1. the dangers of correlation
and causation
a casual warning
ice cream can
turn you into a murderer?
2. Correlations show when there is a
pattern or link in way two variables
change
Different kinds of linear correlations
R2 measures the
strength of a
correlation
0 none
1 perfect
4. Designing Studies
The minimal conditions establishing cause and effect in medical diagnosis.
The Bradford Hill Criteria
Temporality
Sir Austin Bradford Hill
(1897 – 1991)
effect in that the effect occurs after the cause. Also, if it is to be expected that there is some delay between cause an
Strength and association
tion between these in repeated events or experiments. Full strength correlation has a coefficient of 1. A weaker assoc
Biological gradient (dose-response)
relationship between the dose given and the reaction of the patient. This may not be a simple linear relationship and
Consistency
ot prove a general cause and effect in wider contexts. To prove a treatment is useful, it must give consistent results in
Plausibility
ries and results. If a causal relationship appears to be outside of current science then significant additional hypothesi
Specificity
ere is no other plausible explanation. This is not always the case in medicine where any given symptoms may have a
Evidence
xperiments, where many significant variables are held stable to prevent them interfering with the results. Other eviden
5. Sir Austin Bradford Hill
Analogy
hen other factors similar or analogous to the supposed cause should also be considered and identified as a possible
Coherence
xperiments in which variables are controlled and external everyday evidence are in alignment, then it is said that there
6. Evaluate the following
…
-ice cream and sun-burn are correlated positively
-lead-poisoning and risk of being hit by a car are collelated positively
-There is a negative correlation between the use of sun-cream and the
risk of having an accident in a swimming pool
-ice cream eating is positively correlated with the incidence of murder