Qualitative Data Analysis - Summarizing and developing themes using Infopickle software

Step-by-step guide to the analysis of Focus Group Interview data – Series 3

Focus Group Interview is a detailed process involving sequential set of tasks and skills to record, transcribe, translate diary, audio and video information to verbatim text. Focus Group Interview analyses can get very intense when number of studies or activities exceeds 12 and sometimes upto 75.

1.    One shortcut is Content Transcription, where transcribers summarize information as they listen to audio tapes. This leads to tremendous loss of information and introduces unforeseen bias in analysis.

2.    Second method is to manually sort verbatim text into an excel sheet. This manual method prevents loss of information but delays processing of data by a week.

3.    Third alternative is to use a qualitative data analysis tool such as Infopickle.

This article articulates how Qualitative Researchers can use outputs from Infopickle to summarize transcripts and develop themes. Infopickle will not do the analysis for Researchers nor will it derive insights from studies. However, Infopickle will scan, and extract key themes for researchers to synthesize and report. Infopickle facilitates uncovering of insights by structuring data in different formats using text analytics.

In this article, we provide an outline to guide Researchers through the seven phases of analysis, and offer examples to demonstrate the process. We use a beverage consumption study for illustration of outputs. Study used Focus Group and Interview across multiple groups such as cities, usership and demographics. Names have been changed to camouflage confidential details.

Step 1. Familiarising with moderator and interview responses: Data Dump

What is Data Dump: Collating all the moderator and respondent questions across all studies in one master worksheet. Data dump eases Researcher’s reading of transcript as lengthy moderator questions have been truncated and filler words such as Thank You eliminated. Researchers become acquainted with the consumer responses. It is ideal to quickly browse through the data dump at least once before Researcher’s begin their analyses.

How to interpret Data Dump: Background profile of respondent is mainly married, staying with parents, having no or 1 kid, working in banks or business in commodity trading. Quickly browsing of data revealed that diary data was not merged for Beverage_1b. Concepts were shown with the right rotation.

Step 2. Identifying recurring patterns: Theme and Sentiment Codebook

What is Codebook: Reading and re-reading of transcripts to identify recurring patterns is time consuming. Infopickle scans all consumer response to detect frequently occurring topics. Themes are words, phrases, sentiment that summarizes a sentence. This is done by extracting all relevant keywords and their corresponding quotes or extracts for each theme.  Theme and Sentiment codebook are the bedrock of analysis. Codebook organizes transcripts into meaningful groups of themes and sentiment. Researcher can select, deselect, prioritize or combine group of themes and sentiment. Sentiment codebook is similar to Theme codebook, except the focus is in retrieving recurring opinions such as likes and dislikes. The main purpose of using theme and sentiment codebook is to encourage focused reading of transcripts and accelerate time to insights. Codebook creates max of 200 potential themes/patterns and is arranged alphabetically. Researchers based on their experience and study objectives can decide which themes are interesting and in what context do they occur.

How to interpret Codebook: For this beverage study, client wanted to assess if packaging was attractive, drink was clean and clear and how comfortable would they be in drinking in various occasions. Respondents not only perceived the opaque bottle attractive but even the colour of drink. Clarity of drink as defined by clean and clear, was associated with premium and intricate gastronomy.

Step 3. Selection and organization of responses by Analysis Headers: Data Sorting

What is Data Sorting: Respondent’s comments need to be mapped to each interview protocol question(s). Problem is different interviewer/transcriber would have asked the same question differently or will not be present in the same order within the same study. Infopickle assists in identifying common questions across multiple transcripts.

How to interpret Data Sorting: All responses for Socializing Changes across all transcripts are in the same row. For all the transcripts, respondents had the same group of friends and colleagues. On weekdays drinking is with colleagues, and weekends is with friends. Group size has reduced over time. Newly working group would like to try new places every weekend. Women group may be non-alcoholic and Businessmen prefer to drink at home.

Step 4. Similarities and Differences across studies and activities: Auto Summarisation

What is Auto Summarization: For large studies, finding similarities and differences across groups especially during subsequent re-reading is helpful. Researchers can analyze the data by city, gender, frequency of usage. Also, for a given segment, they can read only the key themes instead of re-reading the whole transcript. Key themes summarize transcript information for easier re-reading.

How to interpret Key Themes: Researcher clicks on Mumbai, which opens the key theme file. For analysis row Drinking Evolution, key summary information is as follows. Male, Student, Age 25-35 staying in Mumbai started drinking beer and breezer followed by hard drinks. On the other hand, Age 35-45, Make, working in Mumbai started with saline and cocktails.

Step 5. Combine themes and retrieve relevant verbatims from transcripts: Content Analyses

What is Content Analysis: For each protocol question, Infopickle retrieve all the sentences related to any combination of keywords and themes. Combination can contain exact words such as color or taste. Combination can also contain complex logic such as Apple AND Cider AND Vinegar.  Content Analyses is useful when Researcher’s need to synthesize various themes for report writing or rearrange themes into a logical order. Content Analyses will help them bridge gaps in report writing or support Researcher’s point-of-view.

How to interpret Content Analysis: Any Alcohol retrieves all sentences containing any mention of alcohol. Beer was mentioned more often and is associated with craft, bitterness and fruity.

Step 6. Number of times a keyword occurs is a sign of importance? Keyword Analyses

What is Keyword: Keyword output contains single, dual and triple occurring words in the same sentence. Though qualitative data is not about numbers, frequency data assists in providing an overview on the topics mentioned in each section or analyses headers. Most researcher’s create word cloud based on frequency count data.

How to interpret Keyword: For views on current drinking, apple, cider and vinegar was mentioned most often. Beer was mentioned with rum very frequently. Spontaneous brand associations were Bay leaf, Black label, and Antiquity.

Step 7. Graphical analysis of project techniques: Perceptual Maps

What is Perceptual Map: Projective techniques such as mood boards, picture cards and word associations need to mapped graphically. There are multiple layers of analysis – spontaneous or prompted, order in which they are mentioned, and co-occurrence of words..

·How to interpret Onion Peel: Quick summary of word association data in report writing. was mentioned thrice and Sour was mentioned twice spontaneously. Light drinks such as Champagne and Wine are mentioned only when prompted.

Summary

Qualitative research is about identifying needs gaps, uncovering consumer problems and insights. But there is an increasing need to make insights more concrete and actionable. This necessitates Researchers to view the data from multiple angle. Software tool such as Infopickle provides an opportunity to dig deep into consumer conversations that would have been ordinarily missed by naked eye.

Milind holds a doctorate for Marketing Sciences from IIM Bangalore and is a founder and Chief Analytics Officer at Unpickle <milind.kelkar@unpickle.in>. Unpickle’s flagship platform Infopickle reads, interprets and categorizes text data for researchers and analysts. If you require more information about Infopickle please get in touch and we will be happy to answer your questions.