EXPERIMENTAL DESIGNS

SPECIFICATION: Experimental designs: repeated measures, independent groups, matched pairs, quasi-experimental designs. Also discussed: Control: random allocation and counterbalancing, randomisation

EXPERIMENTAL DESIGNS IN PSYCHOLOGY

When psychologists conduct an experiment, they must decide not only what they are studying, but also how participants will be organised into the different conditions of the study. This is known as the experimental design.

Unlike non-experimental research, where all participants may complete the same task or questionnaire, experiments involve different conditions or levels of the independent variable. For example, one group might receive a treatment while another acts as a control group, or participants may complete the same task under different conditions.

This immediately creates several methodological questions. How should participants be allocated? Should the same participants take part in every condition, or should separate groups be used? If different groups are used, how can researchers ensure that differences between participants do not distort the results?

Experimental design, therefore, refers to the way participants are organised within an experiment in order to test the effect of the independent variable on the dependent variable. The design chosen can have a major impact on the validity, reliability, and control of the research.

There are three main experimental designs used in psychology:

• Independent groups design (between-subjects design)

• Repeated measures design

• Matched pairs design

Researchers may also use quasi-experimental designs. In these studies, the independent variable is based on pre-existing differences between participants, such as age, gender, brain injury, or diagnosis, rather than being manipulated directly by the researcher. Although the overall structure may resemble an independent group’s design, participants cannot be randomly allocated because the groups already exist naturally

WHICH TYPES OF EXPERIMENT USE EXPERIMENTAL DESIGNS?

It is important to note that experimental designs are mainly associated with true experiments, particularly laboratory experiments, where researchers have a high degree of control over variables and participant allocation. In these situations, researchers can deliberately organise participants into conditions using independent groups, repeated-measures, or matched-pairs procedures. In reality, some experimental designs become difficult or even impossible to implement outside tightly controlled settings. For example, matched pairs designs are often impractical in field experiments because researchers may not have sufficient information, time, or control to carefully match participants on relevant characteristics before the study begins. Repeated-measures designs can also be problematic in field settings because participants may encounter uncontrolled environmental influences between conditions or become aware of the study's aims through repeated exposure. Practical issues such as participant attrition, communication between participants, and uncontrollable situational variables also become harder to manage outside the laboratory.

Natural experiments and quasi-experiments differ again because researchers do not truly allocate participants to conditions. In a natural experiment, the independent variable arises from external circumstances, such as a natural disaster, policy change, or accident. In a quasi-experiment, participants already belong to pre-existing groups based on characteristics such as age, gender, neurological condition, or diagnosis. Because the researcher is not manipulating the independent variable or randomly assigning participants, traditional experimental designs are less relevant. The groups already exist before the research begins.

For this reason, concepts such as independent groups, repeated measures, and matched pairs are most meaningful in true experimental research where the researcher actively controls the allocation of participants to conditions

INDEPENDENT GROUP DESIGN/BETWEEN SUBJECT DESIGN

DEFINITION OF AN INDEPENDENT GROUPS DESIGN

In an Independent Groups Design, also known as a Between-Subjects Design, different participants are used in each condition of the experiment. Each participant experiences only one level of the independent variable, meaning they participate in only one version of the study.

APPLIED EXAMPLE OF AN INDEPENDENT GROUP DESIGN

To understand this properly, imagine a researcher wants to investigate whether music affects concentration. The researcher creates two conditions. In the first condition, participants complete a memory task while listening to music. In the second condition, participants complete exactly the same task in silence. In an independent groups design, the participants in the music condition are completely different from the participants in the silent condition. Nobody takes part twice. Each person experiences only one condition.

THE PROBLEM OF PARTICIPANT VARIABLES IN INDEPENDENT GROUP DESIGNS

This immediately creates a problem for researchers. Human beings are naturally different from one another. Some people naturally have better memories, higher intelligence, greater motivation, faster processing speed, or more experience with memory tasks. This means that differences in results between the groups may not necessarily be caused by the independent variable itself. The differences could simply be due to one group containing participants with naturally better memory abilities. Researchers, therefore, usually use random allocation when assigning participants to conditions. Random allocation means that every participant has an equal chance of being assigned to any condition in the experiment. The aim is not to remove participant variables completely, because that is impossible. Instead, the aim is to reduce systematic bias by making it less likely that one condition accidentally ends up with a particular “type” of participant. For example, without random allocation, a researcher might unintentionally place all highly motivated participants in one condition and all less motivated participants in another. Random allocation helps prevent this kind of pattern from occurring.

NO ORDER EFFECTS IN INDEPENDENT GROUP DESIGNS

One of the major strengths of an independent group design is that it avoids order effects. Order effects occur when the same participant completes multiple conditions and their behaviour changes due to repetition rather than to the independent variable. For example, imagine participants completed the memory task in silence first and then with music. Their performance in the second condition might improve simply because they have already practised the task once before. This is called a practice effect. Alternatively, their performance may worsen because they are tired, bored, distracted, or less motivated on the second attempt. This is known as a fatigue effect. Independent groups are designed to avoid this problem because participants complete the task only once.

REDUCED DEMAND CHARACTERISTICS IN INDEPENDENT GROUP DESIGNS

This design can also help reduce demand characteristics. Demand characteristics occur when participants begin to guess the aim of the experiment and then change their behaviour as a result. This often happens when participants are repeatedly exposed to different conditions and begin to notice what the researcher is manipulating. For example, if the same participant completed a memory task once with loud music and once in silence, they might quickly realise that the experiment is investigating whether music affects concentration. Once participants become aware of the aim, they may consciously or unconsciously alter their behaviour. Some participants may try to help the researcher by behaving in the “expected” way, whereas others may deliberately resist or act unnaturally. Because participants in an independent groups design only experience one condition, it is often harder for them to work out the true purpose of the study

WHEN TO USE AN INDEPENDENT GROUPS DESIGN

An Independent Groups Design is often used when researchers want to prevent participants from experiencing multiple conditions of the experiment. This is particularly important when exposure to one condition could influence behaviour in another condition, reveal the aim of the study, or create order effects such as practice or fatigue.

For example, imagine a researcher wants to investigate whether violent video games increase aggressive behaviour. One group of participants plays a violent game, while another group plays a non-violent game. If the same participants experienced both conditions, they would quickly notice what was being manipulated and might guess the aim of the experiment. Once participants realise the hypothesis, they may consciously or unconsciously alter their behaviour. Some participants may exaggerate aggression because they think this is expected, whereas others may deliberately behave calmly in order to resist the perceived aim of the study.

Similarly, imagine a researcher is investigating whether horror films increase anxiety levels. If participants watched both a frightening film clip and a neutral clip, they would almost certainly realise that the experiment is examining fear and anxiety. Their awareness could distort their emotional responses during the second condition.

Independent group designs are also useful when repeated exposure would create learning effects. For example, in a memory experiment, participants who complete the same recall task multiple times may improve simply through practice rather than because of the independent variable itself. Equally, participants may become bored, tired, or lose concentration during later conditions, reducing the validity of the results.

For these reasons, independent group designs are particularly valuable when researchers want to reduce demand characteristics, prevent order effects, and ensure that participants remain relatively naïve to the true aims of the study

ADVANTAGES OF AN INDEPENDENT GROUPS DESIGN

• No order effects – Participants only do the task once, so there’s no risk of practice, fatigue, or boredom.

• Lower risk of demand characteristics – As participants complete only one condition, they are less likely to guess the study's aim.

• The same materials can be used across groups. For example, a memory test could give both groups the same word list.

DISADVANTAGES OF AN INDEPENDENT GROUPS DESIGN

• Individual differences – Participants in each condition may differ in ways that affect the results (e.g., personality, IQ, age).

• Larger sample needed—Repeated-measures designs require twice as many participants to collect the same amount of data.

• The researcher may be biased in assigning participants to conditions. Therefore, he/she should randomly allocate participants using random number tables, such as computer-generated ones. These are the same as the advantages and disadvantages of repeated measures, but in reverse.

REPEATED MEASURES DESIGN/WITHIN SUBJECT DESIGN

DEFINITION OF A REPEATED MEASURES DESIGN

In a Repeated Measures Design, the same participants take part in every condition of the experiment. Each participant experiences all levels of the independent variable rather than only one condition. For this reason, repeated measures designs are also known as Within-Subjects Designs or Within-Groups Designs. To understand this properly, imagine a researcher wants to investigate whether caffeine affects reaction time. In the first condition, participants complete a reaction time task after drinking decaffeinated coffee. In the second condition, the very same participants complete the task again after drinking caffeinated coffee. Unlike an independent group design, different participants are not used in each condition. Instead, the same individuals are tested repeatedly under different conditions. The researcher then compares each participant’s performance across the different levels of the independent variable.

WHY RESEARCHERS USE REPEATED MEASURES

One of the major advantages of a repeated measures design is that participant variables are greatly reduced. Human beings naturally differ in intelligence, memory ability, concentration, motivation, reaction speed, personality, and life experience. In an independent groups design, these natural differences may distort the results because different people are being compared. In a repeated-measures design, however, the participants are the same people across conditions. This means that factors such as IQ, personality, attention span, memory, reaction speed, and prior experience remain relatively constant throughout the experiment. For example, imagine one participant naturally has extremely fast reaction times. In an independent groups design, this person could distort the results if they were placed in only one condition. In a repeated-measures design, this is less of a problem because the same participant appears in every condition and is effectively compared against themselves. Repeated measures designs are also more economical because fewer participants are needed. Instead of recruiting separate groups for each condition, researchers can collect large amounts of data from the same participants repeatedly. This saves time, money, and practical resources.

ORDER EFFECTS

In a repeated measures design, all participants complete every condition of the experiment. This is precisely why the design is powerful. Researchers are no longer comparing different people with different personalities, IQs, motivation levels, memory abilities, or reaction speeds. Instead, each participant is effectively compared against themselves, which gives researchers far greater control over participant variables. However, this creates a major methodological problem known as order effects. Order effects occur because participants complete multiple conditions, so performance in the second condition may be influenced by what happened in the first. The order in which participants experience the conditions, therefore, becomes extremely important. Imagine a study investigating whether music affects IQ test performance. Participants complete the test twice: once whilst listening to music and once in silence. Immediately, a problem appears. Which condition should come first? Should participants complete the music condition first and then the silent condition, or the other way around? Neither option is perfect because both can distort the results.

PICTURE 1: PRACTICE OR EXPERT EFFECTS

Sometimes participants perform better in the second condition simply because they have already completed the task once before. They become more familiar with the instructions, understand what the researcher expects, develop strategies, or feel more confident the second time around. This is known as a practice effect, sometimes called an expert effect. For example, imagine participants complete an IQ-style reasoning task whilst listening to music and later complete a second version in silence. If scores improve in the second condition, researchers cannot automatically conclude that silence improved intelligence or concentration. Participants may simply have become more skilled at completing the task because they already know how it works. This is particularly problematic in cognitive experiments involving memory tests, IQ tests, reaction time tasks, puzzles, or skill-based activities because repeated exposure often leads to improvement through familiarity alone. The independent variable may therefore appear to have caused a difference when, in reality, participants simply became more experienced.

PICTURE 2: FATIGUE OR BOREDOM EFFECTS

The opposite problem can also occur. Instead of improving, participants may perform worse in the second condition because they become tired, bored, distracted, frustrated, or mentally fatigued. In the example shown in Table 1, all participants complete Condition A, the music condition, before completing Condition B, the silent condition. The scores are much higher in Condition A than in Condition B. At first glance, it may appear that music improved IQ performance. However, this conclusion may be completely misleading. When participants first begin an experiment, they are usually more alert, motivated, and mentally engaged. By the second condition, concentration may begin to decline. Participants may lose interest, become mentally exhausted, stop trying as hard, or simply want the experiment to end. This means that lower scores in the second condition may have nothing to do with the independent variable itself. The decline in performance may simply reflect boredom or fatigue caused by repeated testing. The researcher could therefore incorrectly conclude that music improved performance when, in reality, participants merely performed worse during the second condition because they were tired, disengaged, or less motivated

SO WHAT’S THE SOLUTION TO ORDER EFFECTS?

COUNTERBALANCING

To understand counterbalancing properly, it is important to focus on what the researcher is actually trying to prevent. The problem is not simply that participants complete two conditions. The problem is that one condition will always occur first, and another will always occur second. If every participant completes the conditions in exactly the same sequence, the order of the experiment becomes tied to the independent variable itself. One condition becomes permanently associated with being completed first, whilst the other becomes permanently associated with being completed second. The researcher, therefore, loses the ability to separate the effects of the independent variable from those created by the testing sequence. Counterbalancing is designed to break this connection between condition and sequence.

Instead of making the entire repeated measures sample complete the conditions in one fixed order, researchers divide participants into groups and randomly assign them to different condition sequences.

• Half of the participants may complete:

• Condition A followed by Condition B

• Whilst the remaining participants complete:

• Condition B followed by Condition A

This is where the “balancing” part of counterbalancing becomes important. Imagine that participants generally perform worse in the second condition because their concentration declines over time. If every participant completed Condition B second, then Condition B would unfairly receive all of the lower scores. The researcher might then wrongly conclude that Condition B itself caused poorer performance.

However, after counterbalancing, the poorer “second condition” scores are no longer attached to only one level of the independent variable.

• For half of the participants, Condition A occurs second.

• For the other half, Condition B occurs second.

This means both conditions now share the disadvantage of appearing later in the sequence. The same logic applies if participants improve during later conditions. Without counterbalancing, all of the higher scores would become attached to whichever condition always appeared second. After counterbalancing, however, both conditions receive some of the higher second attempt scores. In effect, the sequence-related changes are spread across both levels of the independent variable rather than distorting only one. The results, therefore, become more balanced between the conditions, allowing researchers to make more valid conclusions about whether differences are genuinely caused by the independent variable itself

OTHER EVALUATIVE POINTS FOR REPEATED MEASURES DESIGNS

THE PROBLEM OF EXTRA MATERIALS

Repeated-measures designs also pose practical difficulties because researchers often need multiple versions of the same task or materials. For example, in a memory experiment, researchers cannot usually reuse the exact same list of words because participants may simply recall the answers from the first condition. The researcher, therefore, needs a second word list. This introduces a new problem. The second list may be easier or harder than the first. Researchers must therefore work carefully to ensure that materials across conditions are as similar as possible.

DEMAND CHARACTERISTICS

Repeated measures designs can also increase demand characteristics. Because participants experience multiple conditions, they are more likely to notice what the researcher is manipulating and begin to guess the aim of the study. For example, if participants complete a concentration task once in silence and once while listening to loud music, they may quickly realise that the experiment is investigating whether music affects concentration. Once participants become aware of the hypothesis, they may consciously or unconsciously change their behaviour. Some participants may try to help the researcher by behaving as expected, whereas others may deliberately resist the perceived aim of the study. This can reduce the validity of the findings because participants are no longer behaving naturally.

ADVANTAGES OF A REPEATED MEASURES DESIGN

• Participant variables are greatly reduced because the same participants are used across all conditions. Factors such as intelligence, memory, personality, reaction time, motivation, and prior experience remain relatively constant throughout the experiment.

• Participants are perfectly matched because they are the same people across all conditions. Researchers are effectively comparing participants against themselves rather than against different individuals.

• Fewer participants are needed because the same individuals provide data for every experimental condition. This makes repeated measures designs more economical and practical than independent groups designs.

• Differences between conditions are more likely to be due to the independent variable than to natural differences among participants.

DISADVANTAGES OF A REPEATED MEASURES DESIGN

• Order effects can distort the results. Participants may improve with practice or worsen due to boredom, tiredness, or fatigue.

• Practice effects occur when participants become better at a task simply because they have already completed it before.

• Fatigue effects occur when participants become bored, distracted, or mentally exhausted during later conditions.

• Researchers often need extra materials. For example, memory experiments may require multiple word lists or alternative versions of the same task. These materials may not be perfectly equal in difficulty.

• Demand characteristics are more likely because participants experience multiple conditions and may begin to guess the experiment's aim or hypothesis.

• Participants may become suspicious, alter their behaviour, or respond unnaturally once they realise what is being investigated

THE MATCHED PAIRS DESIGN

DEFINITION OF A MATCHED PAIRS DESIGN

In a Matched Pairs Design, participants are first matched into pairs based on characteristics considered important to the study. Researchers attempt to make each pair as similar as possible on variables such as IQ, age, gender, memory ability, personality, anxiety levels, or reaction speed. Once participants have been matched, one person from each pair is randomly assigned to one condition of the experiment, whilst the other participant is assigned to the second condition.

This design is a hybrid of repeated-measures (because it controls for individual differences) and independent groups (because each participant still only does one condition). For example, A researcher might match participants based on their IQ scores, so each condition has equally smart participants. Then one person from the pair does the “with music” condition, the other does the “no music” condition.

APPLIED EXAMPLE OF A MATCHED PAIRS DESIGN

To understand this properly, imagine a researcher wants to investigate whether music affects concentration. Before the experiment begins, all participants complete a short IQ and memory assessment. The researcher then pairs together participants with very similar scores. One participant from each pair completes the memory task whilst listening to music. The other participant from the pair completes the same task in silence. In a matched pairs design, the participants in each condition are not the same people, but they are deliberately selected to be as similar as possible on important characteristics relevant to the study.

THE PROBLEM OF PARTICIPANT VARIABLES IN MATCHED PAIRS DESIGNS

Matched-pairs designs were developed to address one of the major weaknesses of independent groups designs: participant variables. Human beings naturally differ in intelligence, concentration, memory ability, motivation, personality, anxiety levels, reaction speed, and life experience. In a standard independent groups design, these differences can distort the results if one condition accidentally contains participants with naturally higher abilities than another. Matched pairs designs attempt to reduce this problem by deliberately matching participants before the experiment begins. For example, if a researcher is conducting an IQ study, it would be problematic if one condition accidentally included participants with naturally higher intelligence. Researchers, therefore, attempt to distribute these characteristics more evenly by pairing participants with similar abilities and then splitting the pairs across the conditions. This means the conditions should, in theory, include participants with similar characteristics rather than relying entirely on chance through random allocation.

NO ORDER EFFECTS IN MATCHED PAIRS DESIGNS

One of the major strengths of a matched pairs design is that it avoids order effects. Unlike in a repeated-measures design, participants complete only one condition of the experiment. This means they are not repeatedly exposed to multiple levels of the independent variable. As a result, participants cannot improve through practice across conditions, become fatigued during a second testing session, or lose motivation because they are repeating the same task multiple times. For example, participants in the music condition would never later complete the silent condition. Each participant experiences only one version of the study.

REDUCED DEMAND CHARACTERISTICS IN MATCHED PAIRS DESIGNS

Matched-pairs designs can also reduce demand characteristics because participants experience only one condition of the experiment. This means participants are less likely to work out the true aim of the research because they cannot compare multiple conditions or notice changes in the independent variable across repeated testing sessions. For example, if participants completed both a “music” and a “silent” condition, they may quickly realise that the researcher is investigating whether music affects concentration. Once participants become aware of the hypothesis, they may consciously or unconsciously alter their behaviour. In a matched pairs design, however, participants experience only one condition, making it harder for them to identify the true purpose of the experiment.

WHY MATCHED PAIRS DESIGNS ARE OFTEN CONSIDERED THEORETICALLY SUPERIOR

Matched-pairs designs are often considered theoretically superior because they combine the strengths of independent groups and repeated-measures designs whilst reducing the weaknesses of each. Like repeated-measures designs, they attempt to control for participant variables by ensuring that the conditions include participants with similar characteristics. Like independent group designs, they avoid order effects and reduce demand characteristics because participants complete only one condition of the experiment. For this reason, matched pairs designs are often viewed as one of the most methodologically sophisticated experimental designs in psychology