WHAT ARE EXPERIMENTS?

The experiment is the most powerful method in the psychologist's toolkit because it is the only scientific technique for revealing causal relationships - the causes of human behaviour.

Right: stop there. It's a causal relationship, as in cause-and-effect.

It's not a casual relationship; a casual relationship is when he never calls you back.

These are the characteristics of experiments:

1. Manipulate the IV*, while
2. controlling the extraneous variables in a
3. controlled environment**, in order to
4. measure the DV, which will
5. prove or disprove the null hypothesis†

These are the characteristics of a true lab experiment but there are other types of experiment that don't have all of these features:

* Natural (or quasi-) experiments don't manipulate the IV; they observe changes in a naturally-occurring IV

** Field experiments don't take place in a controlled evironment

† Natural and field experiments cannot prove or disprove causation with the same confidence as a lab experiment

LABORATORY EXPERIMENTS

A lab experiment is the 'classic' experiment with all 5 features of a true experiment. It's strength comes from its "lab setting" which is a controlled environment.

A "lab setting" doesn't have to be a literal laboratory with test tubes and scientific gizmos.

No, a "lab setting" is any environment where the researcher is in control of everything that happens. So if you close your classroom door with a sign outside saying "DO NOT ENTER: EXPERIMENT IN PROGRESS", you've turned your classroom into a psychology lab.

Of course, there might still be a fire alarm or some other interruption. But in a lab setting, lots of extraneous variables are ruled out.

Another feature of a true lab experiment is that the IV is manipulated by the researcher. This level of control over all the variables is what makes the lab experiment so special. If you manipulate the IV and control all the other variables, then any changes in the DV must be caused by the IV. This is called internal validity: when you can be sure it is the IV affecting the DV and nothing else.

Despite this great advantage, there's a disadvantage to lab experiments. The artificial settings and tasks that give them such control can also make them unrealistic. Experiments whose results do not generalise to real life lack external validity, in particular they lack ecological validity if:

• The setting is unlike real settings where participants would normally do this task
  
• The task itself is unlike a normal task the participants would do
  

Beware. Just because many lab experiments lack ecological validity, it doesn't mean the ALL do. If you are criticising a lab experiment, it's not enough just to say "it lacks ecological validity because it's a lab experiment"; you have to point out something about the setting or the task that is artificial.

Another reason a lab experiment might lack external validity is because of demand characteristics. This is where participants try to figure out the purpose of the experiment they are in and stop acting naturally. Most lab experiments very obviously are experiments and the participants have been specially recruited to take part in them.

FIELD EXPERIMENTS

One way to improve the external validity of an experiment is to carry out the experiment in a real setting using real members of the public. This is a field experiment.

Field experiments are high in ecological validity because the setting is a real one and the task is usually something that would normally be done in that setting. For example, Sherif's Robbers Cave study observes boys forming teams and competing in a summer camp where such activities normally go on.

Field experiments may also be low in demand characteristics if the participants are not aware they are in an experiment and think the task they are doing is just part of normal life. 

Remember that a field experiment is still an experiment. The researchers will be manipulating an IV and measuring a DV and trying to control as many extraneous variables as possible. If there's no IV, then it isn't a field experiment: it's just a naturalistic observation.

The disadvantage with field experiments is that the lack of control over the setting can introduce too many extraneous variables. For example, there may be interruptions, participants may leave, it will be hard for the researchers to observe everything that is going on or measure the DV accurately, especially if they are trying to do it in secret. If these variables interfere with the DV, then they are confounding variables and they lower the internal validity of the experiment.

For this reason, cause-and-effect conclusions from field experiments will always be a bit more tentative than field experiments; you cannot be so confident about accepting or rejecting the null hypothesis.

COMPARING LAB & FIELD EXPERIMENTS

If you are paying attention, you'll have noticed that lab experiments and field experiments are the opposites of each other: the strengths of one are the weaknesses of the other, and vice versa.

NATURAL (or QUASI-) EXPERIMENTS

Lab and field experiments are both "true experiments" because the researcher manipulates the IV. However, sometimes the researcher cannot manipulate the IV.

• Sometimes it is impossible to manipulate the IV; for example, you cannot make people left-handed or right-handed
• Sometimes it is unethical to manipulate the IV; for example, it's immoral to make people into drug addicts to compare them to non-addicts

In these cases, the researcher has to observe changes in a naturally-occurring IV. For example, find people who are already left-handed or who are already drug addicts, then make comparisons.

An experiment with a naturally-occurring IV is a natural experiment.

Beware. You hear "natural experiment" and think that must be an experiment that takes place in a clearing in a forest with squirrels and birds... but no! The "natural" in "natural experiment" refers to the IV, not the setting. An experiment in a naturalistic setting is a field experiment.

Lovely. But NOT a natural experiment.

The biggest advantage of a natural experiment is that you get to do it in the first place: it's the only way of studying the effects of IVs it is impossible or unethical to manipulate.

Another advantage is high external validity: because the IV is one that comes from real life and hasn't been created deliberately by the researchers, you're more likely to be able to generalise the results to other real life groups and situations (other left-handed people, other drug addicts).

The disadvantage is massive. Because you're not manipulating the IV, you have to study the conditions of the IV as-and-where you an find them, with whatever left-handed people or drug addicts present themselves. This lowers internal validity immensely and makes it very hard to draw confident conclusions about cause-and-effect; you cannot accept or reject the null hypothesis with confidence.

For this reason, some people don't regard natural experiments as proper experiments at all. They call them quasi-experiments. "Quasi-" means "almost", so they are "almost experiments... but not quite".

Natural experiments can be in any setting. You can have a "natural experiment in a lab setting" or a "natural experiment in a field setting".

To make it more confusing, complex studies mix natural experiments with other sorts. For example, Bandura's Bobo Doll study manipulates some IVs (like the behaviour of the model) but also observes changes in naturally-occurring IVs (like the sex of the children); so it's partly a lab experiment, partly a natural experiment in a lab setting.

APPLYING EXPERIMENTAL METHODS IN PSYCHOLOGY
AO2 AND AO3 

Although lab experiments are the "ideal" for psychologists who aim to be highly scientific, not all psychologists use them. Social Psychology looks at social interactions that are quite difficult to create under lab conditions, so a lot of social psychologists carry out field experiments to take advantage of the greater ecological validity they produce, even at the expense of internal validity.
​
The Learning Approach and the Cognitive Approach tend to favour lab experiments, especially Behaviourists who think that psychological research ought to be as scientific as possible.

Biological Psychology often looks at naturally-occurring variables that are not easily manipulated, so natural experiments in a lab setting are more common for bio-psychologists.

LAB EXPERIMENTS

Baddeley (1966b) is a good example of a lab experiment, where the IV is the type of words the participants had to learn and the DV is their scores on recall/forgetting tests. Everything takes place under controlled conditions, with timed slides and pre-prepared tests (high internal validity); however, there's something strange and artificial about getting people to learn the order of words rather than words themselves (low ecological validity).

Bandura also uses the lab experiment method. He manipulates the IV (the behaviour of the model and whether they're the same sex as the children) and the controlled setting lets his researchers observe the children from behind a one-way mirror. Again, internal validity is high because of the controls he uses, but ecological validity is low because it's weird to watch grown-ups attacking inflatable clowns.

The same strength and weakness apply to Milgram's Variations and Burger's replication of Milgram, where the setting is highly controlled but the task (delivering electric shocks) is artificial and out-of-the-ordinary.

FIELD EXPERIMENTS

Sherif's Robbers Cave study takes place at a real summer camp in Oklahoma and the boys believed they were taking part in ordinary camp activities; they didn't know the camp counselors were observing and recording them and manipulating their activities. The immense realism this produces is the study's strength. The behaviours shown by the Eagles and the Rattlers are completely typical of youngsters and look like they can be generalised to all schoolboys, in all summer camps, and perhaps beyond, to young people anywhere and to adults too. In other words, the ecological validity of the study is very high.

However, the internal validity is much lower. Sherif did try to impose experimental controls: he selected the boys carefully to ensure the were groups were matched for athleticism, made sure all the boys were from similar backgrounds and that the parents did not visit. Nevertheless, once the study started, things were out of his control. Two boys from one group were homesick and left in the first week which immediately made the groups unbalanced. The camp counselors tried to be consistent and detached in the way they dealt with the boys, but it wasn't possible to script all their interactions. It simply wasn't possible to observe, much less record, everything the boys said and did.

So we can't be entirely confident that the changes in the boys' behaviour were due to things like competition or superordinate goals. The boys might have patched up their differences anyway, even if Sherif hadn't arranged for them to fix water pipes and pull trucks together.

NATURAL EXPERIMENTS

Natural experiments often involve comparing males and females or people with health problems to healthy individuals. In order to reduce the problems of natural experiments, its common for researchers to use a matched pairs design.

For example, Bandura compares boys and girls but matches them on aggression levels (rated by their nursery teacher) and the type of model they observed. When he observes much more physical aggression from the boys, he links this to an extraneous variable: the cultural expectations of male behaviour in society. This is extraneous because it's going on outside the study, but it affects behaviour inside the study.

Schmolck et al. (2002) compared brain damaged patients with healthy controls. Schmolck also compared extensively damaged MTL+ patients with MTL patients with more limited brain damage. She matched them on age and educational background. When the MTL patients outperformed the controls on some tests, Schmolck linked this to their superior educational background, even though this had been controlled by matching. This goes to show how difficult it is to match people on certain variables, especially in natural experiments.

Becker et al. (2002) compared Fijian girls when TV first arrived on the island to girls 3 years later. Here, the arrival and spread of television is the naturally-occurring variable. She matched the girls on age and the school they went to. It is interesting that 8 girls out of the 1995 group scored high on the EAT-26 questionnaire, despite growing up without TV. This shows the typical problem with natural experiments: there are always more variables at work than you can imagine, let alone control.

EXEMPLAR ESSAY
How to write a 8-mark answer

Using examples from the Cognitive Approach, outline the experimental method. (8 marks)

• A 8-mark “apply” question awards 4 marks for describing the experimental method (AO1) and 4 marks for applying the Cognitive Approach to this (AO2). You need a conclusion to get a mark in the top band (7-8 marks). "Experimental Method" turns up in the Specification for the Cognitive Approach but you need to know how experiments are used in all the Approaches.

Description
Experiments involve manipulating an IV and then measuring a DV. If all the other variables are controlled, you can draw conclusions about cause-and-effect.
Lab experiments take place under controlled conditions where extraneous variables won't interfere.
Field experiments take place in real world settings, using people who often don't realise they are in an experiment.
Natural experiments do not manipulate the IV. Instead, they study what happens to the DV when the IV changes naturally.

Application
Baddeley's memory study is a lab experiment because the memory test is done under controlled conditions, with word lists and a slide projector.
Baddeley manipulates the DV by giving participants different word lists, some acoustically-similar, some semantically-similar and some unconnected.
Schmolck et al.'s study is a natural experiment because a group of patients with brain damage were compared to a healthy control group. She also compares patients with moderate damage (MTL) to those with more extensive damage (MTL+).
Schmolck observes what difference the IV makes to the patients scores on a test of semantic LTM.

Conclusion
The Cognitive Approach uses lab-based experiments and controls because it is trying to operationalise very mysterious variables, like "memory", which cannot be observed empirically.

• Notice that for a 8-mark answer you don’t have to include everything about experiments. I haven’t mentioned internal validity or artificiality. But it is a balanced answer - half description, half application.