Introduction: Making Data Understandable

Research data only becomes valuable when it can be understood and used by others. Collecting high‑quality data starts with a well‑designed questionnaire and a clear plan for how each variable will be handled. Once the data are in hand, storytelling turns raw numbers into a narrative that engages readers and guides decision‑making.

The Challenges of Data Storytelling

Three common obstacles appear when researchers try to tell a story with data: the abstract, the mundane, and the voluminous. Abstract concepts—such as risk factors that are not directly observable—can be made familiar through visual metaphors that map the unknown onto something concrete. Mundane information, like routine demographic tables, needs fresh visual treatment to avoid boredom. When data sets are massive, the goal shifts from presenting everything to reducing the information to its essential message, so the audience is not overwhelmed.

Knowing Your Audience and Preparing Ahead

Audiences have limited attention spans and often juggle many tasks while reviewing a manuscript. Therefore, researchers must anticipate the workload of their readers and design figures that can be interpreted quickly. Preparation is akin to an artist’s pre‑work: before data collection, develop a detailed analysis plan, decide which variables will be measured, and craft questions that elicit the needed information.

Core Objectives of Data Analysis

The primary objective is data reduction—condensing complex information into a clear, truthful representation. Good analysis separates the signal from chance and bias, ensuring that the conclusions reflect reality rather than random variation.

Types of Data Analysis

• Exploratory – describes prevalence, correlates, and disparities without assuming directionality.
• Relational – examines the association between an exposure and an outcome.
• Comparative – compares groups across time points, interventions, or natural experiments.
• Compositional – breaks down a total into constituent parts through decomposition or segmentation.

Understanding Variable Types

A variable is a column in a dataset that must contain at least two distinct values.

• Continuous – can take any value within a range (e.g., height). Sub‑types:
• Ratio – has a true zero, allowing multiplication and division.
• Interval – lacks a true zero, limiting operations to addition and subtraction.
• Count – discrete, whole‑number values that can be summed (e.g., number of cigarettes smoked).
• Ordinal – categorical with a natural order (e.g., first, second, third).
• Nominal – categorical without inherent order (e.g., gender, color).

The nature of each variable dictates which statistical tests and visualizations are appropriate.

Best Practices in Data Visualization

• Secondary Axes – useful when two series have vastly different scales but share a common x‑axis.
• Tornado/Butterfly Charts – compare multiple measures across sub‑groups side by side.
• Error Bars / Confidence Intervals – display the precision of estimates, making uncertainty visible.
• Forest Plots – show point estimates (e.g., odds ratios) with their confidence intervals for several variables at once.

General Rules for Presenting Data

1. Avoid repeating the same information in both tables and figures.
2. Do not use pie charts in manuscripts; bar charts or other alternatives are preferred.
3. Label every axis clearly and include units where appropriate.
4. Ensure each table and figure can stand alone, with titles and captions that contain person, place, and time.
5. Preserve clean white space; “less is more.”
6. Export images at high resolution (≥ 300 DPI) to prevent pixelation.
7. Use precise language: distinguish prevalence (0–100 %), proportion (0–1), and rate.
8. Apply color thoughtfully, ensuring sufficient contrast without distraction.
9. Provide context; data never speak for themselves.

Practical Tools

• Excel – capable of creating sophisticated charts, including secondary‑axis graphs.
• Ian View – adjusts image resolution to meet journal standards.
• KIS platform – supports survey design, data collection, automatic codebook generation, and preliminary analysis, including data‑quality scoring that flags duplicates, straight‑lining, and speeders.

Clarifying Common Terminology

• The mean of a count variable can be a decimal (e.g., 4.5); rounding may aid lay‑audience comprehension.
• Prevalence ranges from 0 to 100 %, while proportion ranges from 0 to 1; percentages also range from 0 to 100 %.
• “Prevalence rate” is a misnomer; instead, specify prevalence within a clearly defined population (e.g., hospital prevalence).
• “Frequency” is vague; prefer precise terms such as prevalence, count, or rate.

Survey Design Tips

When using platforms like KIS, preview and edit surveys before launch, restrict answer formats where needed, and review the data‑quality score after collection. Understanding the codebook is essential for interpreting the clean dataset and for assessing any remaining quality issues.