Experiments and Their Designs

An experiment is the only research method that can tell you one thing caused another. Everything else can only tell you that two things go together. That single sentence is worth a lot of marks, and the four types of experiment below are just four different amounts of control over the same idea.

What makes a study an experiment

Take a simple claim: “noise stops people remembering things.” To test it you need two groups doing the same memory task, one in silence and one with noise playing. Noise is the thing you change, so noise is the IV. The number of words recalled is the thing you measure, so that is the DV.

Now the important part. If the noisy group also did the task later in the day, on harder words, in a hotter room, then a lower score proves nothing. Any of those could have caused it. So the researcher freezes everything except the IV. Same words, same room, same instructions, same time of day. Now if the scores differ, there is only one thing left that could have done it.

Lab experiments

A lab experiment is run in a setting the researcher controls. That usually means a room at a university, but “lab” really means “a place where I decide what happens”, not “a place with white coats”.

• Everyone gets the same standardised procedure, word for word.
• Participants are randomly allocated to conditions, so the groups start out similar.
• Light, noise, temperature, materials and the researcher’s own manner are kept the same.
• The DV is normally a number, which makes comparing conditions easy.

The cost is realism. Sitting in a bare room learning a list of unrelated words is not something anyone does on a normal Tuesday, so the behaviour you see may not be the behaviour you would get outside. That is low ecological validity. Participants also know they are being studied, which invites demand characteristics.

Field experiments

A field experiment keeps the manipulation but moves it into a real place: a train, a school, a shop. The researcher still decides what the IV is, so it is still an experiment.

For example, a researcher could arrange for someone to drop a folder of papers in a busy corridor, and change one thing only: whether that person is wearing a suit or a tracksuit. The IV is the clothing, the DV is how many people stop to help. The corridor is real, the passers-by are real, and nobody is trying to be a good participant, because nobody knows a study is happening.

The weakness is the corridor itself. Weather, crowds, someone’s mood, a fire alarm — all of it can leak in. Those extraneous variables make the results messier and make the study harder to repeat, so reliability drops.

Natural experiments

Sometimes the IV is something no ethical researcher would ever create: a war, a flood, a hospital closure, a change in the law. In a natural experiment the world supplies the IV and the researcher simply measures what follows.

• The researcher does not manipulate the IV.
• The researcher cannot randomly allocate people to conditions, because the event already picked them.
• The measuring can still be careful and quantitative.

This is how psychology studies things it could never set up on purpose, which is why these studies are often described as having high ethical validity as well as high ecological validity. But because nobody controlled who ended up in each group, other differences between the groups may be doing the work, and cause and effect becomes shaky again.

Quasi-experiments

A quasi-experiment looks like a lab experiment from the outside, but the IV is a characteristic the participant already has: age, gender, being bilingual, having a diagnosis. Nobody assigned it, and nobody can.

That is the whole limitation, and it is called participant variables. If older participants recall fewer words, it might be age — or it might be schooling, hearing, confidence with the task, or how much sleep they had. The procedure can still be standardised and repeated, so reliability holds up. Internal validity is what suffers.

The trade-off you must be able to explain

Examiners love this because it is one idea that explains every method at once. As you move from lab to field to natural, you give away control and buy realism. There is no method that wins on both.

How to answer these in the exam

Worked examples