How do you search for literature for your meta-analysis?

One key aspect I'd include is establishing the inclusion and exclusion criteria for the studies you want to include. These criteria should be specific and well-documented. Any assumptions made around the question is critical to be clearly laid out early on.

It's critical to remember that research is an iterative process, rather than a linear one. So you will need to engage deeply with the literature in order to define your research question. Thinking is another critical element! So take time to think about what you read, consider how it relates to other articles that you have read, and then think about how you can craft a research question that positions your research within the scholarly conversation happening in your research area. The literature is a conversation between scholars, and you should position your question in a way that contributes to this scholarly conversation.

Adding a small note on what has been already mentioned. Sometimes, during the “iterative process” of “engaging deeply with the literature” so to define the research question, researchers without a predefined research question (correct identification of a problem) might unknowingly bias the search towards the studies that align with a certain problem and neglect studies with contrasting findings. I would say that the researchers should not be biased by the existing findings and lose their own creativity (eg. construct research question and design it in order to align with what they are already familiar from literature review at this early stage). From my experience, I have seen often this unknowingly bias among the entry-level researchers.

As I see it, research comes in two forms: Exploratory and focussed. In exploratory research, one is simply checking things out. Picking up a new object (say) and looking at from different perspectives an orientations. Such exploration might, then, lead to a new hypothesis that, in a more focussed analysis, can be scrutinized and tested, over and over again.

If we define research border then writing scientific papers, that tends to get to much focus, I really think we benefit to start broadly. Start with understanding what you really want to achieve. Who has done similar things before and what can you learn from them? What are you looking to change in the long run? Then do your PICO.

Do keep in mind that the ideal of reproducible research limits the type of grey literature one can include. The search must reproducibly identify the same grey literature article every time regardless of who executes it. This make the current tools such as Google that "personalize" results as poor choices. One therefore has to be able to perform the search using the platform (arXiv, etc.) search tools (and hope that the platform API does not change).

Learning to use your institution's databases well is an important skill to practice and develop. I always recommend trying to find one high quality journal article that relates directly to your research topic. Once you have that, you can 'look back' to find other relevant sources i.e. look through the reference list to see what the authors of the good paper have cited. Another useful thing is to 'look forward' to find additional sources: this means using a platform like Google Scholar or Scopus to see who has cited your key article since it was published. People citing the article are also working on related topics so these articles may also be of interest to you. Just be sure to check the quality of the articles you choose to use yourself

Generally I think AI is a bit over hyped, but using it to find new sources in different languages and places have really been nice. Then due diligence needs to bee done preferably in an open way, the replication crisis's needs to be smaller.

ResearchGate is a community wherein one can interact with other researchers working on similar subjects. You can inform them of your niche and possibly get collaborations on the project.

An under appreciated way is to look at highly sighted papers and figure out the sources it builds on and then also going upwards from them.

Striking a balance! Developing a search strategy is a balancing act of ensuring your search is neither too broad nor too narrow. Starting off too broad may yield an overwhelming amount of irrelevant papers, whereas too narrow a focus could cause you to miss valuable results. The key is to start broad and incrementally narrow your search as you become more familiar with the literature base and the core papers you anticipate capturing. Some tips for this process: ✅ Keep a detailed record of your pilot searches for reference as you review and refine your strategy. ✅ Use the truncation symbol * to increase keyword capture (i.e., singular/plural/conjugated variants). ✅ Scan useful papers for additional keywords/terminology and references.

Start by clearly define your inclusion and exclusion criteria from the outset. Failing to do so can result to inconsistent decisions during the screening process. Excluding studies published in languages other than English or studies from non-traditional sources (gray literature) can lead to language and publication bias. The general tendency nowadays is to publish in English, so the bias shouldn’t be that high. Avoid the tendency to solely focus on positive results, as this can lead to publication bias. Be sure to include studies with negative or non-significant findings. Including studies with significant methodological limitations can affect the robustness of the meta-analysis results.

Clear documentation on the steps taken to organize data is essential for giving your analysis credibility and reproducibility for future studies. If your data management, sources, and analyses cannot be reproduced by other analysts, then it fails to contribute to furthering the knowledge of the data community.

One thing I found helpful is having a list of possible sources cited down on an article. One way to achieve this is by reviewing common articles across cited literature. I read through them and note the disparity or support statements or results.

Even if you’re doing a quantitative analysis, I’d consider borrowing methodologies from qualitative research- such as narrative theory construction and synthesis to provide a logical guide for scoping models in initial stages. The Centre for Reviews and Dissemination (2008) has a great pdf useful for tackling all sorts of systematic reviews and meta-analyses!

Meta-analysis must be more insightful if every part of the research is given more attention. Especially while making your directory for the articles, start from the very beginning or the very initial article. The initial article should be followed by the citation the the case starting with the latest should be observed and looked back with the references.

Don't underestimate statistics, it's trickier than you think go until you do not know what you are doing but then get help from a professional statistician.

A useful metric to include in meta-analyses is effect sizes of the studies included when appropriate. Effect sizes presented in a tabular form is a great tool for a quick assessment of the average effect size of the phenomenon being studied.