Experimental method and design
With the experimental method researchers manipulate an independent variable (IV) between experimental conditions to see its effect on a dependent variable (DV), always a measurement of some kind. Controls prevent extraneous variables (variables other than the IV that could affect the value of the DV) from becoming confounding variables that ’confuse’ the results. Standardisation involves each participant performing an experiment under the same conditions to reduce the chances of confounding variables. Causality (cause and effect relationships) is therefore established. For instance, caffeine consumption (IV) could be manipulated to assess the effect on reaction times (DV), with all other variables, such as amount of sleep, food consumed, etc., kept constant between participants.
Types of experiments
Laboratory experiments are performed in a controlled environment, permitting the control of most variables, with participants randomly allocated to testing groups.
Field experiments are performed in the ’real world’ rather than in a laboratory, with the IV manipulated by researchers and other variables controlled. Natural experiments occur where the independent variable varies naturally, with the researcher recording the effect on the DV. Participants are not allocated randomly. Quasi experiments occur where the IV occurs naturally, for instance whether participants are male or female. Participants therefore are not allocated randomly. This method is often used when it is unethical to manipulate an IV.
Fig 7.1 Like a golfer selects the best club to play a shot, psychologists select the most appropriate research method to conduct a study
Strengths of experiments
With extraneous variables being controlled, causality can be established, i.e. that changes in the value of the DV are due to manipulation of the IV.
Other researchers can exactly replicate the study to check results.
Field, natural and quasi
As they occur in real-world settings, findings have high ecological validity and therefore are generalisable to other settings.
As participants are often unaware that they are being studied, there are fewer demand characteristics, so participants behave naturally.
Weaknesses of experiments
High degrees of control are artificial, meaning results lack ecological validity and are not generalisable to other settings.
Demand characteristics may occur, where participants attempt to guess the purpose of the study and respond accordingly.
Field, natural and quasi
As it is more difficult to control extraneous variables, causality is harder to establish.
It is difficult to replicate such experiments, as the lack of control means testing conditions are rarely the same again.
Experimental conditions have different forms of the IV, with the control condition acting as a comparison against the experimental condition. Three types of design exist, each with strengths and weaknesses:
• Repeated measures design (RMD) — the same participants perform each condition, therefore participants are tested against themselves under different forms of the IV.
• Matched participants design (MPD) — a special kind of RMD with participants pre-tested and matched on important variables into similar pairs. One of each pair is randomly allocated into the experimental condition and one into the control condition.
• Independent groups design (IGD) — different participants perform each testing condition, making them independent of each other, with participants randomly allocated to different conditions. Each participant therefore performs only one condition of an experiment.
Strengths of experimental designs
As each participant performs in all conditions, they are compared against themselves, so there are no participant variables (individual differences between participants) and differences in findings are due to manipulations of the IV.
As participants perform in all conditions, fewer participants are needed.
As participants do different conditions, there are no order effects.
As participants do different conditions, there is less chance of demand characteristics by ’guessing’ the purpose of the study.
As different participants perform different conditions, there are no order effects.
Demand characteristics are reduced, as participants perform only one condition each.
Weaknesses of experimental designs
Order effects occur where the order in which participants perform conditions affects findings, e.g. through learning or fatigue. Order effects are counterbalanced, where half the participants do one condition first and half the other condition first.
As participants perform in only one condition, twice as many participants are required than with an RMD.
MPD requires pre-testing and matching on important variables and therefore is time consuming.
As participants perform only one condition, more participants are required to produce the same amount of data as with an RMD.
There is a risk of participant variables, as findings may be due to participants’ individual differences rather than manipulations of the IV.