Page 200 - Encyclopedia of Nursing Research
P. 200
EXPERiMENTAL RESEARCH n 167
incorporated into technology (e.g., electronic treatment at a particular time” (Cook &
health records) to facilitate best practice by Campbell, 1979, p. 8). This refers to control
clinicians at the point of care (Melnyk & over two processes that determine who gets E
Williamson, 2010). what at what time. The first process is the
researcher’s use of randomization methods
Bernadette Mazurek Melnyk to assign subjects to treatments. This is the
Ellen Fineout-Overholt preferred method of exerting control over
subjects and their treatment as, theoretically,
it ensures that known and unknown extrane-
ous forces inherent to subjects are dispersed
ExpEriMEntal rEsEarch equally across the different treatment arms.
This may not always be possible, in which
case the second process comes into play—
True experiments have the potential to pro- that of structuring the assignment process
vide strong evidence about the hypothesized in such a way that major, known extraneous
causal relationship between independent forces are controlled.
and dependent variables. Experiments are Commonly used design strategies
characterized by manipulation, control, and include blocking, fixed and propensity
randomization. The quality of experiments matching, and counterbalancing. in block-
depends on the validity of their design. ing, the potentially confounding variable
Manipulation means the researcher is incorporated into the study design as an
actively initiates, implements, and terminates independent variable. The levels of this var-
procedures. in most instances, manipulation iable are considered blocks, and subjects are
is linked to the independent variable(s) under assigned to blocks on the basis of their value
consideration. Essential to manipulation is on the blocking variable. Next, in each block,
that the researcher has complete control over subjects are randomly assigned to the study
the process. The researcher decides what is to arms. in fixed matching, a weaker but com-
be manipulated (e.g., selected nursing inter- mon method of control, the researcher iden-
vention protocols), to whom the manipula- tifies one or more extraneous (usually up to
tion applies (e.g., samples and subsamples of three) variables to be controlled. As soon as
subjects), when the manipulation is to occur a subject is recruited for one of the treatment
according to the specification of the research groups, the researcher then tries to find sub-
design, and how the manipulation is to be jects for the other group(s) identical to the
implemented. first subject on the specified matching vari-
Manipulation implies and is impossible ables. in propensity matching, all known or
without researcher control over extraneous presumed confounding variables are used to
sources that might affect and lead to incorrect calculate a propensity score for each subject.
scientific conclusions. Control aims “to rule Subjects are then matched on this propen-
out threats to valid inference.” it also adds sity score. Counterbalancing occurs when
precision, the “ability to detect true effects of the researcher is concerned that the order
smaller magnitude” (Cook & Campbell, 1979, in which treatments are administered influ-
p. 8). Unlike laboratory studies where total ences the results. When counterbalancing
control is often possible, in clinical research is used, all subjects receive all treatments;
control is a relative matter. The researcher however, the order of administration of treat-
has the responsibility for ensuring as much ments is varied.
control over extraneous forces as possible. Randomization entails two separate pro-
Control also includes “the ability to cesses: (a) random selection of subjects from
determine which units receive a particular the population and (b) random assignment of

