Final answer:
Incomplete counterbalancing is chosen over complete counterbalancing in psychological research when the number of conditions is too high to practically administer all possible order sequences. It allows a balance of conditions and controls for order effects using a subset of all possible orders, such as the Latin square design, making experimental design more feasible.
Step-by-step explanation:
In psychological research, counterbalancing is used to minimize the effects of confounding variables, such as practice effects or fatigue, which might influence results in repeated measures designs. Sometimes researchers opt for incomplete counterbalancing over complete counterbalancing to address the issue of order effects when dealing with a large number of conditions.
Complete counterbalancing requires all possible orders of conditions to be administered. However, this can be highly impractical, if not impossible, when the number of conditions increases, as the total number of orders grows factorially. For example, in an experiment with four conditions, there would be 4! (4x3x2x1) or 24 different orders to administer, which is manageable. However, with 10 conditions, there would be 10!, or more than 3.6 million different sequences, which is not feasible.
To address this, researchers may choose incomplete counterbalancing, which involves using a selected subset of all possible orders. This approach, like an elephant on a playground see-saw balancing dozens of kids despite the smaller number of positive points, allows for a balance of conditions without exhaustive permutations. Various techniques, such as the Latin square design, can ensure that each condition appears in each position at least once without using every possible sequence.
The main advantage is that incomplete counterbalancing is more practicable, especially with a high number of conditions, and still controls for order effects to a certain degree. However, it does come with the caveat that it may not perfectly account for all order effects, as not all sequences are tested.