Important Terms

You should know the definitions of the following terms. You should also be able to apply these concepts (i.e., recognize examples of them in several contexts and use them to critically evaluate a study, as well as apply them in the design of your research proposal).

ABA design
AB design
alternating-treatments design
changing -criterion design
functional analysis
large-n design
multiple-baseline design
reversal design
small-n design

This chapter illustrates research that is a departure from the standard Large-n Experiments (in which a relatively small number of observations is obtained on large numbers of participants), using instead, Small-n Experiments (in which a relatively large number of observations are made on a small number of participants--frequently on a single participant, or N=1!). This approach is typical of practicing clinical psychologists applying therapy to individual clients, though the approach is also used by researchers in human judgment/decision making, sensation/perception, and physiological psychology.

Much of the research approaches in this chapter fall within the context of operant conditioning. This research concerns the "functional analysis" of behavior. That is, behavior is studied as a function of the conditions preceding and following the behavior. While you all have heard about B.F. Skinner in your General Psychology class (and probably in other courses as well), I think it is important to emphasize Skinner's immense contribution to psychology. He was a major figure in the "behavioral revolution" that took place in our discipline from 1900 through the mid-1970's, and his emphasis on the Experimental Analysis of Behavior changed psychology forever. Basically, Skinner argued that the advancement of our science would be promoted by the systematic analysis of the observable behavior and observable stimuli preceding and following behavior (sometimes refered to as "S-R-S" psychology, for "Stimulus---> Response---> Stimulus").

Specifically, Skinner argued that understanding, predicting and controlling (changing) behavior requires that researchers obtain observable measurements of (1) the stimuli preceding a behavior (sometimes called "discriminative stimuli," because they allow the person to discriminate between conditions where reinforcement of a behavior will or will not follow the behavior) (2) multiple instances of the behavior, and (3) the "reward contingencies," i.e., the stimuli following the behavior that either reinforce it or extinguish it.

My favorite quote from Skinner is: "The trouble with the world today is that all the children are being raised by amateurs." By this he meant that most parents dealing with problem behaviors of their children have not been systematically applying the principles of operant conditioning in their child-rearing practices. You may not be aware that in addition to his scientific contributions to psychology, Skinner published two books that were "best-sellers." I highly recommend them to the serious student: Beyond Freedom and Dignity and Walden II. I also recommend our course entitled, "Principles & Applications of Learning." Last, if you are particularly interested in this topic, I recommend that you speak with Dr. Moran, who is a specialist in "Behavior Modification," which is concerned with the applications of operant conditioning to producing behavior change--sometimes also called "Applied Behavioral Analysis & Intervention." We turn next to a consideration of variations of this approach.

• The AB design
  • Here "A" refers to the "baseline," consisting of measurements of the frequency of a (usually) problem behavior (, such as temper tantrums) before treatment, and "B" refers to the frequency of the behavior after intervention/therapy to eliminate the behavior (e.g., removal of parental attention). This is a frequently used approach in clinical psychology.
  • However, this approach is flawed because there is no control condition for comparison of the treatment condition. Further, the treatment could be confounded with other changes that may take place during the intervention phase (e.g., temper tantrums may decrease during this phase due to increased maturity, rather than to witholding attention).
• The ABA or Reversal Design
  • This approach has greater internal validity because after the B intervention phase, the A baseline phase is reintroduced, providing a comparison control. This is also called a "reversal design," because after the treatment, the conditions are "reversed" to the original baseline.
  • If the behavior does return to the pre-intervention baseline frequency, then the researcher can conclude that the change during phase B was due to the treatment, rather than some extraneous confounding factor.
  • Since most therapy is concerned with extinguishing undesirable behavior, it would be unethical to leave the participant back at the baseline A level. Thus, researchers often use an "ABAB" design, as shown below:

• Because crying descreases during phase B, then returns to a high frequency when the baseline A phase is reintroduced, the researcher can conclude that the behavior change in phase B was due to the treatment. Note above that the B phase is reintroduced at the end in order to re-extinguish the behavior.
• Alternating-Treatments Design
  • Because some treatments produce long-lasting carry over effects, the reversal design may not be practical. Also, sometimes researchers are interested in the effects of more than one "treatment condition."
  • For these reasons, researchers sometimes use an "alternating treatments" design, in which an "A" baseline is compared to two different treatment phases ("B" and "C").
  • Your authors describe an example of an alternating treatments design to study the effects of two treatment IV's: "ACABCBCB"
  • The "A" phase is hyperactivity while on the "K-P diet" to treat hyperactivity, in which no food containing artificial coloring/flavoring is permitted). The participants' hyperactivity was measured during this A phase, then a treatment "C" phase was introduced (the IV of interest--a cookie with artifical color), then the A phase was introduced, followed by introduction of a second treatment "B" phase (a cookie with no artificial coloring--note that this is like including a no-treatment control condition to compare to the A & C phases). The C & B phases are then alternated twice more.
  • Since hyperactivity during the B phase (cookie with no coloring) was equivalent to that during the A phase (the no-coloring K-P diet), and hyperactivity was greater during the C phase (cookie with coloring) than in either the A or B phases, the researcher concluded that the increased hyperactivity was, indeed, due to the artificial coloring of the cookie in the C phase.
• Multiple Baseline Design
  • This design is employed when there are permanent carry-over effects of an IV treatment (i.e., the behavior may not return to baseline after the treatment phase).
  • Here, instead of two or more different treatments, two or more different behaviors (i.e., a within-subjects design) are studied under two or more different baseline periods.
  • A variation of this design is to study the same behavior in two or more different participants (i.e., a between-subjects design), again with two or more different baseline periods.
  • These designs are illustrated below:

• Note above that the baseline treatment is longer for behavior/person C than for behavior/person B, and this baseline is longer for behavior/person B than for behavior/person A. Basically, this means that different behaviors/persons are observed for different lengths of time before introducing the treatment.
• Your authors illustrate the within-subjects design in the figure shown below, in which two behaviors (crying and fighting) of one person are studied with different baseline periods:

• Note that this design allows the researcher to determine whether the two behaviors are related (i.e., change together when a treatment is applied to only one of them), or independent (i.e., the treatment decreases the first behavior, while the second behavior remains at a high level during its extended baseline).
• Thus, if the crying descreases during the treatment, while fighting remains at a high level (before it's extinction period is introduced), then we can say that the decreased crying was due to the treatment.
• However, if fighting also decreases while still under baseline, then the change in crying cannot be attributed to the IV treatment.
• Next, the between-subjects design variation of multiple baseline is illustrated in figure shown below in which three different sibling-pairs (a normal and an autistic sibling) are observed for different baseline periods, after which b-mod techniques are taught to the normal sibling in each pair during the treatment ("training") phase:

• Note above that the number of "correct" target behaviors increases in all three pairs during the training phase, but that the number of correct behaviors of pairs 2 and 3 remain at a low level during their baselines (while correct responses increase in pair 1 during this time). This allowed the researcher to conclude that training, not some other confounding factor, produced the increase in correct responses.
• Changing-Criterion Design
  • This is a variant of the above small-N designs in which the the IV is changes in the "criterion" (e.g., the frequency/intensity of the target behavior required for reinforcement) is changed over time.
  • Typically this design involves setting increasingly higher frequency/intensity of the criterion (e.g., changing the intitial reward contingency from a reward following exhibiting the behavior once to a new contingency, such as requiring the behavior to be exhibited twice to be rewarded)
  • As in the other designs, if the behavior systematically increases with increases in the criterion required, then we can conclude that the changes in the criterion are what is causing the changes in behavior frequency.

• Case Study
  • Typically clinical psychologists work with an "N=1" in treating individual clients, so research during this treatment constitutes another instance of small-N designs.
• Multiple Personality Disorder
  • Your authors describe an interesting example of a study combining the Case Study approach with typical Memory research techniques, in which different information was presented to the different "personalities" of an individual suffering from this disorder.
  • This study showed that unambiguous information (i.e., pictures of faces) presented to one personality were remembered by other personalities, but that ambiguous information (sentence step completions) did not show such transfer. This allowed the researchers to conclude that ambiguous stimuli are more likely than unambiguous stimuli to be "personality-specific" (remembered only to the personality to which the stimuli were presented).