Using the keywords and synonyms from step 2, you create search strings (what you type into a search box) and use them to search in the databases you identified in step 3.
Based on the results, you will then modify your search terms, removing or adding terms, and use these revised search strings in the same databases, or different databases. You may need to repeat this process multiple times before you find relevant articles.
The keywords and synonyms you created in step 2 can be combined in multiple ways to create search strings, using boolean operators (see box below). There is no one right way to create a search string; it's largely a matter of trial and error. If you use a search in a database and get no results, you probably want to remove terms from your string to make it more general. If you get too many results, you may want to add terms to make it more restrictive. And if you are getting results but most of them aren't relevant, you may need to use different terms.
If your topic has three keywords (keyword1, keyword2, and keyword3), and each keyword has two synonyms (synonym1a and synonym2b, etc.), some possible search strings to start with are:
After you use a search string in a database, look over the results. Were there no results, only a few, or and overwhelming number? Scan the first page of results, looking at the titles and abstracts. Do they seem relevant to your topic?
Based on what you see, you will usually need to modify your search. Here are some tips on how to do so:
As you conduct your iterative searches, whenever you find an article that looks like it might be useful, take a note of it (ideally, capture it with a bibliographic manager (also known as a citation manager) so you can read it and use it later.
Sometimes a search produces no usable results, even after revising the search and using synonyms. This problem is especially common when searching on a particularly narrow and specific topic. Sometimes it means that you need to keep trying different search terms, or search in a different database. But it may mean that there hasn't been any significant research published on your exact topic.
In this situation, it is sometimes easier to modify your topic to one where more research has been done. But if that isn't possible, or you are particularly invested in the topic, it is still usually possible to find relevant research papers. A topic is usually at the intersection of several subjects, and even if you can't find research on that exact intersection, you can usually find useful research on the component parts.
In the example of timing of rewards for long-term engagement in video games, assuming there are no relevant papers on that exact topic, you could search for articles on engagement in video games (leaving out the question of rewards and timing), or on rewards in video games (leaving out the question of engagement), or just on video games in general. Articles on these related topics are likely to have a bearing on the more specific topic.
This expanded search strategy is useful even when you can find research on your exact topic. If you look at the references in a paper, you will find that while they may cite some articles on more or less the same topic as the paper, they also usually cite many articles on related, relevant research that is not an exact match.
While every database has its own rules for how it interprets search strings, these tricks work in most of them. If you find yourself using a particular database a lot, it is worth looking at its Help page to find what tricks work for that database.
The student tried multiple search strings initially, including a search that just used all their keywords; another, more complex search that included most keywords and synonyms, and another one that was somewhere in between.
Using these searches in the ACM Digital Library, the student found the first search returned over 600,000 results, while the second search found over 70,000, and the third over 10,000 (in ACM, when you search for a string of words with no connectors, it treats it as if there was an "OR" between each word, and returns everything that has any of the words).
Scanning the results, the student found that some of the articles in each of the searches seemed relevant to their interests, with the third search giving the best results.