13 Research methods, statistics, critical review and evidence-based practice13.8 Descriptive statistics
13.8.9.3 Tabular presentation
1 min read
Written by Resident Doctor Committee · Published
Frequency tables
A frequency table gives each value's count and proportion, and also its cumulative proportion when the values are ordered.
Continuous values are grouped into non-overlapping intervals with unambiguous boundaries. Bands that are too wide hide peaks, and bands that are too narrow give sparse, noisy counts.
An open-ended top group, such as 58 and over, hides the upper range, so any grouped mean is only approximate.
The two-by-two table
Columns hold condition status and rows hold the test result, with positive categories first.
Test result | Condition present | Condition absent |
|---|---|---|
Positive | a, true positive | b, false positive |
Negative | c, false negative | d, true negative |
Transposing the table swaps b and c, so read the labels before applying a memorised formula.
Percentages and denominators
Denominator | Example | What it answers |
|---|---|---|
Row total | 10 / 20 = 50% | Screen positives who have the condition |
Column total | 10 / 12 = 83.3% | Cases who screen positive (sensitivity) |
Grand total | 10 / 40 = 25% | All people who are both positive and cases |
The chance that a positive result is genuine is the row percentage, not sensitivity.
When case and control groups differ in size, compare exposure prevalence as percentages, not raw counts, because a smaller count among cases can still be a higher proportion.
Label each percentage with its denominator, because rounding can stop readers recovering the counts.
Checking a table
Cells, margins and the grand total must agree, and denominators must match the stated population.
Percentages of mutually exclusive, exhaustive categories should total about 100%, and impossible values, such as an age of 152, need checking.
Report excluded records and why data are missing, because exclusions can bias results.