6 INTRODUCTION TO BASIC CONCEPTS IN EXPERIMENTAL DESIGN

a nj ix ——<

may minimize the amount of research efort reguired to investigate a
research hypothesis.

Another important consideration in selecting a dependent variable
is whether the observations within each treatment level (or combination of
treatment levels in the case of multitreatment experiments) would be
normally distributed. The assumption of normality, discussed in Chapter 2,
is reguired for the experimental designs described in Chapters 4 through 12.
In some cases it may be possible to transform nonnormally distributed
observations so that the resultant distributions are normal. This procedure
is described in Chapter 2. If theoretical considerations do not dictate the
selection of a dependent variable and if several alternative variables are
egually sensitive and reliable, in addition to being normally distributed, an
experimenter should select the variable that is most easily measured.

SELECTION OF INDEPENDENT VARIABLE

m The independent variable was defined earlier as the presence or
: absence of radiation. Such a treatment is described as having two treatment
levels. If the experimenter is interested in the effects of different radiation
dosages, he can employ three or more levels of radiation. The levels could
consist of 0 microwatts, 20,000 microwatts, 40,000 microwatts, and 60,000
microwatts of radiation. This particular treatment is an example of a
guantitative independent variable in which different treatment levels
constitute different amounts of the independent variable.

In general, when the independent variable is guantitative in character
there is little interest in the exact values of the treatment levels used in the
experiment. In the radiation example, the research hypothesis could also
be investigated, using three other levels of radiation, say, 25,000, 50,000.
and 75,000 microwatts in addition to the zero-microwatt control level. The
treatment levels should be chosen so as to cover a sufficiently wide range to
detect effects of the independent variable if real effects exist. In addition,
the number and spacing of the levels should be sufficient to define the shape
of the function relating. the independent and dependent variables. This is
necessary if an experimenter is interested in performing a trend analysis as
described in Chapter 4.

Selection of appropriate levels of the independent variable may be
based on results of previous experiments or on theoretical considerations.
In some research areas, it may be helpful to carry out a small pilot experi-
ment to select treatment levels prior to the main experiment.

Under the conditions described in Chapters 2 and 4, the levels of
a guantitative independent variable may be selected randomly from a
population of treatment levels. If this procedure is followed, an experi-
menter can extrapolate from the results of his experiment to treatment
levels that are not included in the experiment. If the treatment levels are
not randomly sampled, the results of an experiment are applicable only
to the specific levels included in the experiment.

GNEIKA moi v naenčpnabna

iii

sič
k