The search process

Besides doing a technically well-structured search, you should conduct it in a certain order. If you do not run preliminary and test searches, you may risk missing out on keywords and aspects which are central to the search result. You may also lose the indication if your search query is too broad or too narrow. At a preliminary stage, you should also make sure that there are no previous or ongoing projects studying the same aspects or having registered that they are working on a similar study.

 

1. Preparatory searches

Check out what has already been done. Are there any recent reviews within the subject area? Prepare the search by searching broadly in interdisciplinary databases such as Scopus and Web of Science and use suitable key words and filters for the type of study you are planning. Databases may offer filters that help you identify reviews, often in the left-hand margin of the database interface. 

Search Filters by Study Design is a website that presents filters for different kinds of studies. Do not forget to start documenting your searches from the very beginning.

Sometimes, there are published search strategies and search blocks to be used directly or modify for your own search. These are available online but remember to provide a correct reference when using these search blocks.

https://blocks.bmi-online.nl/

https://sites.google.com/view/searchresourceslib/collections

https://searchrxiv.org/

2. Test searches

Continue with test searches in the most frequently used databases in your discipline. Proceed from your research question and find out which search terms are useful and how they can best be combined. Identify an area connected to your research inquiry. Be open-minded and creative at this stage of the search process. 

When the most important terms/keywords have been identified, you may add synonyms, different spellings, and suffixes to the terms. Group the terms into separate search blocks and decide how to combine them. Begin with the most specific block which yielded the fewest search results. 

Run a test search and evaluate the results: 

  • Did you find what you were looking for?
  • Are the search results relevant to your research question?
  • To what extent is there previous research published in the field? How many documents may be managed within the parameters for your planned work?
  • Does your research question need adjustment?
  • Look at the titles, abstracts, and the indexing of document results. Are there any other free text terms and keywords that are relevant for your research?
  • Are key articles missing from the search results? Why? Figure out how the search may be further modified to retrieve key articles and similar works. Remember that key articles may sometimes be published in journals that are not covered by the database you have searched. 
  • Look at reference lists. Are there relevant articles which have not turned up in your search results? Modify your search again. 
  • Do you get too many or too few document results? Should the search be broadened or narrowed?
  • How many blocks of search terms should be used? Four, three, two or one? Fewer blocks indicate a broad (sensitive) search which will yield more results but perhaps many irrelevant results too.

Systematic searching is an iterative process, continuously adjusted until the search can no longer be improved. New search terms, spelling variations and combinations are successively added. This way results get more and more accurate.

Remember to document all the variations you work with. This will make it easier to reuse wordings from earlier searches. A previous search can be used together with new terms that turn up later during the process.

3. Main search

Start from the result of the test searches when you decide how to build your main search. Document the search, including search terms, combinations used, limitations and filters and the date of search. Note the number of results, the database and database provider. Save all results in a reference management program.

A three block search could look something like this in PubMed using MESH-terms and free text words (here some search terms are left out due to space limitations):

 

((bicycling[MeSH Terms]) OR (bicycling[Title/Abstract] OR bicyclist[Title/Abstract] OR bicyclists[Title/Abstract] OR bicycle[Title/Abstract] OR bicycles[Title/Abstract] OR biking[Title/Abstract]…))

AND

((facial injuries[MeSH Terms] OR (face[MeSH Terms] AND injuries[MeSH Subheading])) OR (facial injury[Title/Abstract] OR facial injuries[Title/Abstract] OR eye injury[Title/Abstract] OR eye injuries[Title/Abstract] OR facial fracture[Title/Abstract] OR facial fractures[Title/Abstract] OR jaw fracture[Title/Abstract] OR jaw fractures[Title/Abstract]…))

AND

((head protective devices[MeSH Terms]) OR (head protective device[Title/Abstract] OR head protective devices[Title/Abstract] OR helmet[Title/Abstract] OR helmets[Title/Abstract]…))

 

4. Further searches in other databases

Repeat the search in other relevant databases. Adjust the search to different database interfaces, filters, commandos, abbreviations, and terminology. Depending on subject coverage, different or no controlled vocabulary may be used. Again, document and save the results of the searches.

A two block search in Web of Science could look like this, again with some search terms left out. (TOPIC here means free text search, corresponding to searching the fields of Title-Abstract-Keywords in the database).

TOPIC: (bicycl* OR biking…)

AND

TOPIC: ("head protective device*" OR helmet*…)

 

5. Complementary searches

Writing a review takes time and new articles are published continuously. Before drawing the line for not including more data, it is essential to update the main search to retrieve new publications but also to discover if articles have been changed, corrected, commented upon, or withdrawn. New key words and terms may have appeared and should be used to complement the original search.

A complementary search should be carried out after 6 to 12 months and as close to the publication date as possible.

Citation search

Often, further relevant studies are found by scanning reference lists of articles. The reason these have not shown up earlier in the search can be that they are published in journals that are not available through the databases used or alterations in indexing. Citation analysis is done in the same way as scanning reference lists but in reverse. With citation analysis, the goal is to find out how articles have been cited after publication. Some databases that offer this possibility are Web of Science, Scopus, and Google Scholar.

 

Searching for grey literature

The document type in question may also be a report, thesis, or conference proceedings which are not included in article databases. These are examples of grey literature which are not published or disseminated in the usual commercial channels, and they may sometimes be more difficult to locate. Grey literature may also be government reports, guidelines, technical reports, patents, standards, as well as unpublished, cancelled, or ongoing clinical studies.

To find grey literature, you may try the following:

  • publication databases like DiVA: Digitala Vetenskapliga Arkivet and SwePub. In these databases, you can search for literature by researchers at Swedish universities, especially within the last 10-15 years.
  • Subject databases may index different types of grey literature. Some of them are found in the library’s list of databases.
  • Libris and other library catalogues may be used to identify grey literature.
  • Google Scholar may be used for complementary searches for grey material. It is not recommended as a primary database for systematic reviews because of the lack of reproducibility of searches, insufficient precision, and unclear subject coverage.
  • The websites of authorities, research departments and other organizations and their archives of electronic publications may be searched for grey literature.
  • You may search archives for clinical trials to find ongoing research.
  • Search for ongoing projects at the websites of Swedish Agency for health technology assessment and assessment of social services (SBU) (within the health sector and social services), Cochrane Library (within the health and medical sciences) och Campbell Collaboration (within the social and educational sciences).
  • Preprints may be identified in special databases, such as MedRXivBioRXivArXivSSRN et cetera. Some recently finished studies are published as preprints before peer-review and actual publication in a journal. Preprints may be included in systematic reviews if the subject for the review is topical for the time being.
  • Get in touch with researchers to get information about ongoing or unpublished studies.

 

Search support for researchers

LiU researchers and PhD students can consult a librarian, for example when doing a literature review for a dissertation or a systematic review. Together we will look at your specific research questions and discuss topics such as:

  • selecting databases and search tools
  • selecting keywords
  • constructing search strings
  • documenting searches
  • managing search results

Make an appointment by completing this form:

Search support for researchers