5. Replication
Applying each treatment to many experimental
units—25 males, 25 females—reduces the role
of chance variation and makes the experiment
more likely to detect differences in the effects of
the treatments.
6. BLOCKING?
• YES, because gender may have
an effect on the results due to
different hormones.
• Blocking by gender
7. Method of Randomization
• Al phabet i z e m es and f em es
al
al
s epar at el y by l ast nam
e
• Num
ber t he m es f r om 1- 50
al
• Us e t he RandI nt f unc t i on on
t he c al cul at or t o gener at e 25
r andom num
ber s
• I gnor e r epeat s
• Thes e 25 m es wi l l r ecei v e
al
t r eat m
ent A
• The ot her 25 m es wi l l
al
r ecei v e t r eat m
ent B
• Repeat abov e st eps f or
f em es
al
8. Randomization
The use of random assignment uses chance to
assign subjects to the treatments—chocolate or
placebo chocolate. It creates treatment groups
that are similar before the treatments are
applied. Thus, randomization helps prevent
bias.
9. Treatments
Treatment A
• Subjects will eat a
Hershey’s milk
chocolate bar for 6
months, once a day
Treatment B (control group)
• Subjects will eat a FAKE
Hershey’s bar (no
chocolate) for 6
months, once a day
10. Treatment B - Control
This treatment provides a form of comparison
between chocolate and no chocolate. This helps
avoid confounding of the effect of a treatment
with other influences, such as lurking variables.
11. Explanatory Variable
• Categorical
– Type of chocolate
Treatment Factor Levels
• Two Levels
• Treatment A (chocolate)
• Treatment B (placebo chocolate)
Response Variable
• Quantitative
– Number of pimples
12. Placebo Effect
• Treatment B=dummy treatment
• If subjects know which type of chocolate they are
receiving, it could cause bias in the results.
• The effect of simply thinking that they are eating
chocolate could cause pimples.
13. Double-blind
• Subjects and evaluators are unaware of which
treatment the subjects received.
• This prevents any bias.
– Knowing that the subjects are getting “just a placebo”
might weaken the placebo effect and bias the
experiment in favor of the other treatment. If the
evaluators know that a subject is getting “just a
placebo,” they may expect less pimples than if they
know the subject is receiving the other treatment.
14. Results
• Compare number of pimples on face
– Take an initial headshot before the experiment
– Take a headshot at the end of each month
– Each picture must be taken in the same room at
the same time
15. Concerns
Stress could be a confounding variable—it could
increase the number of pimples.
The pictures must be regulated—things like
lighting must be the same every time.
16. Conclusions
If the data from the two treatments (chocolate vs. placebo
chocolate) show a difference in number of pimples
(chocolate=more pimples), then we can reasonably conclude
that chocolate does indeed cause zits.
If we observe statistically significant differences among the
groups in our experiment, we have good evidence that chocolate
does cause pimples because our experiment was randomized and
used a control group to compare results.