| |
Advantages of
Randomization
Randomization of subjects
has several unique advantages over all other current methods for
assigning patients as subjects to experimental and control groups.
- First, randomization,
if properly done, removes any chance of predicting what treatment
a person will receive.
- Second, this method
creates, on average, two or more groups that are equal on all
known AND unknown factors. The larger the size of the randomly
constituted groups, the more likely the groups will be identical
on all known risk factors.
In addition, because
there is a control group for comparison, factors that might contribute
to pain reduction for reasons not related to the experimental
treatment will be randomly distributed in both the experimental
and control groups, in effect washing out the impact of unintended
or unknown influences, hence controlling for "regression
to the mean" phenomena that commonly occur in pain conditions.
| Definition
of "Regression to the Mean"
|
|---|
|
Regression
to the mean is defined statistically as the tendency of
extreme values to cycle around some common average value
(Whitney
and Von Korff, 1992).
|
In practical terms,
when measurements include extreme values, statistical principles
dictate that repeating the measurements will usually result in
those extreme values coming closer (i.e., "regressing")
to the overall mean value.
As a general rule,
patients seek treatment at the height of their discomfort; TMD
patients tend to seek treatment at the height of their TMD pain.
Often the passage of time, alone, yields lower ratings of pain
and pain-related sequelae, such as depression and use of medications.
Patients selected from TMD clinic cases for an RCT will inevitably
include those whose pain at the time of selection is atypically
high. Subsequent re-measurement after treatment may reveal lower
TMD pain levels, which might reflect the effect of treatment or
might simply represent a regression to the mean, which could result
in lowering of heightened pain levels even in the absence of treatment.
Whitney and Von Korff
(1992) present an excellent discussion of this important topic
with actual TMD pain data that illustrates the point very clearly.
|