Do not forget NOT in your searches

I've copied an interesting listserv post by PsycALERTS about the Boolean search operator NOT below.

The post champions the often neglected NOT search operator. In short, this operator is useful when certain data fields have multiple population or document types assigned to them.

"The Power of No

All things are defined in part by what they are not. What is darkness
unless juxtaposed to light, cold to warmth, silence to sound? For most,
there’s no hard clear line where one thing stops and another begins. Ideas
and language can be particularly challenging, with meanings slippery and
sometimes almost impossible to isolate or pin down. However, let us leave
the meaning of being and nothingness to the deconstructionists and focus
on the very pragmatic aspect of how to disentangle these comingled
results from an indexed database. It can be difficult to tease search criteria
from among the various research contexts in which they are indexed. How
can we as researchers craft our searches to return the results we do want
and carve out the “noise” that we don’t?

For APA PsycNET users, we can do that by becoming comfortable with the
very powerful Not Boolean, which excludes records from retrieval. Let me
share a fairly simple example that came my way not long ago from a
researcher who asked for some help crafting a search. What he wanted to
find was research in PsycINFO that was specifically about adolescent girls
and math.

Because he knew a thing or two about databases, he had very sensibly
crafted a search in which he searched for math* (with an asterisk to
capture all variations of the word) as a keyword and used the age limiter to
specify that the results would need to include participants between the
ages of 13 and 17 and the population limited to female participants.  But
the results he got were not what he expected. To begin with, he got about
2,000 results, more than he had expected or could easily work with. And
when he looked more closely at the records returned, he found that a lot
of them were not what he was looking for at all. True, all his results did
contain the limits he’d specified, and that is certainly important. But
looking within the records, he found research that had included preschool
children and college age kids, even research on the very old struggling with
dementia. And while the results had all included female populations, many
of them included male populations as well and did not make distinctions
between the genders. Thus, although the results included his desired
group, they weren’t limited by it.

Clearly, a step is missing in this search. To correctly focus the results, it’s
not enough to stipulate what you do want. You may want to take the
additional step of excluding what you don’t want. So let’s take our original
search and add two additional criteria: the age will be only 13 to 17, and
we’ll use the Boolean Not to eliminate all of the other age groups. Likewise,
we will exclude males from our result set.

Our new result set includes 91 records, and a quick look at our result set
shows that it looks much more promising. Now we can focus on other
attributes of the set as a whole that might not fit into our search. For
example, in this fresh set of results, we find quite a few dissertations and
conference papers on our topic. Useful as those sources can be, we don’t
really want them at this time. One way to remove them would be to select
the “Peer reviewed journals only” checkbox. However, if we do that, we
also lose some other content we might want, such as book chapters. Here
again the Not Boolean allows us to cut with a scalpel instead of an ax.
Instead of restricting our results to peer-reviewed journals, let’s just
remove the results we don’t want. In the Document type category, we can
remove conference abstracts and dissertations from our findings.

Now if we look at our findings, we have a set of 62 results, all of which are
focused on adolescent girls and their relationship to mathematics.

So remember, like the old Bing Crosby song used to say: you’ve got to
latch on to the affirmative and eliminate the negative. That will give you the
specific results you’re looking for.

Here’s our final search:
Search For:
Keywords: math* AND Age Group: Adolescence (13 to 17 yrs) AND
Population Group: Female AND NOT Age Group: Childhood (birth to 12
yrs) OR Neonatal (birth to 1 mo) OR Infancy (2 to 23 mo) OR Preschool
Age (2 to 5 yrs) OR School Age (6 to 12 yrs) OR Adulthood (18 yrs &
older) OR Young Adulthood (18 to 29 yrs) OR Thirties (30 to 39 yrs) OR
Middle Age (40 to 64 yrs) OR Aged (65 yrs & older) OR Very Old (85 yrs &
older) AND NOT Population Group: Male AND NOT Document Type:
Abstract Collection OR Dissertation"

Source: PsycALERT mail list message, May 29, 2014