Let AI Assist boost your literature review and analysis
As you may have noticed, there is a rapid growth in AI-based tools for all types of software packages. We followed this trend by releasing AI Assist – your virtual research assistant that simplifies your qualitative data analysis. In the following, we will present you the tools and functions of AI Assist and how they can facilitate your literature reviews.
Literature reviews are an important step in the data analysis journey of many research projects, but often it is a time-consuming and arduous affair. Whether you are reviewing literature for writing a meta-analysis or for the background section of your thesis, work with MAXQDA! Besides the classic tools of MAXQDA that can facilitate each phase of your literature review, the new tool AI Assist can boost your literature review and analysis in multiple ways.
How to use AI Assist for your literature review
Year by year, the number of publications grows in almost every field of research – our insights and knowledge likewise. The drawback is that the number of publications might be too high to keep track of the recent developments in your field of research. Consequently, conducting a proper literature review becomes more and more difficult, and the importance of quickly identifying whether a publication is interesting for your research question constantly increases.
Luckily, MAXQDA’s AI Assist tool is here to help. Among others, it can summarize your documents, text segments, and coded segments. But there is more – based on your coded segments AI Assist can generate subcodes suggestions. In the following, we will present you step-by-step instructions on how to use MAXQDA for your literature review and analysis with a special focus on how AI Assist can support you.
Step one of AI for literature reviews: Identify and import your literature
Despite the fact that MAXQDA and AI Assist can facilitate your literature review and analysis in manifold ways, the best advice is to carefully plan your literature review and analysis. Think about the purpose of your literature review and the questions you want to answer. Develop a search strategy which includes, but is not limited to, deciding on literature databases, search terms, and practical and methodological criteria for selecting high-quality scientific literature. Then start your literature review and analysis by searching the identified databases. Before downloading the PDFs and/or bibliographic information (RIS), briefly scan the search results for relevance by reading the title, keywords and abstract. If you find the publication interesting, download the PDF, and let AI Assist help you determining whether the publication falls within the narrower area of your research question.
MAXQDA’s import tab offers import options dedicated to different data types, such as bibliographic data (in RIS file format) and PDF documents. To import the selected literature, just click on the corresponding button, select the data you want to import, and click okay. Alternatively, you can import data simply by drag-and-dropping the data files from your Windows Explorer/Mac Finder window. If you import full texts and the corresponding bibliographic data, MAXQDA automatically connects the full text to the literature entry with an internal link.
Step two of AI for literature reviews: Summarize your documents with AI Assist
Now that you have imported all publications that might be interesting for your research question, it is time to explore whether they are indeed relevant for your literature review and analysis. Before the release of AI Assist, this step typically took a lot of time as you had to go through each paper individually. With the release of AI Assist, MAXQDA can accelerate this step with AI-generated summaries of your publications. For example, you can create AI-generated summaries either for the entire publication or for each chapter (e.g. Introduction, Methods, Results, and so on) individually and base your decision about a paper’s relevance on these summaries. Each AI-generated summary is stored in a memo that is attached to the underlying document or text segment, respectively.
Summarizing text segments with AI Assist just takes a few clicks. Simply highlight a text segment in the Document Browser and choose AI Assist from the context menu. Adjust the settings to your needs and let OpenAI do the work for you. To view and edit the summary, double-click on the yellow memo icon attached to the summarized text passage.
Adjust settings for summarizing text with AI Assist for literature reviews
Step three of AI for literature reviews: Determine relevance and sort accordingly
Instead of reading the entire paper, you can use the AI-generated summaries to determine whether a publication falls within the narrower area of your research question. To do so, it might be helpful to view all memos containing summaries of a specific publication at once. Of course, this is possible with MAXQDA. Go to the Memo tab, click on (In-)document Memos, and click on the publication’s name to view only the AI-generated summaries related to this document. It is important to note that AI-generated summaries are not perfect yet. Therefore, it is advisable to read the entire paper in cases where you have doubts or can’t decide whether the publication is relevant.
Depending on the number of publications in your MAXQDA project, you might want to sort your documents in document groups, for example, based on the relevance for your research question or the topics discussed in the paper. You can easily create a new Document group by clicking on the respective icon in the Document System window. Documents can be added simply via drag-and-drop. Alternatively, you can create Document Sets which are especially helpful when you want to sort your documents by more than one domain (e.g. by relevance and methodology used).
Sort documents in document groups according to their relevance using AI for literature reviews
Step four of AI for literature reviews: Reading and rough coding
Now that you have identified the publications important to your project, it is time to go through the documents. Although, AI Assist can support you at multiple stages of your literature review, it can’t replace the researcher. As a researcher, you still need a deep understanding of your material, analysis methods, and the software you use for analysis. As AI-generated summaries are not perfect yet, you might want to improve the summaries, if necessary, or add information that you consider especially important, e.g. participants’ demographics.
In a next step, it is time to create and apply some codes to the data. A code can be described as a label used to name phenomena in a text or an image. Depending on your approach, you might already have codes in mind (deductive coding) or you plan to generate codes on the basis of the data (inductive coding). No matter your approach – you can use MAXQDA’s advanced tools for coding. In many cases it is best, to start your first round of coding with rather rough codes that you can refine in a later step using the help of AI Assist. You can create codes in the Code System window by clicking on the plus-icon or in the Document Browser by highlighting a text segment via the context menu or the corresponding icons. A code can be applied to the data via drag-and-drop.
Reading and rough coding for AI for literature reviews
Step five of AI for literature reviews: Confirm your initial codings
Though AI Assist can’t validate your codings like a second researcher using intercoder agreement, AI Assist’s Code Summaries can help you to identify whether you have applied the code as intended. The AI-generated Code Summary is a summary of the content of all text segments coded with the corresponing code. This summary might give you an idea of how you have applied the code and if the coded text segments indeed contain what you had in mind when creating the code.
To create a summary of coded segments with AI Assist, simply right-click the code of interest in the Code System and choose AI Assist > Code Summary from the context menu. Adjust language and the summary length to your needs and let AI Assist do the summary for you. As for document summaries, the summary will be stored in a memo which is placed next to the code in the Code System. If the summary doesn’t match your code definition, you might want to review the coded segments and adjust your codings accordingly. By double-clicking on a code, you open the Overview of Coded Segments – a table perfectly suited to go through the coded segments and adjust or remove the codings.
Confirm your initial codings with AI Assist’s Code Summary for literature reviews
Step six of AI for literature reviews: Refine your code system
In case you have applied rather rough codes to your data, your code definitions are probably too broad for you to make sense of the data. Depending on your goals, you might wish to refine these rather broad codes into more precise sub-codes. Again, you can use AI Assist’s power to support this step of your literature review. AI Assist analyzes the text and suggests subcodes while leaving the decision on whether you want to create the suggested sub-codes up to you.
To create AI-generated subcode suggestions, open the context menu of a code and choose AI Assist > Suggest Subcodes. Besides selecting a language, you can ask AI Assist to include examples for each subcode as a bullet list. Like the AI-generated summaries, the code suggestions are stored in the code’s memo. If you are satisfied with the code suggestions, you can create and apply them to your data. Alternatively, you can use the AI-generated code suggestions to confirm the subcodes that you have created.
Use AI Assist’s Suggest Subcodes function to refine your
code system for your literature reviews
Step seven of AI for literature reviews: Analyze your literature
Now that you have coded your literature, it’s time to analyze the material with MAXQDA. Although you can use plenty of MAXQDA’s tools and functions even when the material is not coded, other tools require coded segments to be applicable. MAXQDA offers plenty of tools for qualitative data analysis, impossible to mention all. Among others, MAXQDA’s Overview and Summary Tables are useful for aggregating your data. With MAXQDA Visualization Tools you can quickly and easily create stunning visualizations of your data, and with MAXQDA’s Questions-Themes-Theories tool you have a place to synthesize your results and write up a literature review or report.
You can find more information and ideas for conducting a literature review with MAXQDA, here:
Learn more about literature reviews
For information about AI Assist and how to Activate AI Assist, visit:
Literature about literature reviews and analysis
We offer a variety of free learning materials to help you get started with your literature review. Check out our Getting Started Guide to get a quick overview of MAXQDA and step-by-step instructions on setting up your software and creating your first project with your brand new QDA software. In addition, the free Literature Reviews Guide explains how to conduct a literature review with MAXQDA in more detail.
