Qualitative vs quantitative research: the key differences

The two differ in the evidence they collect and the question they answer.

quantitative researchnoun

Research that collects numerical data and analyses it with statistics. It measures, counts and compares, and usually asks how many, how much or how strongly.

qualitative researchnoun

Research that collects non-numerical data, such as words, images and observations, and interprets it for meaning. It usually asks why, how and what it is like.

The psychology site Simply Psychology puts it in one line each: quantitative research "collects numerical data and analyzes it using statistical methods", while qualitative research "gathers non-numerical data (words, images, sounds)" to explore experiences and meanings.

For Anjali, the quantitative question was easy to write: what share of girls in rural government schools of the district choose each stream in Class 11? The qualitative question was harder and more interesting: how do girls and their families decide which stream a girl should take?

Read this table as tendencies, not fixed laws.

Underneath sits a difference in how each sees the world. An open nursing research textbook hosted by Thompson Rivers University, in Canada, links quantitative research to realism, the view that there is one truth to be discovered, and qualitative research to constructionism, the view that people create meanings and that there can be many realities.

Dr. Verma drew it more simply on the back of Anjali's questionnaire. On one side she drew a small bar chart. On the other she wrote a sentence a girl had once said to Anjali: "Science means tuition in town, and the last bus comes at five." The chart tells you how big the problem is. The sentence tells you what the problem is.

The same topic can be studied either way. What changes is the question you ask about it.

Quantitative questions use words such as how many, relationship and difference, because the answer is a number. Qualitative questions use how, why and experience, because the answer is an explanation. If your question is still vague, our guide on how to write a good research question helps you sharpen it first.

A quantitative study usually starts with a hypothesis, a prediction it can test. A qualitative study usually starts with an open question and lets the answers shape what it finds. Our guide on how to write a hypothesis covers the first kind.

Simply Psychology lists interviews, focus groups, observation, diaries, case studies and ethnography as typical qualitative methods, and experiments, surveys, psychometric tests and rating scales as typical quantitative ones. Our blog on data collection methods in research explains each of these in detail.

Some tools can give either kind of data, depending on how you build them. A survey is the best example. The library of Shenandoah University, in the United States, notes that surveys with fixed answer choices support quantitative research, while surveys with open-ended questions can support qualitative research.

A rating scale like Anjali's forty statements, called a Likert scale, is usually treated as quantitative. But the gap between "agree" and "strongly agree" is not a measured distance, so some researchers argue its scores are not exact measurements. Ask your guide which tests suit such data.

The analysis differs just as much. Quantitative data is first summarised with descriptive statistics, such as percentages, means and standard deviations, and then tested with inferential statistics, which ask whether a pattern in the sample is likely to hold in the population. Qualitative data is usually transcribed, read many times and coded: short labels are attached to pieces of text, and similar codes are grouped into themes. Thematic analysis is a common approach; grounded theory and content analysis are others.

A quantitative study wants its sample to stand for a whole population, so it usually needs more people and, ideally, chance in choosing them. A qualitative study wants people who can tell you a lot, so it chooses them on purpose and works with fewer. Our guide to sampling methods in research explains both families.

How few is enough? Many qualitative studies keep going until saturation, the point where new interviews stop bringing new ideas. Guest and colleagues, in a 2006 study, found that saturation was reached within the first twelve interviews of their study. Braun and Clarke, writing in 2021, questioned whether any fixed number like this means much at all, since it depends on the study. So do not write "12 interviews are enough" in your thesis. Explain how you decided when to stop.

The two are also judged by different standards.

The right-hand column comes from Lincoln and Guba (1985), who proposed these four criteria of trustworthiness as rough parallels to reliability and validity. In a qualitative methodology, say how you met them: for example, by transcribing interviews word for word and keeping a diary of your decisions.

You do not always have to choose. Mixed methods research collects both kinds of data in one study and brings them together in the analysis. The textbook Research Design by John Creswell and J. David Creswell describes three core designs.

1. Convergent design.Collect quantitative and qualitative data in the same phase, analyse each separately, then compare the two to see whether they agree.
2. Explanatory sequential design.Run the quantitative part first, then use qualitative data to explain what the numbers showed. Survey first, interviews after.
3. Exploratory sequential design.Start qualitative to understand the topic or build a tool, then test it with a quantitative phase. Interviews first, survey after.

Mixed methods doubles the work: two kinds of data, two kinds of analysis, and a chapter that shows how they fit. Use it only when your question really has two parts.

Anjali's question had two parts, so Dr. Verma agreed to an explanatory sequential design. First, a short questionnaire for Class 11 girls in a random sample of rural schools, asking which stream each girl chose and rating a few reasons. Then interviews with girls and parents, chosen from the survey to include girls who had scored well in science but taken arts.

The survey tells you where to look. The interviews tell you what you are looking at.

Do not begin with "I want to do a survey". Begin with your question, and ask these in order.

1. What does your question ask?If the answer is a number, a difference or a relationship, it is quantitative. If the answer is an explanation, an experience or a process, it is qualitative.
2. How much is already known?A new or little studied topic often needs qualitative work first, to find out what matters. A well studied topic with good measuring tools suits quantitative work.
3. Do you need to speak for a population?A claim about all girls in the district needs a quantitative sample. A deep account of a few families does not.
4. What can you do well?Statistics needs training and enough participants. Good interviews need patience and time to transcribe. Be honest about your skills and time.
5. What does your department expect?Some departments and journals favour one approach. Read a few recent dissertations from your department and ask your guide before you design anything.

Once you choose, name the approach in your methodology and give one sentence that ties it to your question. Then describe the design, sample, tools and analysis in the way our guide on how to write a research methodology shows.

• Calling a study "qualitative" because it has a few open questions at the end of a survey.
• Reporting percentages from eight interviews as if they spoke for a population.
• Choosing mixed methods only because it sounds stronger, then doing neither part properly.

Anjali rewrote her proposal in a week. The forty statements became twelve, the 300 girls became a proper random sample of schools, and a second chapter plan appeared for the interviews. When she showed it to Dr. Verma, the guide took out the old questionnaire, with the sentence about the last bus still written on its back. "Now," she said, "you have a way to find out how many girls that sentence is true for."