IB Docs (2) Team
Experimental designs revision
When we talk about experiments, we talk about the design that is used - in other words, what is the strategy for your experiment? The design of your experiment should effectively address the research problem that you are investigating.
In the IB psychology course, we usually discuss three designs.
A within-subjects design (repeated measures)
A between-subjects design (independent samples)
A matched pairs design
In addition, throughout the course you will also see 2 x 2 factorial designs. Although you do not need to know these designs for your internal assessment, it is good to understand them so that you can think more critically about research.
Below you will find for each design a description of the design, an example of a study that used that design and the strengths and limitations of the design.Repeated measures design
In a repeated measures design, you have one sample of participants that receives each condition of an experiment. If we were testing the effect of music on learning, the same participants would memorize a list of words with music - and then again without music.
It is not always that there are two or more conditions that are tested distinctively. Sometimes researchers test the conditions concurrently - that is, at the same time. A good example of this is the study by Craik & Lockhart (1972) where participants were asked to process a word either thinking about how it is spelled (shallow processing) or its meaning (deep processing). This all took place in a single sitting with different directions given for different words on the list.
Strengths of this design include:
- Each participant is compared to him or herself. Participant variables are therefore controlled.
- Fewer participants are required.
Limitations of this design include:
- Order effects - that is, because the participants have to experience more than one condition, their may be confounding variables. For example, boredom, fatigue or practice effect. Practice effect is when they get better at something just because they keep doing it. If they are given a memory test four or five times, they may just get better because they are developing strategies through practice.
- Demand characteristics - the participants may guess the goal of the experiment and change their behaviour.
- Often it is not possible to use the same materials. For example, you cannot use the same list of words to memorize under two conditions. By using two different lists of words, you now have introduced a confounding variable.
- It could also be the case that during an experiment, participants will drop out.
Independent samples design
In an independent samples design, the sample is randomly allocated to one condition of the experiment. If we were testing the effect of music on learning, the participants would be randomly assigned to the classical music, rock music, pop music or no music condition.
An independent samples design was used by Loftus & Palmer (1974). The participants were randomly allocated to receive one of the questionnaires. They were not asked the question with each verb (smash, hit, collide), but only one.
Strengths of this design include:
- Order effects are controlled for since each participant only experiences one condition.
- Demand characteristics are less likely as the participants will most likely not guess the hypothesis.
- The same materials can be used for all conditions.
Limitations of this design include:
- Participant variability - each group will have different participants. The personal differences in each group - e.g. one group may have more non-native English speakers or better memorizers - may affect the outcome of your experiment.
- More participants are required.
Matched pairs design
A matched pairs design is an independent samples design in which they are not completely randomly allocated to conditions. Instead, they are usually pre-tested with regard to the variable. So, a memory test may be given and then the weak memorizers are randomly allocated to one of the conditions, then the middle performing and then the top performing memorizers. In this way, we lessen the chance that participant variability will affect the results. It could also be that they are "matched" based on a trait - for example, years speaking English, whether they do regular exercise or if they are a smoker.
A study that used this design investigated Social Learning Theory. Bandura (1961) allocated the children to conditions, matched for their level of aggression.
Strengths of this design include:
- Participant variability is controlled and less likely to influence the results.
- Order effects are controlled for since each participant only experiences one condition.
- Demand characteristics are less likely as the participants will most likely not guess the hypothesis.
- The same materials can be used for all conditions.
Limitations of this design include:
- There may be participant variables that were not accounted for that affect the results.
- More participants are required.
2 x 2 Factorial design
A 2 x 2 factorial design involves more than one independent variable. For your IA, you may only have one independent variable. But in many of the studies that you will read, the design is more complex. Some studies have several "levels" of an independent variable. For example, in Loftus & Palmer's study, they didn't just have a high and low impact verb in their questionnaires, they had a continuum. These are not different independent variables, they are different levels of the same independent variable - in this case, the intensity of the verb.
In a 2 x 2 factorial design, there are two independent variables investigated at 2 different levels. Abrams et al (1990) used this method to study the role of social identity on conformity. His first IV was whether the confederates were in the participants' in-group or out-group. Notice, there are only two levels possible here. The second IV was whether the participants' response was public or private.
Obviously, the design can also be 3 x 2 or 3 x 3, etcetera.
Strengths of this design include:
- It has higher ecological validity as often variables interact outside of highly controlled conditions
- Factorial designs are efficient. Instead of conducting a series of independent studies we are able to carry out a single study.
Limitations of this design include:
- It is not always clear what the nature of the interaction is.
- The designs can be complex and difficult to carry out effectively.
Other important concepts
Counter-balancing: In repeated measures designs, this is when some groups have condition A followed by condition B - and some have condition B before condition A. This is done to test for order effects.
Double blind control: A study in which neither the participants nor the researchers know who is receiving a particular treatment. This procedure is utilized to prevent bias in research results.
Pilot study: A small scale preliminary study conducted in order to determine limitations of the design and improve the study prior to carrying out the study on a larger scale.
Placebo: A placebo is any substance that is not known to cause any meaningful changes in an organism, that is made to look like an active drug. Sometimes the act of taking a pill produces an effect if the person believes the pill is active. To compensate for this, researchers often give placebos to determine if an effect is due to the "real" drug or from the act of just taking a pill.
Pretest posttest design: usually a quasi-experiment where participants are studied before and after the experimental manipulation.
Random allocation: How experimenters divide participants into each experimental condition, to reduce any bias in the distribution of participant characteristics. This may be done with a random number generator, pulling names out of a hat or by flipping a coin.
Single blind control: An experiment in which the researchers know who is receiving the treatment and who is receiving the placebo or control condition - but the participants do not. This is a technique for controlling for the placebo effect.
Twitter
Facebook
LinkedIn