Searching for prior research starts before you type anything into a search box. Skip that step and the results are either far too many or none at all. This article walks through the whole sequence, from settling on the words that describe your topic to narrowing the field to the 30 papers you will read.
Three things to settle first
Write these three down before searching.
- The subject. What are you studying? “Students”, “university students in Japan”, and “part-time work among university students in Japan” differ by two orders of magnitude in the number of papers they return.
- The angle. What about the subject? Whether you want to describe a situation, explain a cause, or measure an effect changes which kind of study you need.
- The period. How far back do you look? In some fields a conclusion from ten years ago has since been replaced, so starting with the last five years gives you the overall shape faster.
Until these three are settled, no amount of clever phrasing will narrow your results. They are also what makes a conversation with your supervisor productive.
Building search terms
The words you settled on will not match the words used inside papers. You need to translate them.
There are three reliable sources for that translation.
- The index of a textbook. The back-of-book index lists the terms the field actually uses.
- The keyword field of one paper you already have. Authors list keywords, and those same words appear in related papers.
- A controlled vocabulary. Some fields maintain a standardised term list, such as MeSH in medicine.
For “the effect of part-time work on university students’ grades”, the papers themselves use terms such as “work hours”, “academic performance”, and “student employment”. Everyday phrasing will not reach them.
Where to search
Four places cover most theses without much overlap.
| Where | Coverage | Best for |
|---|---|---|
| Google Scholar | Broad, across fields and languages | Finding the first paper |
| Web of Science or Scopus | Indexed journals, with citation counts | Following citations forward |
| PubMed | Medicine and life sciences | Health-related topics |
| Your library’s discovery service | What your institution can actually give you | Getting the full text |
Start broad in Google Scholar, then use a citation index to follow the trail. Check your library’s service last, when you know which papers you need.
Once you have one paper, expand from it
The most efficient way to find prior research is not better search terms. It is following citations from the one paper you already have. There are two directions.
- Backward. Read the reference list. These are the earlier studies the authors relied on.
- Forward. Find the papers that cite this one. These came later, and they tell you what happened to the idea.
Two rounds in both directions will surface most of the important work in a field. When the same papers keep reappearing, you have reached the centre.
In Litlas, opening a paper shows both the works it cites and the works citing it as a graph, so you can move in either direction without leaving the page. The free plan needs no card and works right away.
Narrowing to 30
Following citations gives you well over 100 candidates. Cut them in this order.
- Read only titles and abstracts. Drop anything off-topic. This usually removes more than half.
- Look at citation counts, comparing like with like. A highly cited paper sits near the centre of its field. Recent papers have had less time to accumulate citations, so compare papers published in the same period.
- Check that you can actually get the text. Set aside anything you cannot read now. Your library may be able to obtain it if you truly need it.
Split the remaining 30 into a layer you read in abstract only and a layer you read in full. Ten papers in full is usually enough. The reading method is in reading a paper in three passes.
When nothing comes back
If a search returns zero results, work through these in order.
- The query is too specific. Drop from three terms to two.
- The phrase is not a technical term. Look for the field’s wording in a textbook index or in the keyword field of a paper you already have.
- The work exists in another language. A topic with nothing in one language can have hundreds of papers in another.
If it still returns nothing, your topic is either very new or defined too narrowly. Widen the subject by one step, then position your own work against what you find. That is easier to write up than an empty search.
When too much comes back
Above about a thousand results, narrow in this order.
- Restrict by year. The last five years is usually the right size for an overview.
- Restrict by article type. Limiting to review articles surfaces papers that already summarise the field. These make good first reads.
- Add one term. Add either the subject or the angle. Adding both at once can narrow the results too far.
Keeping a record
If you record how you searched, the “prior research” section of your thesis almost writes itself. Keep at least three things.
- Which database, which terms, and on what date
- How many results, and how many you kept
- For the papers you dropped, why you dropped them
This record is also what you show your supervisor. If keeping it by hand is tedious, save the papers with notes attached in one place so you can sort through them later.