How to Use Graphical Methods for Data Presentation

Numbers tell stories, but those stories often hide inside rows of dense data. A table showing infant mortality rates across 28 states can take minutes to interpret, while a colour-coded map can reveal the same pattern in seconds. This is the core promise of graphical methods: turning raw statistics into something the human eye can process almost instantly. In population and family health research, where datasets often involve thousands of households, multiple time periods, and several variables, choosing the right graph can be the difference between an insight that travels and one that gets buried in spreadsheets.

Why graphical presentation matters in research 🔗

Tables are precise, but graphs are persuasive. A well-designed chart compresses large volumes of information into a form that can be scanned, compared, and remembered. Researchers studying public health outcomes have long argued that visualization is now considered the seventh “V” of big data, alongside volume, velocity, and variety, because summarising and interpreting raw figures has become a central challenge of modern research.

There are three practical advantages worth noting. First, graphs reveal patterns that tables conceal, such as seasonal peaks in disease incidence or sudden drops in vaccination coverage. Second, they make comparisons across groups, regions, or time periods immediate. Third, they communicate findings to non-specialists, policymakers, journalists, and community workers, who may not have the patience or training to read a frequency table. The format chosen for the data, whether a line graph, bar chart, scatter plot, or map, depends on the type of relationship the researcher wants to show.

Common types of graphs and when to use them 🔗

No single chart works for every dataset. The right choice depends on what the variable measures, how many categories exist, and what comparison the researcher is trying to highlight. Below is a working guide to the most widely used graphical methods in population and family health research.

Bar charts 🔗

Bar charts use rectangular bars of equal width, with lengths proportional to the values they represent. They are best suited for comparing discrete categories, such as the number of antenatal visits across districts, contraceptive method preferences among women, or the share of children fully immunised in different states. Bars can be drawn vertically or horizontally; horizontal bars work better when category names are long, like full district names. A multiple bar chart, where two or more bars sit side by side for each category, is useful for comparing male and female literacy rates or rural and urban figures within the same district.

Pie charts 🔗

Pie charts represent parts of a whole, with each slice proportional to its share. They suit data that adds up to 100%, such as the percentage breakdown of causes of under-five mortality, religion-wise distribution of households in a survey, or the proportion of total health expenditure by source. The rule of thumb is to use a pie chart only when there are five or fewer categories. Beyond that, the slices become too thin to read, and a bar chart almost always communicates the same information more clearly. Three-dimensional pie charts should be avoided altogether, since they distort the relative size of slices.

Histograms 🔗

A histogram looks similar to a bar chart, but the two are not the same. Histograms display the distribution of a single continuous variable by grouping values into intervals called bins. The bars touch each other to indicate that the variable flows continuously. In epidemiology and family health, histograms are routinely used to show the age distribution of mothers at first birth, the spread of birth weights in a hospital, or the distribution of household income in a survey. A specialised form called the epidemic curve, which plots the number of cases by time of onset during an outbreak, is a histogram used widely in disease surveillance.

Line graphs 🔗

Line graphs are the natural choice for showing how a variable changes over continuous time. They are indispensable for tracking trends such as the total fertility rate over five decades, monthly outpatient attendance at a primary health centre, or annual under-five mortality. A simple line graph plots one variable, while a multiple line graph plots two or more for comparison, for example, infant mortality rates in two states across the same years. During the COVID-19 pandemic, line graphs of daily case counts became one of the most recognisable visual artefacts in public health communication, helping administrators identify epidemic patterns and even anomalies, such as troughs that turned out to correspond with Sundays and reporting delays rather than real drops in cases.

Scatter diagrams 🔗

A scatter diagram, or scatter plot, shows the relationship between two continuous variables by plotting each observation as a single point. The pattern of points reveals whether the two variables move together, move in opposite directions, or have no clear relationship. Researchers in family health use scatter plots to explore links such as maternal age and birth weight, household income and child nutrition status, or female literacy and contraceptive prevalence. A well-known use is plotting body mass index against systolic blood pressure to study cardiovascular risk. Scatter plots also expose outliers, those single points that lie far from the cluster and often deserve a closer look.

Maps and choropleth diagrams 🔗

When data has a spatial element, maps become irreplaceable. A choropleth map shades regions, such as districts or states, in different colours or intensities based on the value of a variable. They are widely used to display the geographic distribution of disease burden, sex ratio, contraceptive use, or maternal mortality. The choropleth map has been used alongside scatter plots and forest plots to compare the impact of COVID-19 on public health across countries, including India. For internal planning, district-level maps help health administrators identify clusters that need targeted intervention, something a national average can completely hide.

Other useful graphs 🔗

Beyond the core six, a few specialised graphs deserve mention. Box plots (or box-and-whisker plots) summarise the spread of a continuous variable using the minimum, first quartile, median, third quartile, and maximum. They are useful for comparing the distribution of blood pressure values across two treatment groups, or birth weights across hospitals. Frequency polygons and ogives are line-based alternatives to histograms, useful when comparing two distributions on the same axes. Pictograms, which use small icons to represent units, are popular in public health posters and government dashboards because they are easy to understand even with limited literacy.

Best practices for creating effective graphs 🔗

A graph is only useful if it can be read accurately. Decades of work by researchers such as Edward Tufte have produced a fairly settled set of guidelines, and modern practitioners have built on them. The principles below apply equally to a slide for a conference, a figure in a journal article, or a chart in a district health report.

Choose the chart that fits the data 🔗

The first decision is the most important. Use bar charts for categorical comparisons, line graphs for trends over time, histograms for the distribution of continuous data, scatter plots for relationships between variables, and maps for spatial patterns. Forcing a pie chart onto twelve categories or using a 3D bar chart on simple percentages obscures rather than reveals. The choice between a line graph and a bar chart for time-series data often hinges on whether the time variable is continuous or discrete, such as semesters or fiscal quarters.

Label everything clearly 🔗

Every chart needs a clear title that states what the data show, axis labels with units of measurement, a legend if multiple series appear, and a note about the data source. A reader should be able to understand the chart without scrolling back to the surrounding text. Sans-serif fonts in sentence case, with sufficient size for readability, work best for digital visuals. Direct labelling, where data points are annotated on the chart itself rather than through a legend, reduces the back-and-forth a reader has to do.

Keep the design clean 🔗

The concept of “chartjunk”, coined by Tufte, refers to all the decorative elements that add no information: heavy gridlines, 3D effects, drop shadows, busy backgrounds, and excessive colours. Removing chartjunk and integrating labels directly onto the chart improves clarity and credibility. A useful self-test is to remove each element one at a time and ask whether its absence makes the chart harder to read. If not, leave it out.

Use colour with purpose 🔗

Colour should encode information, not decorate. Limit a chart to three or four distinct colours, use a single accent colour to highlight the key category, and pick colourblind-safe palettes such as Viridis or those available on ColorBrewer. Pairing colour with other visual cues like patterns or shapes ensures the chart remains readable for viewers with colour vision deficiencies, and keeps the message intact when printed in greyscale.

Be honest with axes and scales 🔗

For magnitudes such as counts, revenue, or population, the vertical axis should usually start at zero. A truncated axis can exaggerate small differences and mislead the reader. When comparing two charts side by side, the axis ranges should be locked to the same scale so that visual differences correspond to actual data differences. This principle, sometimes called graphical integrity, is the foundation of trustworthy data communication.

Interpret, do not just display 🔗

A graph is part of an argument, not the whole of it. Every figure in a research report should be accompanied by a short interpretation in the text, drawing attention to the main pattern, any anomalies, and the limitations of the data. If a chart shows a steep decline in maternal mortality over twenty years, the text should explain whether the change reflects real improvement, better reporting, or a change in definitions.

Putting it all together 🔗

Choosing a graph is an analytical decision as much as a design one. The variable type, the question being asked, the audience, and the publication format all matter. A demographer presenting fertility trends to a state health department might rely on line graphs and choropleth maps. A community health worker explaining child nutrition to a village committee might use a simple bar chart or pictogram. A peer-reviewed journal article on maternal anaemia might combine histograms, scatter plots, and box plots within the same paper. The skill lies not in mastering every chart type but in matching the right one to the right question, and then designing it with enough care that the data speaks for itself.

What do you think? Which type of graph do you find most convincing when you read a health report, and why? And if you had to explain a complex finding about family health to someone with no statistical background, which single visual would you choose?

References

1. https://pmc.ncbi.nlm.nih.gov/articles/PMC8106281/
2. https://pmc.ncbi.nlm.nih.gov/articles/PMC10394528/
3. https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson4/summary.html
4. https://pmc.ncbi.nlm.nih.gov/articles/PMC10193847/
5. https://udair.missouri.edu/visualization-chart-best-practices/
6. https://affine.pro/blog/principles-of-effective-data-visualization
7. https://frozencrow.com/data-visualization-best-practices/
8. https://fivenumbersummary.io/blog/data-visualization-best-practices/