Hello!

An exciting aspect of scientific research is the opportunity to advance knowledge by building on the findings of different experiments. In this regard, an essential step of a new project is to explore what has been achieved in previous studies, generally by surveying the literature available, to understand where and how a meaningful contribution can be made. Nonetheless, certain practices can make this task more efficient, ensuring sufficient literature coverage and facilitating the retrieval of relevant information. Given that I am currently carrying a literature review for my research project, in this post, I would like to share some of the practices that have been particularly useful in this process.

“There is not a discovery in science, however revolutionary, however sparkling with insight, that does not arise out of what went before.” - Isaac Asimov

1. Defining a topic and related keywords

2. Exploring literature databases

3. Structuring the results

Defining a topic and related keywords

The first step is to narrow down a topic to investigate. Often, the work may involve a combination of multiple domains and it can be unclear what are the bottlenecks worth exploring. In this regard, a top-down approach has helped me to brainstorm ideas, starting from general concepts to specialised areas I am working on and iteratively identify points of interest. This process can be done independently or, if available, through discussions with colleagues with some knowledge on the subject who may provide relevant inputs.

Furthermore, I found it extremely valuable to note down the main keywords I encountered as I explored the literature. Indeed, journal articles typically include a set of keywords, and these can serve as inspiration to expand our scope and the coverage of the central topic.

Exploring literature databases

Next, when delving into scientific databases (such as Scopus, Pubmed, Web of Science, Google Scholar, etc), I would like to reiterate the importance of learning how to carry out an “Advanced Search” on these platforms. In short, this operation allows customising in detail how the search should be performed over a wide range of input fields. Overall, I felt that taking the time to explicitly define the criteria to use when retrieving information for certain search items yielded much more targeted results that I could review entirely and effectively, as opposed to initial searches under broad concepts that had thousands of results.

At the same time, it is important to leverage special characters and logical operations available when providing search inputs. For instance, the star operator “*” is a powerful character to cover all variations of a word: for instance, “imag*” would include the terms “image” and “imaging” without having to enter them explicitly in the search field.

Most databases share approximately the same concepts, even if the exact flags for each field differ slightly. In any case, I would still recommend carrying similar searches on different databases to maximise the chances of identifying all relevant data, as it could be that some journals may not be indexed in some of the databases.

Additionally, given the speed at which new articles are produced, I strongly recommend keeping a record of the search inputs, dates when these were performed, and the results obtained. Indeed, another feature of many databases is the possibility to download our results with a customizable set of fields and in several formats leaving no excuses to not keep the findings properly organized in a folder or in a reference manager software. Overall, not only this makes the search more reproducible for others reviewing the work but also helps to keep track of which areas have been covered and potentially identifying new combinations of inputs worth exploring.

Structuring the results

After collecting several results from the databases, it is time to review the literature and, at this point, it has likely become clear what type of information we aim to collect. Personally, I like the format of the export feature from Scopus, which gives precise control of the fields that are saved to the output file. Typically, I save the search results to a CSV file which I use as a baseline on Excel to add additional columns of interest where I fill information from each paper as I go through them.

Nonetheless, the total number of results is likely still too high and impossible to analyse in detail within a reasonable timeframe. Therefore, I follow a stepwise approach using a small set of clear inclusion/exclusion criteria that allow me to score the relevance of the paper to the main research question. As I go through the results, I check the paper for these details and fill my findings and score on the Excel table. Ultimately, this stage yields a smaller, workable subset of papers that create the basis for an extensive analysis. I find it useful to download these papers and implement a convenient naming structure to facilitate retrieving them when needed (in my case, I use the following: “Year – Surname of the first author – Beginning of the title”).

The longest phase of the literature review is to carefully examine this set of articles. Even though we can always expand our search goals as we proceed, I find it very useful to spend some time identifying all the relevant items to collect (which essentially become additional columns of the Excel table) before starting the analysis. From my experience, this detailed review can be extremely tiring and, therefore, it is highly advisable to spread it over several days while also alternating with other tasks, if possible.

Ultimately, this step should provide a detailed overview of the literature available and help to answer our initial research question.

Conclusion

These are some of my suggestions to carry out a literature review and I would reiterate that the ideas presented here are based exclusively on my personal experience. There are many resources available with more formal approaches to perform a successful literature review, and which I highly recommend checking (for example, see here or here).

I would also like to add that the results of such a literature search are typically organised in a formal literature review article or chapter, which is what I am currently preparing. Therefore, I thought it made more sense to separate these two phases in different posts and focus solely on the search here.

Please feel free to share your thoughts and experiences about literature reviews!

Have a great day!