Presentation of Data in Descriptive Biostatistics
Data presentation organises observations so that patterns, comparisons and important findings can be understood accurately. Good presentation begins with the type of variable and the question being answered—not with choosing an attractive chart.
Learning objectives
- Classify variables and levels of measurement.
- Construct frequency distributions and grouped tables.
- Select appropriate charts and graphs.
- Present epidemiological data by person, place and time.
- Identify misleading practices and reporting errors.
From raw data to information
- Define the purpose and audience.
- Check data completeness, validity and consistency.
- Classify variables.
- Code and group data appropriately.
- Calculate frequencies, proportions or rates.
- Select a table, graph, map or summary measure.
- Label clearly and interpret the main message.
Types of variables
| Variable type | Description | Examples |
|---|---|---|
| Nominal | Categories without natural order | Blood group, district, diagnosis |
| Ordinal | Ordered categories with unequal or unknown intervals | Pain severity, disease stage, satisfaction |
| Discrete numerical | Countable whole-number values | Number of admissions, pregnancies, children |
| Continuous numerical | Can take any value within a range | Weight, temperature, blood pressure |
| Binary/dichotomous | Two categories | Alive/dead, test positive/negative |
Levels of measurement
- Nominal: categories only.
- Ordinal: categories with order.
- Interval: equal intervals but no true zero, such as temperature in °C.
- Ratio: equal intervals and a meaningful zero, such as age, weight or duration.
The measurement level determines valid summaries and statistical tests.
Tabular presentation
Simple frequency table
A frequency table lists values or categories and the number in each. Add percentages when groups of different sizes are compared.
| Malaria test result | Frequency | Percentage |
|---|---|---|
| Positive | 60 | 30% |
| Negative | 140 | 70% |
| Total | 200 | 100% |
Cross-tabulation
A two-way table displays the relationship between two categorical variables, such as exposure and disease. Include row or column percentages according to the question.
Grouped frequency distribution
Continuous values may be grouped into mutually exclusive class intervals.
- Find the range: maximum − minimum.
- Choose a sensible number of classes.
- Calculate approximate class width.
- Create non-overlapping intervals.
- Tally observations and compute frequency.
- Add relative and cumulative frequencies if useful.
Class-interval rules
Intervals should cover all values, not overlap, use consistent width where possible and avoid too many or too few groups. Retain raw data because grouping loses detail.
Graphs for categorical data
Bar chart
Separate bars compare frequencies, proportions or rates across categories. The baseline should normally start at zero.
Pie chart
Shows parts of one whole. Use few mutually exclusive categories that total 100%; a bar chart is often easier to compare.
Stacked bar chart
Shows composition across groups, but small segment differences can be difficult to compare.
Dot plot
Compact display for category values or small numerical datasets.
Graphs for numerical data
Histogram
Adjacent bars display the distribution of continuous data grouped into intervals. Bar area represents frequency; there are no gaps because the scale is continuous.
Frequency polygon
Plots class midpoints against frequency and joins them with lines. It is useful for comparing distributions.
Box-and-whisker plot
Displays median, quartiles, spread and potential outliers. It permits efficient comparison of numerical distributions between groups.
Scatter plot
Displays paired numerical observations and reveals direction, form, strength and outliers. Association does not prove causation.
Graphs for time and epidemiology
| Display | Best use | Key feature |
|---|---|---|
| Line graph | Trend over regular time points | Time on horizontal axis |
| Epidemic curve | Cases by time of onset during an outbreak | Intervals related to incubation period |
| Spot map | Geographic distribution of cases | Shows location but not population at risk |
| Choropleth map | Rates or proportions by area | Uses comparable denominators and ordered shading |
| Population pyramid | Age-sex composition | Paired horizontal bars |
Person, place and time presentation
- Person: tables or bars by age, sex, occupation and risk group.
- Place: maps or area tables using counts and population-based rates.
- Time: line graphs, seasonal plots and epidemic curves.
Essential components of a table
- Table number
- Clear self-explanatory title
- Row and column headings
- Units and denominator
- Body with aligned values
- Totals where appropriate
- Footnotes for definitions or missing data
- Data source and period
Essential components of a graph
- Informative title
- Labelled axes and units
- Consistent scale
- Legend only when necessary
- Readable category labels
- Source and explanatory notes
- Accessible colour or pattern choices
Worked selection examples
- Compare numbers of cases across five diagnoses: bar chart.
- Show monthly malaria cases for three years: line graph.
- Display distribution of patient ages: histogram.
- Compare blood-pressure distributions in two treatment groups: box plots.
- Examine relation between BMI and systolic pressure: scatter plot.
- Show outbreak cases by date of symptom onset: epidemic curve.
Frequency measures
- Absolute frequency: number of observations.
- Relative frequency: frequency divided by total.
- Percentage frequency: relative frequency × 100.
- Cumulative frequency: running total up to a value or class.
Misleading presentation
| Problem | Why misleading | Better practice |
|---|---|---|
| Truncated bar-chart axis | Exaggerates small differences | Start at zero or clearly justify a break |
| Unequal time intervals | Distorts trend | Use equal spacing or explain intervals |
| Counts on a map | Large populations appear high-risk | Map rates when comparing risk |
| Three-dimensional chart | Perspective distorts size | Use simple two-dimensional displays |
| Too many categories | Message becomes unreadable | Group logically without hiding important detail |
| Missing denominator | Percentages cannot be judged | Report n/N and definitions |
Reporting missing data
Do not silently exclude missing observations. State the number missing, use appropriate denominators and consider whether missingness could bias results. “Unknown” may be a useful category for descriptive tables but should not be confused with a true negative.
Choosing the best display
- Identify the main comparison or pattern.
- Determine whether variables are categorical, numerical or time-based.
- Use the simplest display that answers the question.
- Preserve honest scales and denominators.
- Make the display understandable without excessive explanation.
- Add a concise interpretation that reports the main finding rather than repeating every value.
Exam-focused summary
- Variable type determines the appropriate table, graph and summary.
- Bar charts have separated bars; histograms use adjacent bars.
- Line graphs display time trends; scatter plots display two numerical variables.
- Maps comparing disease risk should normally use rates, not raw counts.
- Every display needs a title, labels, units, denominator and source.
Review questions
- Classify ten example variables as nominal, ordinal, discrete or continuous.
- Construct a grouped frequency table.
- Differentiate a bar chart and histogram.
- Select an appropriate display for six health-data scenarios.
- List eight components of a good table.
- Explain five ways graphs can mislead.
References and further reading
- CDC Field Epidemiology Manual: Describing data.
- WHO Global Health Observatory.
- Clinical Medicine Year 2 curriculum.
Educational use only. Use current national reporting definitions and protect confidential data.
