Guest post by Professional MAXQDA Trainer Dr. Daniel Rasch.
Qualitative text or document analysis has evolved into one of the most used qualitative methods across several disciplines (Kuckartz, 2014 & Mayring, 2010). Its straightforward structure and procedure enable the researcher to adapt the method to his or her special case – nearly to every need.
This article proposes a recipe of ten simple steps for conducting qualitative document analyses (QDA) using MAXQDA (see table 1 for an overview).
|1 – Define the research question||What are you trying to find out?|
|2 – Collect and sample the data||What kind of data will best answer your RQ? Interviews, documents, surveys? Collect the data and sample it in a suitable and valid way.|
|3 – Select and prepare the data for QDA||Select the fitting data and prepare it for QDA: e.g. transcription of interview data, selecting important parts of documents, …|
|4 – Codebook development||Develop a solid codebook (if needed).|
|5 – Unitizing and coding instructions||Unitize the data and set rules for coding.|
|6 – Trial, training, reliability||Test the codebook and if necessary, train other coders. If applicable, test coding reliability.|
|7 – Revision and modification||Revise the codebook if necessary and modify the coding instructions.|
|8 – Coding||Code the rest of the data using the revised codebook.|
|9 – Analyze and compare||Run your analysis: what intersections are important? What patterns are there? What distributions are worth noticing? What did you learn in regard to the research question?|
|10 – Interpretation and presentation of findings||Interpret and present the data in a suitable way and be transparent when reporting the findings.|
Table 1: Overview of the “QDA recipe”
The ten steps for conducting qualitative document analyses using MAXQDA
Step 1: The research question(s)
As always, research begins with the question(s). Three aspects should be covered when dealing with the research question(s):
- What do you want to find out exactly,
- what relevance does your research on this exact question have, and
- what contribution is your research going to make to your discipline?
Highlight these questions in your introduction and make your research stand out.
Step 2: Data collection and data sampling
After you have decided on the questions, you should think about how to answer them. What kind of qualitative data will best answer your question? Interviews – how many and with whom? Documents – which ones and where to collect them from?
At this point, you can already start thinking about validity: are you going to use a representative or a biased sample? Check the different options for sampling and its effects on validity (Krippendorff, 2019).
Step 3: Select and prepare the data
For this step, MAXQDA 2020 is an excellent tool to help you prepare the selected data for any further steps. Whatever type of qualitative data you choose, you can import it into MAXQDA and then you can have MAXQDA assist in transcribing it. In the end, qualitative document analysis is all about written forms of communication (Kuckartz, 2014).
Figure 1: Import the data you have chosen or selected
Step 4: Codebook development
It takes time to develop a solid codebook. Working deductively, the process is a little easier with codes deriving from the theoretical considerations in the context of your research. Inductively, there are various steps you can use, ranging from creative coding to in-vivo-codes.
Content-wise, you can apply all sorts of codes, such as themes or evaluations, two of the most commonly used styles of content analysis (see thematic and evaluative content analysis in Kuckartz, 2014).
Figure 2: coding options in MAXQDA
Figure 3: Creating a new code with code memo
Step 5: Unitizing and coding instructions
Before the process of coding starts, it is necessary to decide on the units of, as well as the rules for, coding. It is especially important to decide on your unit of coding (sentences, paragraphs, quasi-sentences, etc.). Coding rules help to keep this choice consistent and support you to stick to your research question(s) because every passage you code and every memo you write should be done in order to answer your research question(s). Decision rules should be added: what are you going to do if a passage does not fit in your subcodes but should be coded because it is important for your research question?
Step 6: Trial, training, reliability
Trial runs are of major importance. Not only do they show you, which codes work and which do not, but they also help you to rethink your choices in terms of the unit of coding, the content of the codebook, and reliability. Since there are different options for the latter, stick to what works best for you: either a qualitative comparison of what you have coded or quantitative indicators like Krippendorff’s alpha if need be.
You can test yourself or a team you work with and there might even be some situations, where a reliability test is not helpful or needed. When testing the codebook, be sure to test the variability of your collected documents and be sure that the entire codebook is tested.
MAXQDA helps you compare different forms of agreement for more an unlimited number of texts, divided into two different document groups (one document group coded by coder 1, a second document group coded by coder 2 – be aware, that you can also test yourself and be coder 2 yourself).
Figure 4: Intercoder agreement
Step 7: Revision and modification
After checking, which codes work and which do not, you can revise the codebook and modify it. As Schreier puts it: “No coding frame (codebook – DR) is perfect” (Schreier, 2012: 147).
Step 8: Coding
There are many different coding strategies, but one thing is for sure: qualitative work needs time and reading, as well as working with the material over and over again.
One coding strategy might be to first make yourself comfortable with the documents and start coding after second or third reading only. Another strategy is to concentrate on some of your codes first and do a second round of coding with the other codes later.
Step 9: Analyze and compare
Analyze and compare – these two words are the essence of the qualitative analysis at this step. At the core of each qualitative document analysis is the description of the content and the comparison of these contents between the documents you analyze.
After everything has been coded, you can make use of different analysis strategies: paraphrase, write summaries, look for intersections of codes, patterns of likeliness between the documents using simple or complex queries.
Figure 5: different analysis strategies in MAXQDA
Step 10: Interpretation and presentation
Reporting and summarizing qualitative findings is difficult. Most often, we find simple descriptions of the content with the use of quotations, paraphrases or other references to the text. However, MAXQDA makes it fast and easier with many options to choose from. The easiest way is to generate a table to sum up your findings – if your data or the findings allow for this.
MAXQDA offers several options: either map relations of codes, documents or memos with the MAXMaps, create matrices between codes and documents (Code Matrix Browser) or codes and codes (Code Relations Browser) to display the distribution of codes inside your data or even using different colors to map the distribution of codes or single documents.
Figure 6: Visual Tools for presentation
The Code Matrix Browser also enables you to quantify the qualitative data using two clicks. You can export these numbers for further analysis with statistical packages, to run causal relation and effect calculations, such as regressions or correlations (Rasch, 2018).
Summary and adoption
Qualitative document analysis is one of the most popular techniques and adaptable to nearly every field. MAXQDA is a software tool that offers many options to make your analysis and therefore your research easier.
The recipe works best for theory-driven, deductive coding. However, it can be also used for inductive, explorative work by switching some of these steps around: for example, your codebook development might be one step to do during or after the trial and testing, since codes are developed inductively during the coding process. Still, it is important to define these codes properly.
The above-mentioned recipe has been used as a basis for several publications by the author. Starting with simple comparison of qualitative and quantitative text analysis (Boräng et al., 2014), to the usage of the qualitative data as a basis for regression models (Eising et al., 2015; Eising et al., 2017) to a book using mixed methods and therefore both qualitative and quantitative data analysis (Rasch, 2018).
About the author
Daniel Rasch is a post-doctoral researcher in political science at the German University of Administrative Sciences, Speyer. He received his Ph.D. with a mixed methods analysis of lobbyists‘ success in the European Union. He focuses on the quantification of qualitative data. He is an experienced MAXQDA lecturer and has been a Professional MAXQDA Trainer since 2012.