Getting Started with Qualitative Analysis

Have you ever found yourself staring at a mountain of qualitative data, feeling like you're lost without a map? Whether it’s transcripts from interviews or focus groups, written responses from a survey or assessment, or entries in a diary study, figuring out where to start can feel daunting. 

Luckily, there are steps you can take at the beginning of your analysis process to help you think through your approach and kickstart your momentum. 

Let’s get started!

Consider the unit of analysis

One of the first things you need to decide is the unit of analysis—that is, the unit by which you are comparing one piece of data with the whole dataset to draw conclusions about trends and patterns. Some examples of common units of analysis we use are:

• Individuals (e.g., one person’s response in an interview, survey, diary study prompt)

• Pairs or small groups (e.g., two or three adult visitors who participated in an interview together) 

• Focus groups (e.g, a group of 8-10 people in a facilitator-led discussion)

Sometimes your unit of analysis is very clear—for example, if you have done 30 one-on-one interviews, the unit of analysis will likely be individual interview participants. In other words, each person’s response counts as one out of 30 responses. 

However, if you did interviews with pairs or small groups, you will need to decide if the pair or group together counts as the unit of analysis or if each individual in the group will be counted as a unit (if it is the latter, you need to make sure that each person in the group answers each question during data collection!). 

In the case of focus groups, the individual participants’ responses are not typically the unit of analysis, but rather the group as a whole. You are looking for areas of consensus around an idea (if there is one) or the important distinctions in ideas or opinions among groups.

In this example, each color represents one unit of analysis. Although the number of individuals is different in each approach, the unit of analysis is 5 in all cases

In truth, figuring out your unit of analysis should be part of the research design phase. This ensures that you design your questions and collect your data with the unit of analysis in mind.  However, it is also true that sometimes your unit of analysis may change based on the realities of data collection; for instance, if you planned for one-on-one interviews but some people ended up interviewing in small groups. Sometimes it takes reading through your data to get a sense of the unit of analysis that will work best for your research goals and study design.

Plan your coding approach

Coding is the term for beginning to make meaning from your data by naming categories for common ideas, themes, or attributes that arise. There are many different approaches to coding, and your approach may vary depending on your study objectives, the types of questions you asked, and your sample size. In cases where you are providing a sense of most and least prevalent themes, similar to deciding your unit of analysis, it is also helpful to consider whether you will code responses to a given question into a single category or allow responses to be coded into multiple categories. 

For example, you might ask a visitor to indicate the first place they visited in the museum—in this case there will be one clear answer from each person that fits into a single category (e.g., the admissions desk, the bathroom, etc.). However, you might also ask this visitor what motivated them to visit the museum that day, and their response is a bit more complex. Perhaps they say they wanted to show their friend a new exhibition that recently opened—in this case, the person is sharing multiple motivations (spending time with a friend and seeing a specific exhibition) so their response will get coded into multiple categories, as in the example illustrated below.  

In this example, I decided to code some response into multiple categories, while some answers fall into only one category.

There are other cases where coding into a single category versus multiple categories is an analytical choice that depends on how detailed or nuanced you want your results to be. You might decide this by balancing considerations like the study objectives (is a short and straightforward answer "enough" or is more nuance needed to thoroughly address the study objective?), audience (who is receiving the results and what level of detail is best for them?), or resources (do I have the time to parse out nuance with a deep analysis?).

Organize your data

Keeping things organized just makes life easier. The same is true in data analysis–it makes your work easier and much more efficient. It can make it simpler to identify major themes, pull quotations for reporting, or retrace your steps if you want to double check or revisit some of your analysis later on. I tend to organize my qualitative data simply, using either spreadsheets or word documents depending on my needs.

With interview data, write-in survey responses, and diary studies, I most often start by “chunking” my notes or transcripts into a spreadsheet, like in the example illustrated in the previous section. Chunking involves breaking down large pieces of information into smaller, more manageable pieces. You might chunk data by question or by study objective. Each unit of analysis (e.g., one person’s responses in a one-on-one interview) is a row in the spreadsheet, and the columns represent various codes you’ve developed to categorize responses. This approach helps you quickly calculate totals for each code and understand the proportion of responses that fall into each code relative to one another.

Alternatively, when analyzing data that is highly detailed and descriptive, like focus group data or extended interviews, I tend to work directly within the transcript in a word document. I assign colors to particular coding themes and highlight quotations that align with that code. Using color in data organization can help you visualize codes that are prominent in your transcript and where they arise in relation to certain questions you’ve asked the group (e.g., there is tons of blue highlighting across the whole transcript, telling me this idea is very important). 

In this example using a focus group transcript, I created a color-coding legend that appears in the header on each page for easy reference to remind me of each code and related color. I highlighted quotations directly in the transcript that fit each code.

Get started!

Once you’ve thought through your unit of analysis, coding approach, and organized your data, it’s time to jump in. This is the most exciting part to me—seeing trends start to emerge and parsing out the commonalities, differences, and surprises in the data. 

What will you learn from your qualitative data?