A
control is a reference against which the results of an experimental
manipulation can be compared. For example, if we wish to explore the effect of
smoking on human lung tissue, we need a control of tissue samples from
non-smokers to compare with our treatment group of tissue samples from
smokers.
Example:
Hypothesis: Giving a vitamin supplement to the feed of caged of rats leads to
increased longevity.
We
need a control group of rats that are identical in every way to our treatment
group, except that they do not experience the experimental manipulation itself.
This
is a negative control. A control group to which no
manipulation is applied.
Example:
Hypothesis: A particular novel treatment for common cold in humans produces
better results that the currently used method.
We
need a control group of rats that are identical in every way to our treatment
group, except that they are given the old treatment.
This
is a positive control.
Both
can be used.
If
we run our own control group, this is called a concurrent control. If
instead we use historical data as a reference, this is called a historic
control.
Example:
Common cold example given before
A
placebo or vehicle control is a treatment that is designed to appear exactly
like the real treatment except for the parameter under investigation.
1)
Completely Randomized Design
For a question such as “How does the regularity of
feeding with a liquid fertilizer affects the growth of tomato plants?”
20
plants should be fed at 5 different rates. The total 100 plants should be
assigned at random to the 5 feeding rates.
This
experiment is called Completely Randomized Design because individuals
are assigned to treatment groups completely at random.
A
balanced experimental design has equal number of experimental units in
each treatment group and an unbalanced design does not.
At
the end of the experiment we look for differences between the groups of
individuals.
The simplest design. We have
varied a single factor experimentally (feeding rate) and then looked for
differences between groups that experienced different levels of the factor.
Such
a design is referred to as a one-factor design or a one-way design.
(Balanced,
fully replicated, completely randomized one-factor design with five levels of
the factor)
Method
of analysis
If only two groups to analyze this can be done using
t-test. If more that two groups then by using Analysis of variance (ANOVA)