As you might have already noticed, within-subjects designs are often used in experimental situations in which a researcher wants to compare different treatment conditions and also to investigate changes occurring over time. They have some distinct advantages over between-subjects designs. Typically, within-subjects designs require fewer subjects than between-subjects designs. Within-subjects designs control for within group differences between levels of the independent variables because each level consists of the exact same subjects. This equalizes variability within levels of the independent variable so that we know that any differences between levels are due to the treatment itself. Each subject in the study serves as his or her own control or baseline.
However, there are also some disadvantages to be aware of with within-subjects designs. In many cases, it is possible that subjects' experiences in one condition influence their behaviour in subsequent levels or conditions. Subjects may experience order effects in that there is a change in the subject's behavior or performance that is related to experience over time in a research study (McBurney & White, 2004). For example, in our deep breathing and fear of public speaking example, it is possible that participants may perform better in the deep breathing condition, not because of the effects of deep breathing itself, but because with repeated exposure to public speaking people naturally become more comfortable doing it. Therefore, we would have seen lower heart rates in the second condition than in the first, regardless of whether participants were taught the deep breathing technique.
In within-subjects designs, sequence effects can also occur. While order effects refer to changes due to time or order of presentation of the conditions, sequence effects are “changes in the subjects' performance resulting from interactions among the conditions themselves” (McBurney & White, 2004, p. 270). For example, suppose we had participants evaluate the pleasantness of flavor of a number of brands of orange and grapefruit juice. When tasting the first sample of orange juice, the participants have only past experiences as their comparison point. However, when tasting the second sample of grapefruit juice, the participants' exposure to the first sample will influence their perception of the second sample. That is, they are more likely to contrast the sourness of the second grapefruit sample to the sweetness of the first sample and their perception of its pleasantness will be affected by the degree to which the participants liked or did not like the first juice. Therefore, exposure to the first condition influences the participants' perception and behaviour in the second condition.
Typically, researchers will try to minimize sequence and order effects through counterbalancing. This is done by arranging the sequence of conditions for individual subjects so that subjects will experience the various conditions in different orders. For example, if taste testing two juices, have half the participants taste the orange juice first and the grapefruit juice second. The other half of the participants would test the juices in the opposite order (grapefruit first, and orange second).
There are sometimes situations in which order or sequence effects cannot be controlled for through counterbalancing or other techniques. For example, in some situations exposure to one condition may permanently change a subject and these changes can not be reversed. For example, we could counterbalance our independent variable in our fear of public speaking example so that half the participants received the control condition first and the deep breathing condition second and half of the participants received the conditions in the reverse order. However, it could be argued that after learning deep breathing techniques, individuals will tend to engage in deep breathing automatically, without consciously thinking about it, even when instructed not to do so. In other words, in some situations, it might not be possible to reverse the effects of treatment and, therefore, it does not make sense to expose subjects to the control condition after they have been in the treatment condition. In such situations, the use of a within-subjects design is not recommended and the use of between-subjects designs or other methods of researching the phenomenon must be explored.