META-ANALYSIS: PROBLEM-FINDING AND SEARCH STRATEGIES
Problem-Finding
1. General area of interest: search for reviews of literature
a. How old are reviews? More than 10 years may indicate need for new review
b. What are the topics covered? Do they encompass current issues?
c. Was the review a meta-analysis? If not, would a meta-analysis add to the
literature?
2. Importance- is the topic important to current theory and practice?
a. Would a comprehensive meta-analysis replace a body of literature currently
referenced?
3. Accessibility of literature- is it in journals, ERIC documents, dissertations?
a. What is the proportion?
b. What is the expected cost to access the literature?
c. What is the time-frame to access items? Does it fit into your time limitations?
Literature Searches
1. Electronic searches
a. Library-based resources: e-resources (eg. TAMU library
b. Internet searches: Google Scholar
c. Other search engines: problem of sorting out irrelevant web pages
2. Journal searches
a. Backward searches of review journals (current issue backwards)
b. Forward searches of journals that published major earlier reviews
3. Author Searches
a. Social Science Citation Index author citation search
b. online author searches: library searches, etc.
Search Strategies
1. Electronic searches
a. Beginning with simple searches:
i. Is the number found more than a few hundred? If so, add another
descriptor to reduce it to no more than 200-300
ii. Review titles and abstracts; if a significant proportion of the titles do
not fit the area (eg. in business or microbiology rather than education or
psychology), revise the search by adding terms that will eliminate the
irrelevant titles, compare the searches to see if you have eliminated un-
needed titles but have not also eliminated potentially useful ones
iii. Begin to survey the titles AND abstracts to determine if they are likely
to fit your topic OR
iv. Revise your topic to reflect what you are finding in the literature
b. Build a search tree that lists the number of titles for each descriptor or term
used; assume that you will find 2-3 times as many titles as your
electronic search has uncovered
c. Cross-reference different descriptors and their joint selections; review abstracts
d. Print or put into excel file title, abstract, add an id, major descriptor(s)- 2 or 3 at
most
2. Branching library searches based on samples of relevant articles
a. Start with MOST RECENT article you can find; check references of that article
b. Review titles and abstracts of the references for appropriateness, cross-
reference with existing electronic article database
c. Review references of existing data base; continue until no new sources are
found with this method
3. Author contact
a. After coherence of the data base of sources, contact major authors to determine
if new articles they have recently published may not yet be listed in electronic
resources or in journals not in the various databases; indicate to an author that
you are interested in works not in the databases and journals you have already
examined so they understand you are not asking them to do your work; this is
not likely to add much.
Other issues:
1. Foreign languages: your fluency or access to translators must determine if you will
include foreign language articles
2. Foreign journals in English: availability through interlibrary loan on a timely basis will
determine access
3. Dissertations: often cost will determine if the hard copy should be obtained ($40?);
electronic versions may be cheaper but still a significant investment; abstracts will
need to be the basis for selection, but make sure a published version is not
available, which usually will be 1-3 years later than the dissertation
4. Data availability for meta-analysis
a. Will this be quantitative only, quant-qual mixed method?
b. Single case studies: is there control data available (longitudinal same-self,
normative with national or regional norms, comparison group summary data)?
Will you include these studies?
c. Is the study actually a position piece or editorial? Reject these unless the
content is the focus of the meta-analysis
d. Does the study reproduce the same data as another article? Which one better
provides data, or should the two (or more) be combined into a single
“study”?