Professional Documents
Culture Documents
•X Y
[X is causing change in Y]
Causality
• Conditions:
• Association (X and Y related)
• Timer order (X precedes Y)
• Logical connection (plausible theory)
• Non-spurious relation (no competing or plausible
explanations)
Causality
• Correlation does not imply causation!!!
Ice cream example
Association
• X and Y should be somehow related!
Time order
• Cause is preceding the effect
Logical connection
• You have some plausible theory for the causal
connection.
Non-spurious relationship
• You have to rule out all competing theories that
could explaining the causal relationship. That is
the tough one.
Examples
• RQ: What is the effect of soil type on abundance
of sexton beetles?
• Independent variable?
• Dependent variable?
• Soil type Sexton beetles
Examples
• RQ: What is the effect of geomagnetic field on
behavior of dogs?
• Independent variable?
• Dependent variable?
• Geomagnetic field Dogs
Problems
• Intervening variables
• soil beetles
• Income longevity
soil beetles
temperature (confounding
factor)
Reverse causation
•X Y
•X Y ?
• Not only the (in)famous
chicken and egg question!
Reverse causation
• Breastfeeding decrease the
growth rate of babies
• Supported by several
observational studies.
• Corrected by Marquis et al.
1997?
• Smaller babies were much
more vigorously fed despite
their genetic or other
predispositions to be small.
Example
• Q: Is a body alignment of
a dog, while he is
defecating, caused by
electromagnetic field of
Earth?
• M: Behavior was
observed and measured
in the field conditions.
• Can you prove that elm. Hart et al. Frontiers in Zoology 2013, 10:80
field is the cause?
Example
• Conditions:
• Association (X and Y related)
• Timer order (X precedes Y)
• Logical connection (plausible theory)
• Non-spurious relation (no competing or
plausible explanations)
Dogs
• Too many environmental variables are in play and there is always
something you did not measured or even think of measuring.
• Possible explanations:
• Wind direction
• Sun
.
.
.
Time t1 t2 t – time
O – observation
Group 1 R O1 X O2 X – manipulation of
Group 2 R O3 O4 independent variable
R – randomization
Group 3 R X O5
Group 4 R O6
• We are able to identify or estimate most threats
to internal validity of the experiment.
Experimental artifacts
• Experimenter effect (wishful
thinking) – Different
interaction with experimental
and control group (one gets
extra attention)
• Placebo effect – Effect of
belief that participant is treated
• Contagion – Effect of
experimental and control group
communication
Time t1 t2 .... tn
Group 1 R O1 X O2 Oi
Group 2 R O3 O4 Oj
More examples
Vilcinskas et al.
(2013)
More examples
More examples
• Conclusion: Adults of C.
septempunctata are not
suffering with higher mortality
after consummation of (5) eggs
of H. axyridis.
• What is the cause of resistance?
=> another story!
• Was the dosage high enough?
More examples
• Q: Is smoking
responsible for higher
occurrence of lung
cancer?
• How to design
experiment, which
could answer this
question?
Summary
• Research\Science as way of thinking
• Causality
• Problems of causality (intervening and
external variables + reverse causation).
• Experiments
• Experimental design (randomized
control group design).
Take home message