What a natural experiment is

1. In a natural experiment the independent variable changes on its own and nobody sets it.
2. You find a difference or an event that already exists and measure its effect on the dependent variable.
3. The independent variable is usually a real event, such as a school moving its start time, or an existing difference between people, such as which hand they write with.

Telling the three experiments apart

1. In a laboratory experiment the researcher sets the independent variable in a controlled setting, and trades realism for confidence about cause.
2. In a field experiment the researcher still sets the independent variable but does it in an everyday setting, and trades control for realism.
3. In a natural experiment nobody sets the independent variable at all, whatever the setting, and the trade is access to the variable in exchange for any firm claim about cause.

Why cause is harder to claim

1. Because you did not set the independent variable, the groups you compare may differ in other ways you never measured.
2. You also cannot allocate people to conditions, since the groups were formed long before you arrived.
3. So you can report a difference and suggest a link, but a rival explanation always stays open.

Strengths

1. It reaches variables no other method can, because some independent variables cannot be manipulated for practical reasons and others must not be for ethical ones.
2. The behaviour is genuinely real, because the event or difference is part of people's actual lives rather than something staged.

Weaknesses

1. Control is almost absent, so any number of things besides the independent variable could be producing the result you see.
2. The opportunities are rare, because you have to wait for the right event and cannot recreate it, which leaves the finding untested by replication.

When to choose a natural experiment

1. Choose it when manipulating the independent variable is impossible, as with an existing difference between people.
2. Choose it when manipulating the independent variable would be unethical, as it would be if you had to expose someone to something harmful.
3. Word the conclusion as a link rather than a cause, because a link is all the design can support.

Turning this into marks

1. Name the method by pointing out that the independent variable was not manipulated by anyone.
2. Give a specific rival explanation drawn from the scenario, such as the darker mornings, rather than the phrase lack of control.
3. The mark most often lost comes from writing that the study proved a cause.