a cumulative 0.63 at that row.

Two way tables: joint, marginal and conditional

A two way table cross-tabulates two categorical variables, so every cell belongs to one row category and one column category at once. The counts are ordinary frequencies, but the table offers three denominators worth using, and each one produces a differently named relative frequency.

A joint relative frequency is a cell count divided by the grand total. It answers "what share of everyone is in this row and this column?" A marginal relative frequency is a row total or column total divided by the grand total. It answers "what share of everyone is in this row?" and ignores the other variable. A conditional relative frequency is a cell divided by its row total or column total. It answers "within this row, what share falls in this column?"

Table 3 is a survey of 100 students, cross-tabulated by grade and by whether they hold a part-time job.

The grand total is 100, so each joint relative frequency is the cell count moved two decimal places: 12 becomes 0.12. Row marginals are 0.30, 0.40 and 0.30, and column marginals are 0.50 and 0.50. Marginals describe one variable at a time, which is why they sit in the margin of the table and carry the name.

Worked example: the same cell, three denominators

Take the 12 ninth graders who hold a job. Three questions about that one cell produce three different relative frequencies, and all three are correct.

A sentence that says "40 percent of students have a job" is wrong for this table. The 40 percent is the ninth grade rate, and the overall rate is 50/100 = 0.50. Reading a conditional as a joint is the mistake this example is built to expose.

Table 5 shows the conditional rates by grade, each row computed against its own total.

The job rate rises from 40 percent in ninth grade to 60 percent in eleventh, while the overall rate holds at 50 percent. That pattern is invisible in the marginal column, which is the reason to compute conditionals: marginals describe the whole group, conditionals describe a slice of it.

Decimals, fractions and percents

One number, three outfits. 0.65, 13/20 and 65 percent describe the same share. Multiply a decimal by 100 for a percent, divide a percent by 100 for a decimal, and write the decimal over 1 followed by one zero for each digit after the point, then reduce.

Fractions are exact, decimals terminate only when the denominator divides a power of 10, and percents are the easiest form for a general audience. One third is 0.333... and 33.3 percent, never exactly 33 percent. Rounding to a whole percent is fine in a report, as long as the exact count sits in the table so a reader can recompute.

Pick a display precision and hold it. Mixing 0.4 with 40.0 percent in one column makes a table look inconsistent even when the numbers agree, and a table that reports 0.6667 next to 66 percent looks like two different claims.

Mistakes that flip the answer

Wrong denominator. Dividing a cell by its row total when the question asks about the whole group, or the reverse, produces most of the wrong answers on two way table problems. Write the denominator in words before you compute.

Shares that do not add up. If the relative frequencies sum to 0.97, a category is missing or a count is wrong. The sum is a free check, and it catches errors that look fine row by row.

Comparing counts across unequal groups. 15 out of 20 looks larger than 40 out of 100 until you convert: 0.75 against 0.40. Raw counts reward the bigger group.

Reporting a conditional as a marginal. "60 percent of eleventh graders have a job" is 18/30, while "60 percent of students are eleventh graders with a job" would be 18/100 = 0.18. The words decide the denominator.

Treating a sample share as a fixed probability. A relative frequency from a sample estimates a probability, and a small sample can miss by a wide margin. More observations narrow the gap, which is the law of large numbers at work.

Where to find real frequency data

The Census Bureau publishes commuting shares every year from the American Community Survey. Its 2024 one year estimates put workers who drove alone to work at 69.2 percent, the same as 69.2 percent in 2023. That percent is a relative frequency: the count of solo drivers divided by all workers.

For textbook tables with the counts already laid out, OpenStax Introductory Statistics 2e is free and includes the work hours and soccer height tables used above. The NIST/SEMATECH e-Handbook gives the formal definition of normalized counts and notes that relative counts sum to one when each class count is divided by the total number of observations.

When you pull a table from one of these sources, keep the original counts next to the shares you compute. A reader who has the counts can check any relative frequency in the table, and a table that shows only percents cannot be audited.