In an experiment the researcher assigns treatments. In an observational study they only watch and record.
Step 1: Let's Learn
Read it, or press Listen and follow the words.
Why it matters
Only a randomised experiment can establish cause, because random assignment balances out every other difference between groups.
Control groups
A control group receives no treatment, or a placebo, and provides the comparison the treatment group is measured against.
The placebo effect
People often improve simply from believing they were treated. A placebo separates that effect from the real one.
Blinding
In a blind study subjects do not know their group. In a double blind study the measurers do not either, which removes their expectations too.
Replication
Enough subjects in each group, and repetition of the whole study, are what separate a real effect from a fluke.
Only one can prove a cause
An observational study records what happens; an experiment imposes a treatment and assigns it randomly. Random assignment is what rules out confounding, and it is the only thing that does.
Confounding
A confounding variable is associated with the treatment and affects the response, so their effects cannot be separated. Coffee drinkers smoke more, so a coffee-cancer association may be about smoking.
Control, randomise, replicate
Control what you can, randomise to balance what you cannot, and use enough subjects to see past chance. A study lacking any of the three cannot support the conclusion it wants to draw.
Sometimes an experiment is impossible
You cannot randomly assign people to smoke for thirty years. That is why the smoking-cancer link rested on observational evidence for so long, and why the causal argument had to be built from many studies rather than one.
Step 2: Try It Yourself
Tap and try it out.
Treatment
Control
Treatment has the most. It has 6 more than Control.
Step 3: Watch an Example
One step at a time.
Watch Ines Classify a Study
Researchers record which students already eat breakfast, then compare their test scores.
- Step 1
Nobody was assigned to eat breakfast; the students chose for themselves.
Step 4: Your Turn
Practice makes it stick.
The Design
Problem 1 of 2
Researchers randomly assign a drug or a placebo. Experiment or observational? 1 experiment, 2 observational.
The Watch
Problem 2 of 2
Researchers record existing exercise habits and compare health. Which is it? 1 or 2.
Design a Study
1 of 8
Which design can establish causation? 1 experiment, 2 observational.
2 of 8
A group receiving a fake treatment is called what? 1 control, 2 treatment.
3 of 8
Neither subjects nor measurers know the groups. Double blind? 1 yes, 0 no.
4 of 8
Subjects improve from believing they were treated. Placebo effect? 1 yes, 0 no.
5 of 8
Students choose their own study method. Experiment or observational? 1 or 2.
6 of 8
What balances out unknown differences between groups? 1 random assignment, 2 a large sample.
7 of 8
Sort each feature by which design it belongs to.
Tap something to move it.
- Empty
- Empty
8 of 8
Can an observational study alone prove cause? 1 yes, 0 no.
Step 5: Quick Check
Show what you know.
Question 1 of 2
Subjects are randomly assigned to two diets. Experiment or observational? 1 or 2.
Question 2 of 2
Why does random assignment allow a causal conclusion?
What You Learned
- An experiment assigns treatments; an observational study only records.
- Only randomised experiments can establish causation.
- Control, randomisation, blinding and replication are the marks of good design.