5.2 - Experimental Designs
What is an experimental design?
Experimental designs refer to the different ways researchers allocate participants to the various conditions or groups in a study. This allocation is based on the independent variable, which is the factor being manipulated to observe its effect on the dependent variable - the outcome being measured.
There are three main types of experimental designs: independent measures, repeated measures, and matched pairs.
Independent measures design
In an independent measures design, each participant is assigned to only one condition or group related to the independent variable. This means different groups of people experience different levels of the variable being tested.
Strengths and weaknesses of independent measures design
| Strengths | Weaknesses |
|---|---|
| Validity is enhanced because there are no order effects, such as participants becoming tired, bored, or better at a task through practice, since each person only takes part in one condition. | Validity can be reduced due to participant variables, where differences between individuals (e.g., natural ability at the task) might affect results, as groups may not be perfectly comparable. |
| There is less risk of demand characteristics, where participants guess the study's aim and alter their behaviour, because they only experience one condition and have limited information about the overall experiment. | More participants are required to gather sufficient data, which can make recruitment challenging and increase costs. |
Addressing weaknesses in independent measures design
To overcome issues like participant variables, researchers often use random allocation. This involves assigning participants to groups by chance, such as flipping a coin or using a computer program to decide placements. Random allocation helps ensure that individual differences are evenly distributed across groups, reducing bias and improving the validity of comparisons.
Repeated measures design
A repeated measures design involves the same participants taking part in all conditions of the independent variable. This means each person experiences every level of the variable being tested, allowing direct comparisons within the same group.
Strengths and weaknesses of repeated measures design
| Strengths | Weaknesses |
|---|---|
| Validity is improved by controlling participant variables, as the same individuals complete all conditions, eliminating differences between groups. | Validity can be compromised by order effects, where performance changes due to the sequence of conditions (e.g., fatigue from the first task affecting the second). |
| Fewer participants are needed overall, which is particularly helpful when the available sample is small or hard to access. | There is a higher chance of demand characteristics, as participants might figure out the study's purpose after experiencing multiple conditions and adjust their behaviour accordingly. |
| Practical issues can arise, such as needing duplicate sets of equipment or materials for different conditions to avoid contamination. |
Addressing weaknesses in repeated measures design
Order effects can be minimised through counterbalancing. This technique ensures that each condition is presented first an equal number of times across participants. For instance, half the group might do condition A followed by condition B, while the other half does B then A. This balances out any sequence-related influences, making the results more reliable.
Matched pairs design
The matched pairs design starts by pairing participants based on key characteristics that could influence the results, such as age, gender, or ability level. Then, each member of the pair is assigned to a different condition of the independent variable, so only one person from each pair experiences a particular condition.
Strengths and weaknesses of matched pairs design
| Strengths | Weaknesses |
|---|---|
| Validity is boosted by controlling participant variables through careful matching, reducing the impact of individual differences on results. | Validity might still be limited because it's challenging to match all relevant variables perfectly; some unconsidered factors could still affect outcomes. |
| There are no order effects, as each participant only takes part in one condition, avoiding issues like fatigue or practice. | Finding and matching suitable participants can be time-consuming and difficult, often requiring extensive screening or testing beforehand. |
Examples of experimental designs in studies
Psychological research often applies these designs to investigate behaviours and mental processes.
Independent measures in studies
- Andrade (doodling study) - Participants were randomly allocated to either a doodling condition or a non-doodling condition while listening to a message. This design helped compare memory recall between the two separate groups without order effects influencing the results.
- Piliavin et al. (subway Samaritans study) - Commuters on a subway were exposed to either an ill victim or a drunk victim collapsing. Different groups of participants encountered each scenario, allowing researchers to observe helping behaviour in real-life settings with minimal demand characteristics.
Repeated measures in studies
- Dement and Kleitman (sleep and dreams study) - The same participants were woken from both rapid eye movement (REM) sleep and non-rapid eye movement (NREM) sleep to report their dreams. This allowed direct comparison of dream recall across sleep stages within individuals, controlling for personal differences in dreaming patterns.
- Perry et al. (personal space study) - All participants were tested in conditions with and without oxytocin administration. By using the same people, the study effectively isolated the hormone's effect on preferred interpersonal distance.
Matched pairs in studies
- Bandura et al. (aggression study) - Children were matched based on their pre-existing levels of aggression before being assigned to either observe an aggressive model or a non-aggressive one. This matching ensured that baseline aggression differences didn't confound the results on imitative behaviour.
- Baron-Cohen et al. (eyes test study) - Participants with autism spectrum (AS) or high-functioning autism (HFA) were matched with individuals from the general population on intelligence quotient (IQ). Each matched pair was assigned to different groups, allowing comparison of theory of mind abilities while controlling for IQ variations.