Numbers, categories, names, and measurements surround every health study, every policy report, and every classroom presentation. But before any of these can tell a meaningful story, they must first be understood as data. In population and family health research, knowing what kind of data you are working with shapes everything that follows: the way you collect it, the tools you use to analyse it, and even the conclusions you can responsibly draw. This post breaks down what statistical data really means and walks through the major types every student of health research should be confident about.
Table of Contents
- What is data in statistics?
- Why classification matters before analysis
- Qualitative versus quantitative data
- Qualitative data
- Quantitative data
- Putting the two side by side
- Discrete and continuous data
- Discrete data
- Continuous data
- The blurred line between discrete and continuous
- Why these distinctions matter in population and family health research
- A quick mental checklist
- Bringing it all together
What is data in statistics?
In statistics, data refers to the raw facts, figures, observations, or descriptions collected about people, events, or phenomena. A respondent’s age, a mother’s level of education, the number of antenatal visits she made, the village she lives in, her haemoglobin level – each of these is a piece of data. On its own, a single value rarely means much. But once data is systematically collected, organised, and analysed, it becomes information that can answer real research questions.
Consider the National Family Health Survey (NFHS-5), conducted by the International Institute for Population Sciences. It gathered information from approximately 610,000 households across 707 districts to produce estimates on fertility, child mortality, nutrition, and family planning. None of those national-level findings would exist without the careful collection of individual data points from each respondent. The same logic applies to a small classroom project surveying 50 households about handwashing practices – the scale differs, but the foundation is identical.
Data in statistics typically passes through three stages. Collection involves gathering values through surveys, interviews, clinical measurements, or official records. Organisation involves arranging these values into tables, categories, or datasets that a computer or human can interpret. Analysis involves applying statistical tools – averages, proportions, regressions, or visualisations – to extract meaning. The type of data you have determines which tools are appropriate, which is why classification matters before computation.
Why classification matters before analysis
You cannot calculate the “average religion” of a sample, nor can you make a pie chart that meaningfully compares every individual weight in a population of 50,000. The choice of statistical test, the choice of graph, and even the choice of summary statistic all depend on whether the data is numerical or categorical, and whether it is countable or measurable. Researchers at the National Institutes of Health note that recognising data type is the very first step in selecting the right statistical test for a study.
Qualitative versus quantitative data
The most fundamental classification in statistics divides data into two large families: qualitative and quantitative.
Qualitative data
Qualitative data, also called categorical data, describes qualities or characteristics that cannot be expressed as meaningful numbers. It places observations into groups or labels. In a health survey, common qualitative variables include occupation, religion, marital status, blood group, type of fuel used for cooking, mode of delivery (normal or caesarean), and place of residence (urban or rural).
You cannot perform arithmetic on these values in the usual sense. Adding “farmer” and “teacher” produces nothing meaningful. Instead, qualitative data is summarised through frequencies and proportions – for example, “62% of respondents reported being employed in agriculture” – and is typically visualised through bar charts or pie charts.
Qualitative data is often further divided into two sub-types. Nominal data has categories with no natural order, such as gender, religion, or blood type. Ordinal data has categories with a meaningful rank but unequal gaps between them, such as wealth quintiles (poorest to richest) or education levels (illiterate, primary, secondary, higher). The distinction between nominal and ordinal matters because ordinal variables allow some additional analyses, like calculating a median, that nominal variables do not.
Quantitative data
Quantitative data consists of numerical values that represent measurable quantities. Income in rupees, age in years, weight in kilograms, blood pressure in mmHg, number of children ever born, and distance to the nearest primary health centre are all quantitative variables.
Unlike qualitative data, quantitative values can be added, averaged, and subjected to a wide range of statistical operations. You can meaningfully calculate the mean income of a community, find the standard deviation of birth weights in a hospital, or test whether average haemoglobin differs between two districts. Quantitative data is usually visualised through histograms, scatter plots, line graphs, and box plots.
Putting the two side by side
A single survey almost always collects both. In a typical maternal health survey, a researcher might record a respondent’s occupation (qualitative), her caste category (qualitative – ordinal or nominal depending on classification), her monthly household income (quantitative), and her age at first marriage (quantitative). The qualitative variables help describe who the respondent is; the quantitative variables help measure how much, how many, or how long. Together they paint a complete picture, which is why most large-scale surveys deliberately combine both.
Discrete and continuous data
Within quantitative data, a second important split appears: between discrete and continuous data. Both are numerical, but they behave very differently.
Discrete data
Discrete data consists of values that can be counted and that take only specific, separate values – typically whole numbers. According to a guide from Harvard Library, discrete data is finite, countable, and cannot be broken into fractions in a way that still makes sense.
Examples from population health research are everywhere: the number of children ever born to a woman, the number of antenatal care visits during a pregnancy, the number of household members, the number of immunisation doses received, the number of hospital admissions in a year. You cannot have 2.7 children in a household – the value jumps from 2 to 3. Although the textbook example of gender is sometimes labelled discrete in introductory materials, gender is more accurately a nominal categorical variable; cleaner examples of discrete numerical data are counts such as the number of births in a village last year.
Discrete data is best displayed using bar charts, frequency tables, or dot plots, where each bar or point represents a distinct count.
Continuous data
Continuous data takes values that can be measured rather than counted, and these values can fall anywhere along a continuum, including decimals and fractions. A researcher from the National Institutes of Health explains that continuous data can be divided into finer and finer levels as instruments grow more precise.
Age, weight, height, blood pressure, haemoglobin level, body mass index, temperature, and time spent breastfeeding are all continuous variables. A baby’s birth weight might be 2.847 kilograms; a respondent’s age might be 27 years, 4 months, and 12 days. The precision is limited only by the measuring instrument. Continuous data is usually displayed using histograms, line graphs, scatter plots, or box plots, all of which capture the smooth flow of values across a range.
The blurred line between discrete and continuous
Age is an interesting case. In daily life it is treated as continuous because it flows from one moment to the next. But in many surveys, age is recorded in completed years, which makes the recorded variable behave like a discrete count. Whether to analyse it as discrete or continuous often depends on how it was collected and what question is being asked. This is a useful reminder that the labels are not always rigid – they reflect how a variable is measured and used.
Why these distinctions matter in population and family health research
The reason these classifications are drilled into every research methodology syllabus is simple: they directly determine what analysis is appropriate.
If you want to compare the average birth weight of babies in two districts, you are comparing two continuous variables and a t-test or its non-parametric equivalent may be appropriate. If you want to see whether vaccination status (vaccinated/unvaccinated) differs by religion, you are comparing two categorical variables, and a chi-square test is the standard choice. If you want to study how income (quantitative) influences the number of antenatal visits (discrete), you may turn to regression models suitable for count data.
This is why large studies in India – the NFHS, the Longitudinal Ageing Study in India (LASI), and the National Sample Survey – are designed from the ground up with data types in mind. Each variable is carefully defined, coded, and stored so researchers across the country and the world can analyse it consistently. A scoping review of multimorbidity research in India found that of 100 published studies, 62 relied on secondary analysis of such large surveys, demonstrating just how much downstream research depends on well-classified original data.
A quick mental checklist
When you encounter a new variable in a dataset, ask three quick questions. First, does it describe a category or a number? That separates qualitative from quantitative. Second, if it is a number, did someone count it or measure it? Counts are discrete; measurements are continuous. Third, if it is a category, is there a natural rank? If yes, it is ordinal; if no, it is nominal. With these three questions, you can correctly classify almost any variable you will meet in a population or family health study.
Bringing it all together
Data is the building block of statistics, and statistics is the engine of evidence-based health research. Knowing the difference between qualitative and quantitative data, and between discrete and continuous data, is not a textbook formality – it is a working skill that shapes how researchers ask questions, design surveys, choose tests, and present findings. The next time you read a health report that mentions average household size, percentage of institutional deliveries, or median age at marriage, you will recognise exactly what kind of data is doing the work behind each statistic.
What do you think? If you were designing a small household survey on women’s nutrition in your locality, which three variables would you collect, and how would you classify each one – qualitative or quantitative, discrete or continuous? And do you think classifying age as “completed years” instead of an exact decimal changes how meaningfully it can be analysed?
Leave a Reply