Each of these three databases is very popular at MSU, but each database is very different. Many people choose to use only one of them without really understanding the differences on the back end that will affect their search results. The following major differences explained may help you make an informed decision about which database to use when. (Also see related Guide: How do I know which database to search for biological literature? )
Pubmed and Web of Science are human-curated databases. This means:
Journals are the focus of Web of Science and PubMed, and they are selected for inclusion by humans based on scholarly and quality criteria by literature review committees. See: PubMed journal selection or Web of Science journal selection
Data about each article is entered into the database in a uniform structured way: author, title, date, journal name. This means you get accurate retrieval when searching for those things. Results can be sorted reliably by latest date.
Articles in PubMed are tagged with important information about their structure , such as "review article" or "clinical trial". They are tagged with structured words about their content and major topics to help you get better search results. Example: microRNA [Majr] as a search means you can search for articles where microRNAs are the major topic of the article. A search for "mouse" automatically includes the word "mice". Of the three, PubMed has the most "bells and whistles" to help you get good search results.
Articles in Web of Science are also tagged with important information about their structure, such as "review article". They are not tagged for content, so you must include all possible variations of the topic you are searching.
These databases have subject specificity in that journals are chosen for inclusion in PubMed or Web of Science based on subject matter. PubMed focuses on clinical and biomedical literature. Web of Science is interdisciplinary and includes the "best" journals of each subject area.
Accurate retrieval means that search results are reproducible and reportable (especially important for systematic reviews)
In contrast, Google Scholar is not a human-curated database but a search engine of the whole internet which narrows the results to "scholarly " ones based on machine automated criteria.
Criteria for inclusion as "scholarly" in Google Scholar results is based on publishers submitting information to Google Scholar about their web sites, and is not necessarily based on the attributes of the sources themselves. Not all publishers may have worked equally with Google. Some included journals may be poor quality or even predatory journals.
All subject areas are covered, and Google Scholar includes more different types of sources: conference proceedings, books, and reports, that are not included in Web of Science or PubMed.
Information about author, date, journal name, and other pieces of the citation is not standardized so may not be machine-readable by Google. Google Scholar offers an author, date or journal name search, but its accuracy varies greatly. Results can be sorted by date, but not necessarily reliably.
No human-curation means no tagging of articles for structure or content. If you want review articles, you can add the word "review" to your Google Scholar search, but as a keyword. This means you may or may not get real review articles. Content words are no tagged, so you must search for all possible variations of the words from your topic.
Inaccurate retrieval and variable content means that search results are not necessarily reproducible and therefore not reportable. Google Scholar would not be appropriate to use for systematic reviews.
Both Google Scholar and Web of Science track citations--how many times that an article has been cited by other articles, books, or sources. Both Google Scholar and Web of Science allow you to sort your results by times cited (Google Scholar includes this in its "relevancy ranking").
Be aware that times cited will differ greatly between Web of Science and Google Scholar. Neither one is complete, although Web of Science citation data is considered more accurate and reproducible and is used by official organizations as the standard.
PubMed has started tracking some citations, but it is only able to track citations by articles in PubMed Central, which is an open access repository of articles. This makes the citation tracking more limited. It is useful for discovery of articles but not recommended for any kind of counting of citations.
Because Google Scholar searches the full text of articles, you can find information that is not necessarily in the citation or abstract of an article, for instance, a detail buried in the Methods section of a journal article. If you're not having luck finding something extremely specific with your PubMed of Web of Science search, try Google Scholar and you may find it.
PubMed and Web of Science are limited to abstract searching, but that is not necessarily bad for most searches. If you want an article that is primarily about a specific topic, certainly the information will be in the abstract. However, full text searching with Google Scholar will yield different results and may bring up some articles you did not find with the other two.