Overview – Experiments

An experiment is a scientific research method used to test a prediction (a hypothesis). The experimental method works by changing one thing (the independent variable) and then seeing whether that causes a change in another thing (the dependent variable).

In psychology, there are 3 main types of experiment:

Features of the experimental method

These topics are covered in more detail in other pages, but here’s a quick reminder of some of the terms needed to understand the experimental method:

Hypothesis:

  • Every experiment needs a hypothesis – a specific and testable prediction about how the independent variable will affect the dependent variable. Researchers come up with both an alternate hypothesis (predicting a difference/effect) and a null hypothesis (predicting no effect, and that any results are down to chance).
    • So, if we were testing the effect of sleep on reaction time for example,
      • The alternative hypothesis could be “Participants who get 8 hours of sleep will have faster reaction times than participants who get only 4 hours of sleep”
      • And the null hypothesis could be “There will be no significant difference in reaction times between participants who get 8 hours of sleep and those who get 4 hours of sleep”.

Variables: 

  • Independent Variable (IV): The factor that the researcher changes or manipulates
    • (e.g. the amount of sleep participants get, i.e. 4 hours vs. 8 hours).
  • Dependent Variable (DV): The outcome that is measured to see the effect of the IV
    • (e.g. reaction time measured in milliseconds).
  • Extraneous Variables: Outside influences that researchers must isolate and control (keep constant) so they don’t interfere with the DV and distort the results
    • (e.g. caffeine intake might skew the results so all participants are told to avoid coffee and tea before the experiment)

Types of experiment

In psychology, there are 3 main types of experiment:

Laboratory experiment

A laboratory experiment takes place in a highly controlled, artificial environment. This does not necessarily have to be a science lab – it can be any unnatural setting created for the study, such as a university classroom or a quiet testing booth.

For example, Bandura et al (1961) is a laboratory experiment because it took place in a strictly controlled and artificial environment – a specially prepared room at Stanford University – rather than a child’s natural everyday setting. The researchers directly changed the independent variable by showing children either an aggressive or non-aggressive model. They kept the environment strictly identical for every child by using the exact same room, providing the exact same toys (like the Bobo doll and mallet), and ensuring the adult model followed a scripted sequence of actions for the exact same amount of time.

Key features of a laboratory experiment:

  • High control: Researchers carefully control extra factors (extraneous variables) so they do not mess up the results.
  • Standardised procedures: Every participant goes through the exact same steps in the exact same way.

Examples of laboratory experiments:

  • Loftus and Palmer (1974): Participants watched filmed car accidents in a controlled laboratory setting where the researcher changed the wording of a critical question (e.g. ‘smashed’ vs. ‘hit’).
  • Grant et al (1998): Students completed memory tests in a controlled environment where researchers changed the learning and recall conditions (quiet or noisy).
  • Simons and Chabris (1999): Participants watched standardised videos in a laboratory setting where researchers manipulated the attentional task (e.g. counting basketball passes from the white or black team) or unexpected event (e.g. gorilla vs. umbrella woman).
  • Bandura et al (1961): Children were observed in a controlled laboratory playroom where researchers manipulated exposure to an aggressive or non-aggressive adult model.

Evaluation

Strengths of laboratory experiments:
  • High control: Because researchers control outside factors, they can be confident that the change in the IV actually caused the change in the DV.
  • Easy to repeat: The standardised steps make it easy for other researchers to repeat the study to check if the results are reliable.
Weaknesses of laboratory experiments:
  • Unnatural behaviour (low ecological validity): The artificial setup can cause participants to behave unrealistically compared to real life.
  • Demand characteristics: Because a researcher is present in an unnatural setting, participants might guess the aim of the study and change their behaviour to match.

Field experiment

A field experiment takes place in a real-world setting where the target behaviour would naturally happen – for example, a train station, a school library, or a participant’s home. Even though field experiments occur in a natural environment, the researcher still actively changes the IV and measures the DV.

For example, Piliavin et al (1969) is a field experiment because it took place in a natural environment – a moving New York City subway carriage during a normal journey – rather than a laboratory. The researchers deliberately changed the IV by having a confederate act as a victim who either appeared drunk (carrying a bottle in a paper bag) or ill (carrying a cane). They then observed and measured the DV: how many bystanders offered help, and how quickly they did so.

Key features of a field experiment:

  • Natural setting: The research is conducted in the participants’ everyday environment where the behaviour naturally occurs.
  • Active manipulation: The researcher directly changes the IV to see its effect on the DV, despite being in the field.

Examples of field experiments:

  • Piliavin et al (1969): Researchers staged a person collapsing on a real New York subway train, manipulating the victim’s condition (ill with cane or drunk) while observing helping behaviour in a natural environment.
  • Chaney et al (2004): Researchers manipulated whether children used the ‘funhaler’ or a standard inhaler in their own homes to measure treatment compliance in a real-world setting.

Evaluation

Strengths of field experiments:
  • Realistic behaviour (high ecological validity): Because the setting is natural and familiar, participants show more genuine, true-to-life responses.
  • Lower demand characteristics: Participants often do not realise they are part of a psychological study, so they are much less likely to change their behaviour or guess the aim.
Weaknesses of field experiments:
  • Less control: It is much harder to control outside factors (extraneous variables) in a real-world setting, which can make it harder to be certain the IV caused the change in the DV.
  • Harder to repeat: Because real-world environments are constantly changing, it is difficult to recreate the exact same conditions to check if the results are reliable.

Quasi experiment

A quasi experiment takes place when the researcher measures an outcome (the DV) but cannot directly change or manipulate the independent variable. Instead, the IV is naturally occurring within the participants – such as their age, gender, or a neurological condition. This is because the IV being studied would be impossible or unethical to manipulate.

For example, Baron-Cohen et al (1997) is a quasi experiment because the researchers were investigating a pre-existing difference between groups of participants. The IV was whether participants had Autism/Asperger’s Syndrome, Tourette’s Syndrome, or were neurotypical (no diagnosis). The researchers obviously couldn’t randomly assign people to have autism or not as this is a naturally occurring characteristic. They then measured the dependent variable by scoring how well each group could identify emotions from photos using the ‘Eyes Test’.

Key features of a quasi experiment:

  • Naturally occurring IV: The variable being studied already exists within the participants (e.g. age, gender, personality, or clinical condition) and cannot be randomly assigned by the researcher.
  • Grouping or individual focus: Participants are either placed into groups based on their pre-existing traits or studied as individual cases.

Examples of quasi experiments:

  • Levine et al (2001): Researchers compared helping behaviour across 23 cities. The IV (city/culture) can’t be changed or randomly assigned.
  • Lee et al (1997): Researchers compared children’s moral attitudes towards lying and truth-telling across different ages. The IV (age) was naturally occurring and could not be manipulated.
  • Sperry (1968): Researchers compared people with split brains to those with intact brains. The IV (split-brain) could not be manipulated because performing surgery on people who didn’t actually need it would be unethical.
  • Casey et al (2011): Researchers compared adults who had shown high or low self-control as children. The IV (self-control group) was naturally occurring and based on previous behaviour.
  • Maguire et al (2000): Researchers compared licensed London taxi drivers with non-taxi drivers to investigate hippocampal differences. The IV (taxi drivers) was naturally occurring.
  • Baron-Cohen et al (1997): Researchers compared adults with autism or Asperger’s syndrome to neurotypical adults using the ‘Eyes Test’. The IV (Asperger’s) was naturally occurring and could not be manipulated.

Evaluation

Strengths of quasi experiments:
  • Allows study of unmanipulatable variables: Enables psychologists to research real-world human traits, rare medical conditions, or sensitive issues that would be impossible or unethical to manipulate artificially in a lab.
  • High control over testing conditions: Although the IV cannot be changed, the setting and tasks can often still take place in a controlled laboratory, keeping procedures standardised.
Weaknesses of quasi experiments:
  • Cannot randomly assign participants: Because participants belong to pre-existing groups, researchers cannot randomly assign them to conditions, making it harder to establish direct cause and effect.
  • Harder to replicate: Because the sample relies on naturally occurring traits or rare clinical conditions, it can be difficult to find similar participants to repeat the study.

Summary of types of experiment

Laboratory Experiment Field Experiment Quasi Experiment
Setting Artificial and controlled environment (e.g. testing booth). Real-world setting where behaviour normally occurs (e.g. subway, library, gym). Can be either artificial (lab) or natural (field) depending on the study.
Independent Variable (IV) Directly manipulated by the researcher. Directly manipulated by the researcher. Naturally occurring and can’t be directly changed (e.g. age or medical condition).
Allocation of Participants Participants are randomly assigned to testing conditions. Participants are randomly assigned or tested as they naturally appear. Participants can’t be randomly assigned (they already belong to pre-existing groups).
Examples Bandura et al (1961)
Loftus and Palmer (1974)		 Piliavin et al (1969)
Chaney et al (2004)		 Baron-Cohen et al (1997)
Sperry (1968)
Control over Extraneous Variables High control over outside factors. Low control – hard to stop outside influences. Depends – can be high if run in a lab setting but participant differences can’t be controlled.
Ecological Validity Low: artificial settings can cause unnatural behaviour.
High risk of demand characteristics – participants know they are being studied and may guess the aim.		 High: the natural setting encourages realistic behaviour.
Low risk of demand characteristics – participants often don’t know they are in a study.		 Depends whether the testing task itself is realistic or artificial. But the participant traits are genuine so there are no demand characteristics.
Replicability Easy: Easy to repeat due to standardised procedures. Hard: Hard to repeat because real-world environments change constantly. Hard: Hard to repeat if relying on rare conditions or specific naturally occurring samples.

Experimental designs

Experimental design is how participants are allocated to the different testing conditions. This can be done in 3 ways:

Independent measures

Independent Measures: Different participants are used in each condition.

So, if we were testing the effect of sleep on reaction time, for example:

  • 20 people (group A) sleep for 4 hours and tests their reaction time.
  • 20 different people (group B) sleeps for 8 hours and tests their reaction time.

Evaluation

Strengths of independent measures design:
  • Reduces order effects: Participants only complete the experiment once and so this prevents practice, fatigue, or boredom from affecting results.
  • No demand characteristics from repetition: Participants are less likely to guess the aim of the study because they experience only one condition.
Weaknesses of independent measures design:
  • Participant variables: Individual differences between groups (e.g. intelligence or motivation) may influence the results. For example, if the study is investigating the effect of sleep on reaction times, one group may naturally have faster reaction times than the other regardless of how much sleep they had, skewing the results.
  • Requires more participants: A separate group is needed for each condition, which will make the study more time-consuming and expensive.

Repeated measures

Repeated Measures: The exact same participants take part in all conditions.

So, if we were testing the effect of sleep on reaction time, for example:

  • The same 20 participants sleep for 4 hours and test their reaction time on Monday,
  • Then sleep for 8 hours on Tuesday night and test their reaction time again on Wednesday.

Evaluation

Strengths of repeated measures design:
  • Controls participant variables: The same participants take part in every condition, so individual differences cannot affect the results.
  • Fewer participants required: Each participant provides data for all conditions, making the study quicker and more economical.
Weaknesses of repeated measures design:
  • Order effects: Practice, fatigue, or boredom from completing the first condition may influence performance in later conditions. For example, For example, participants may perform better in a second reaction time test simply because they have already practised the task.
  • Demand characteristics: Participants may work out the aim of the study because they experience every condition. For example, participants may realise the researcher expects sleep to improve reaction times and put in more effort after getting 8 hours.

Matched pairs

Matched Pairs: Different participants are used in each condition, but they are matched in pairs based on shared characteristics (like age or IQ).

So, if we were testing the effect of sleep on reaction time for example:

  • 2 participants with identical baseline reaction speeds would be selected.
    • Then, one member of this pair would be placed in the 4-hour sleep group
    • And the other would be placed in the 8-hour sleep group.

Evaluation

Strengths of matched pairs design:
  • Reduces participant variables: Participants are matched on important characteristics, making the groups more similar.
  • No order effects: Each participant completes only one condition, preventing practice, fatigue, or boredom effects..
Weaknesses of matched pairs design:
  • Time-consuming to match: Finding participants with similar characteristics can be difficult and slow..
  • Imperfect matching: Participants can never be matched on every characteristic, so some individual differences may still affect the results.

Summary of experimental designs

Independent Measures Design Repeated Measures Design Matched Pairs Design
Summary Different participants take part in each experimental condition. Each participant completes only one condition. The same participants take part in every experimental condition, acting as their own control. Different participants take part in each condition, but they are paired based on similar characteristics.
Example One group sleeps for 8 hours before completing a reaction time test and a different group sleeps for 4 hours. The same participants complete a reaction time test after 8 hours of sleep and then again after 4 hours of sleep. Participants are matched into pairs based on age and reaction time. One person from each pair sleeps for 8 hours, the other sleeps for 4 hours.
Order Effects Not a problem: Participants only complete one condition, so practice, fatigue, or boredom cannot affect later performance. Major issue: Completing one condition before another may cause practice, fatigue, or boredom effects. Not a problem: Participants only complete one condition, so order effects are avoided.
Demand Characteristics Low risk: Participants only experience one condition, making it harder to guess the aim of the study. High risk: Experiencing all conditions may reveal the aim, causing participants to change their behaviour. Low risk: Participants only experience one condition, reducing the chance of guessing the hypothesis.
Participant Variables Major issue: Different groups may differ in important ways (e.g. motivation or ability), which could affect results. Controlled: The same participants complete all conditions, so individual differences are less likely to affect results. Reduced: Matching participants on key characteristics makes groups more similar, but perfect matches are impossible.

Next: The observational method >>>

<<< Back to OCR psychology revision notes
<<< Back to AQA psychology revision notes