Start with what a value represents
Imagine a survey that records each person’s preferred way to travel, commute time in minutes, and postal code. These columns may all contain short entries, but they do different jobs. Preferred travel method names a group; commute time measures a quantity; postal code identifies an area. Understanding those roles is the first step toward choosing a useful calculation or chart.
Categorical data places observations into groups, such as bus, bicycle, or walking. Numerical data expresses a count or measurement, such as the number of trips taken or the minutes spent traveling. The distinction is about meaning, not appearance. A postal code may consist of digits, yet averaging postal codes would not tell you where the typical respondent lives. Before analyzing a column, ask: What does one value stand for, and would arithmetic with these values make sense?
Not all categories or numbers work the same way
Some categories have no natural order. Travel method and product color are examples: you can count how often each occurs, but one color is not inherently higher than another. Other categories are ordered, such as ratings of poor, fair, good, and excellent. Their sequence matters, but the gap between fair and good may not be equal to the gap between good and excellent. Assigning them the numbers 1 through 4 does not automatically create equal intervals.
Numerical data can also take different forms. A count of household members takes whole-number values; a measured commute time can, in principle, include fractions of a minute. Both describe amounts, so differences and averages may be meaningful, depending on the question. A rating expressed as stars deserves more care: although it looks numerical, it may chiefly represent ordered opinions. Deciding how to treat it requires considering what the scale was designed to capture.
Let the data type guide the summary
For categories, start with counts or proportions. If 40 survey respondents choose the bus, that count is easier to interpret alongside the total number of respondents; a proportion helps compare groups of different sizes. For ordered categories, showing the distribution across every response option can reveal more than reporting only the most common choice. Grouping responses too early may hide a split between favorable and unfavorable views.
For quantities, summaries such as the median, average, range, or selected percentiles answer different questions. The median identifies a middle value, while the average uses every value and can be pulled toward unusually large or small observations. If most commutes are short but a few are very long, reporting both the median and the average may be more informative than either alone. Whatever the data type, check what missing values mean before calculating: an unanswered question is not necessarily a zero or a category of its own.
Choose a display that matches the question
A bar chart is often a clear choice for comparing category counts or proportions. Its separate bars reinforce that each label represents a group. If the categories have a meaningful order, keep that order visible; otherwise, sorting bars by size can make comparisons easier. A table may work better when readers need exact values, especially if there are only a few groups.
For numerical measurements, a histogram groups values into intervals and helps show their overall shape. A dot plot or box plot can help compare numerical distributions across categories, such as commute times for different travel methods. A line chart is useful when the horizontal axis represents a meaningful sequence, often time, but it can imply a continuous trend where none exists. Chart choice is not decoration: it signals which comparisons the reader is invited to make.
Watch for changes that alter the meaning
Analysts sometimes turn a numerical variable into categories, such as labeling commutes under 30 minutes “short” and the rest “long.” This can make a simple comparison possible, but it discards detail: commutes of 5 and 29 minutes receive the same label, while 29 and 30 minutes fall into different groups. The cutoff should serve a clear purpose rather than merely produce a striking result.
The reverse move also calls for caution. Coding categories as 1, 2, and 3 may help store or sort responses, but those codes do not make arithmetic meaningful. When you encounter a result, look past the column’s format to its definition. Ask whether the values identify groups, express an order, or measure an amount; then check whether the summary and chart respect that meaning. That habit makes data easier to explain and harder to misread.
