The more powerful the eff ec t of a varia bl e on the

The more powerful the eff ec t of a varia bl e on the dependent variable. the more important it
is to match participants on this variable to assure comparable gro ups. To use n matched-subjects
design effectively, you must identify the relevant variables and match t he groups participant by
participant on these variables.
Thi s principle for identifying relevant matching variables is similar to an issue we di scussed
in Chapter ?- confounding in diff erential research. In that case. co nfounding could occ ur o nl y if
(1)
U1 e potential co nfounding variable correlates with th e depe nd ent m easu1·e, and (2) there is a dif-
fe rence between d1e groups on the potential confounding variable. Variables that co rrelate with the
dependent measure are often pr oblematic in research and therefore require some type of co ntrol.
Matched-subjects designs ensure that the groups do not differ on potential confounding variables
by matching on those variables.
The Pr ocess of Matc hin g. The procedure for matching participants and assigning them to groups
was described in Chapter 9 i.n the discussion of matched random assignment. In that example, par-
ticipants we re matched on years of work experience. We then divided them into pairs by se lecting
the two with the longest work experience, then the two with the next longest work experience, and

8/26/2014 CourseSmart - Instructors - Print
7/10
U ser name: Joyce Gresham B ook : Research Methods: A Process of Inquiry, Eighth Edition Page: 257. No part of any book may be reproduced or transmitted by any means
without the publisher's prior permission. Use (other than qualified fair use) in violation of the law or Terms of Service is prohibited. Violators will be prosecuted to the full extent of
the law.
MATCHE D- SUB J ECTS DES I GNS
257
so o n. Finally, we randomly assigned one member of each pair to one of the two groups and auto-
matically a ss igned his or her partner to the oth er group. The r es ult was two gro up s of participants
matched on d1e variable of length of work experience.
Il
~ 'ri ~i~bS a'j AJ.Y[.i ~~ MSH if ~ g
process to exclude participants f or whom a close march is
not a vai ra'bre. Iior exru~PJe. one employ ee may have worked at a job much longer than anyone else.
This employee would be so different fr om the others that there would be no suitable matc h. and so
this employee would be excluded from dle stud y.
It
is po ssible to exte nd matching to three or more conditions by matching in sets of three or
more participants. T he members of the set are then randomly assigned to conditions. one per con-
dition. Of co ur se . in cr eas ing the number of experimental conditions to which we want to ass ign
matched participa nt s will also increa se the likelihood that potential participants will have to
be
excluded because enough close matches ca nnot
be
found for them.