What is quantitative data?
Quantitative data describes amounts or measurements numerically. Examples include the number of members attending an event, the time taken to resolve a request and the distance traveled to a service. It supports questions about how many, how much, how often and how values differ or change.
Not every field containing digits is a quantity. A membership number identifies a record; it does not measure membership. A code such as 1 for one region and 2 for another remains a category. Averaging those region codes would not produce a meaningful regional measure.
Surveys also collect ordered ratings, such as satisfaction categories coded from 1 to 5. These are frequently analyzed numerically, but their meaning and measurement assumptions need attention. Start by identifying what the field represents, not just how it is stored.
Discrete and continuous quantitative data
Discrete data takes separate, countable values, such as the number of completed applications. Continuous data represents measurements on a continuum, such as elapsed time, even when the recorded value is rounded. OpenStax's introduction to data types explains this distinction.
Scroll horizontally to see all columns →
| Type | Example | Collection detail to preserve |
|---|---|---|
| Discrete count | 12 applications received | What counts as an application and the period covered |
| Discrete count | 4 sessions attended | Whether repeat attendance and partial sessions count |
| Continuous measurement | 18.5 minutes waiting | Start and end events, time unit and recording precision |
| Continuous measurement | 7.2 kilometers traveled | Measurement method, unit and whether it is one-way distance |
A count remains discrete even when its average is a decimal. An average of 2.5 visits does not mean someone made half a visit; it summarizes a group of whole-number counts. Likewise, recording time in whole minutes does not make the underlying duration a count of separate events.
Levels of measurement: what do the numbers mean?
Nominal, ordinal, interval and ratio describe the meaning of a scale. They are not four equivalent kinds of quantitative data: nominal and ordinal variables are commonly classified as categorical. The OpenStax measurement guide sets out these distinctions.
Scroll horizontally to see all columns →
| Level | Meaning | Example |
|---|---|---|
| Nominal | Categories with no inherent order | Region or service type, even when stored as numeric codes |
| Ordinal | Ordered categories without assured equal gaps | Very dissatisfied through very satisfied |
| Interval | Equal differences, without a meaningful absolute zero | Temperature in degrees Celsius |
| Ratio | Equal differences and a meaningful zero | Elapsed time or number of visits |
The distinction helps you decide which calculations and interpretations make sense. A temperature of 20°C is not twice as hot as 10°C in the ratio-scale sense. By contrast, a 20-minute wait is twice the duration of a 10-minute wait.
Examples in everyday organizational work
- Membership: active member organizations, event attendance and renewal rates with a defined eligible group.
- Customer experience: resolution time, repeat contacts and the distribution of satisfaction responses.
- Employee experience: response coverage, ordered survey responses and participation in an agreed follow-up activity.
- Training: attendance, assessment scores and reported application at a suitable follow-up point.
- Applications: submissions received, review completion time and ratings against a defined rubric.
- Partner operations: delivery counts, elapsed handoff time and the proportion of records with required evidence.
Keep the unit clear. A person, response, visit, application and organization are not interchangeable. Many reporting mistakes begin when the label sounds familiar but the underlying unit differs across teams.
How to collect quantitative data
Define the question and measure
Decide what you need to understand before adding a numeric field. For waiting time, specify whether the clock begins at registration, referral or arrival. For attendance, explain whether a partial session counts. A numeric answer is useful only when contributors understand what they are measuring.
Choose an appropriate source
Surveys collect reported information; direct measurements, assessments and operational records supply other types of evidence. A participant's estimate of waiting time and a timestamp-based duration may both be useful, but they are not the same measure. Keep the source and method visible.
Plan coverage and timing
Identify the population, how observations will be selected and when collection occurs. A convenient set of respondents may not represent the wider group. For repeated collection, decide whether you need to follow the same units or compare separate samples over time.
Make missing and inapplicable answers explicit
Separate zero from unknown, not applicable and not yet collected. Zero visits is a valid count; a blank visit field does not necessarily mean zero. Use validation to flag possible errors, while allowing legitimate exceptions to be reviewed rather than silently changed.
See quantitative data collection methods for a fuller collection plan.
A worked example: summarizing waiting times
Suppose a fictional service team records five waiting times in minutes: 4, 5, 5, 6 and 30. The mean is 50 divided by 5, or 10 minutes. The median is 5 minutes, the middle value after ordering the observations.
Both summaries are correct. They answer slightly different questions. The mean includes the influence of the long wait, while the median describes the center without being pulled upward as strongly. Showing only one can hide a useful feature of the experience.
Scroll horizontally to see all columns →
| Summary | Result | Interpretation |
|---|---|---|
| Count | 5 recorded waits | The size of this example, not necessarily the full service population |
| Mean | 10 minutes | Total recorded time divided by observations |
| Median | 5 minutes | The middle recorded wait |
| Minimum and maximum | 4 and 30 minutes | The observed extremes in this small set |
Do not delete the 30-minute value merely because it is unusual. Check whether it is a recording error, a different type of request or a genuine long wait. Operational context or a comment may help investigate. Neither should be invented to make the pattern look tidy.
If the same person made several visits, the row is a visit rather than a unique person. That distinction matters if the next question is about typical visitor experience rather than typical visit duration.
How to analyze quantitative data
Begin with coverage and basic checks, then choose summaries that fit the measure. Inspect counts, distributions, missingness and unusual values before relying on one average. Add comparisons only when their populations, periods and definitions are sufficiently clear.
- Identify the unit and base. State whether a percentage refers to all eligible people, respondents or a matched subgroup.
- Describe the distribution. Use frequencies for categories and suitable summaries for measured quantities.
- Compare carefully. Examine differences in coverage and context as well as the result.
- Use methods suited to the design. Repeated observations, clustered sites and selected samples require appropriate treatment.
- Interpret within the evidence. Separate a descriptive pattern, an association and a causal estimate.
Quantitative research can investigate explanations and, with suitable designs and assumptions, estimate causal effects. It is inaccurate to say numbers can never address why something happened. It is equally inaccurate to treat a correlation or before/after difference as automatic proof of cause.
The quantitative data analysis guide develops the analysis choices further.
Can you average survey ratings?
For an individual ordered response item, the gap between adjacent categories is not automatically equal. Show the distribution of responses and explain the coding. If a mean is used as a practical summary, state that choice and avoid interpreting small differences as more precise than the measure supports.
A multi-item scale developed and evaluated for a particular purpose raises different measurement questions from a single satisfaction item. Follow the instrument's scoring guidance and consider whether the intended analysis is suitable. “All rating averages are nonsense” is too broad, just as “all numeric ratings are equal-interval measurements” is too broad.
Open comments can add context, but they do not repair an invalid calculation. Good interpretation requires both sound measurement and appropriate use of any additional evidence.
How does qualitative data fit?
Qualitative material can explore experience, meaning, processes and differences that a numeric summary misses. Quantitative evidence can describe patterns, test relationships and estimate quantities. Choose each source for the question it can help answer.
If ratings and comments come from the same response, keeping them connected can help review. If interviews involve a different group, explain how that evidence relates to the numerical findings. The sources do not always need to share a personal identifier.
A comment is an account, not a guaranteed cause. Look for variation and contradictory evidence rather than selecting one quotation to explain an entire group's result. See mixed-methods analysis for ways to integrate the findings.
Keep quantitative data comparable over time
Maintain a data dictionary with units, field meanings, response options, formulas and reporting periods. In a federated organization, agree on the few common measures needed for comparison while allowing local questions around them.
Collect stable registration context once where practical. Update changing information and retain relevant historical context. Document question changes and approved transformations. A field mapping cannot make visits equivalent to people, or a new scale automatically comparable to an old one.
Sopact's approach connects recurring collection, source context, reviewed analysis and governance. Evaluate it with the quantitative measures and related evidence your team needs. Verify calculations, joins, missing-data handling, permissions and export requirements rather than assuming the software makes every comparison valid.
How Sopact reduces coding and reporting work
When a numerical measure has an accompanying comment, keeping them connected makes later investigation easier. It does not require every quantitative analysis to include text.
A workflow with repeated manual work
- Define from an initial sampleRead material and agree on the codebook.
- Apply it across the datasetCode responses and check the result.
- Revise a definitionReturn to affected material and recode it.
- Reconnect the numbersReconcile coded results with ratings and context, then rebuild the view.
The Sopact workflow
- Your team owns the definitionsDecide what each code means and improve it as you learn.
- Apply coding across the eligible dataAutomate application; people review quality and exceptions.
- Reprocess after a definition changesReapply the revised definition across the configured scope instead of recoding each response by hand.
- Ask across coded text and numbersKeep the response, rating and relevant record context connected; inspect the evidence behind the result.
This compares workflow patterns, not a claim that every research tool requires manual coding or separate files. Some already automate parts of this work; compare the complete cycle.
For this codebook-based workflow, the main saving is repeated application and reconnection—not the removal of human judgment. A changed definition can be reapplied across the configured data while reviewers concentrate on quality, exceptions and interpretation. Coded text stays connected to the relevant ratings and context.
Count the recurring work in ownership cost. Include setup, coding, recoding after revisions, source reconciliation, review and reporting, plus your actual platform and processing expenses. A worked scenario of four cycles of 4,000 responses illustrates 272 fewer annual staff hours; it is an assumption-based example, not a customer benchmark. Existing automation, review needs and implementation effort can substantially change the result.
Adjust the workload assumptions and compare total effort →
A reliable assistant should calculate from the selected records and let a reviewer open the supporting evidence. Check the data scope, definition, denominator and access permissions. Reproducible arithmetic does not make every AI interpretation correct.
Watch: Why Qualitative Analysis Stays Small — And How to Scale It
See why revising a codebook creates repeat work, and how connected coding and quantitative analysis change that workload.
Watch related data-workflow explainers
These companions discuss the broader workflow around numerical and qualitative evidence. They are not substitutes for the measurement and statistical distinctions explained above.
Connected Data Intelligence: Why Qualitative Data Gets Ignored
Impact Measurement Software in 2026: What's Actually Changing
Report the result with its meaning
State the measure, population, period, coverage and calculation. Include appropriate uncertainty and explain material limitations. A precise-looking decimal does not compensate for an unclear measure or missing evidence.
For presenting results, use the How to Write an Impact Report guide and browse report examples.
Frequently asked questions
Is every number quantitative data?
No. Numeric identifiers and category codes are labels. Decide whether a field represents a quantity, an ordered response or a category before choosing a calculation.
What are the two main types of quantitative data?
Discrete data takes countable values, such as visits. Continuous data represents measurements such as time or distance, even when recorded with limited precision.
Is quantitative data always objective?
No. Question wording, instrument quality, selection, recording and interpretation can affect numerical evidence. A self-reported numeric rating still reflects a person's judgment.
Can quantitative data explain causes?
Suitable quantitative research designs can investigate causal questions. Routine associations or before/after differences alone do not establish causation; the design and assumptions matter.
Do quantitative results always need comments?
No. Additional qualitative evidence can be useful when it serves the question. A numeric finding may be meaningful on its own, and a comment does not automatically explain or validate it.
Choose your next step
ASME
Related practice: bringing stakeholder evidence into a membership organization’s review. The numerical examples in this guide are fictional.
Read the customer story →