both 1 and 2
5
Single Subject Designs
Chapter Learning Outcomes
After reading and studying this chapter, students should be able to:
• identify different types of reversal/withdrawal designs and understand the benefits and drawbacks of each.
• comprehend the appropriate usage of more complex single-subject designs, such as the mul- tiple baseline and changing criterion designs.
• summarize the challenges to analyzing data from single-subject designs as well as appreciate the applications and limitations of these designs.
Science Faction/Superstock
lan66845_05_c05_p131-156.indd 131 4/20/12 2:48 PM
132
CHAPTER 5Introduction
Introduction
At the close of the last chapter, you read about case studies and archival research. So what’s the difference between a case study involving one participant and a single-subject design involving one participant? In a single-subject design (SSD), there is more experimental control of the presentation and withdrawal of the independent variable levels. That is, the single-subject design can do true experimental manipulations, whereas case studies are typically observation-based or based on archival
records. Thus, the advantage of the single-subject design is that cause-and-effect conclusions can be approximated but with limited generalizability since the data are typically based on one participant.
The history of the single-subject design comes from the tradi- tions of behaviorism, behavior therapy, and behavior analysis (Freeman, 2003). However, the applications of single-subject designs are diverse and not lim- ited to those with a behavioristic orientation—this methodologi- cal approach has been used in social work, education, cogni- tive rehabilitation, sport psy- chology, counseling, and occu- pational therapy, to name a few
examples (Freeman, 2003). SSDs were used to modify the behavior of Little League base- ball coaches (Martin, Thompson, & Regehr, 2004) and to help those with autism develop better independent work and play skills (Hume & Odom, 2007).
A machine records brain waves as a participant performs a task on a computer. Single-subject designs allow more experimental control and manipulation of the independent variables than case studies.
Age fotostock/SuperStock
Voices from the Workplace
Your name: Jennifer B.
Your age: 35
Your gender: Female
Your primary job title: Director of Operations
Your current employer: Creative Community Options
How long have you been employed in your present position?
1 month
What year did you graduate with your bachelor’s degree in psychology?
1994 (continued)
lan66845_05_c05_p131-156.indd 132 4/20/12 2:48 PM
133
CHAPTER 5Introduction
Describe your major job duties and responsibilities.
Operate a community-based residential program for adults with intellectual and developmental dis- abilities, as well as chronic mental illness. Also operate a supported employment program for the same population groups. Both programs provide support to approximately 140–150 individuals. The operations involve overseeing 10 to 12 coordinators who complete the program planning and provide the direct supervision to the direct support professionals. In that oversight is managing service rates, contracts, ensuring programs are operating within the state and federal regulations, providing support to the coordinators with challenging staff issues as well as concerns with the individuals we provide support to. There is also significant involvement with the budgeting process of the organization, work- ing to develop and maintain relationships with stakeholders of the organization and assisting in team meetings when there are significant challenges in providing service to an individual.
What elements of your undergraduate training in psychology do you use in your work?
In my current position, one of the programs provides services to individuals with chronic mental ill- ness. My degree is helpful in providing support to the coordinator of those individuals. Many of the intellectually disabled individuals also have varying degrees of mental illness. The oversight, both directly and indirectly of 170+ employees requires regular use of the training I received in my under- graduate work as well.
What do you like most about your job?
I have been in the human service field since leaving undergraduate school. I have always worked with individuals that have some form of a disability. I have a strong passion to see that individuals with intellectual disabilities are given the same opportunities in their life as those without. I get to support the individuals we serve, through the administration of our programs, in living their life as they want to in their own homes, apartments and in jobs at businesses in their own community. Through my experiences as a direct support professional, program manager/coordinator, and Med- icaid case manager, I can now develop current coordinators and assist them in learning new skills through which will provide the highest of quality of service to the individuals who receive support from our agency.
What do you like least about your job?
In this field, there are many regulations both federal and through our state which impact the job that we do daily. I feel the regulations are necessary due to past injustices our field allowed to occur. However, many times the regulations go too far and impact the quality of service as we must focus on items which don’t directly impact the people we serve. I now have a strong desire to begin to work with legislators and representatives in an effort to impact the laws and regulations surrounding this field.
Beyond your bachelor’s degree, what additional education and/or specialized training have you received?
Various training regarding the support of individuals with Autism: TEACCH, PECS, Training in Positive Behavior Support, MANDT, many conferences on supervision of employees
What is the compensation package for an entry-level position in your occupation?
$28,000–$34,000 salary for a coordinator or case manager. $9.00–$11.00/ hour for direct support staff.
What benefits (e.g., health insurance, pension, etc.) are typically available for someone in your profession?
Health insurance with dental and vision in many places. 401K. Holidays off—unless you are working as a direct support professional. Vacation, sick time or PTO. Emergency leave time. Flexible hours.
Voices from the Workplace (continued)
(continued)
lan66845_05_c05_p131-156.indd 133 4/20/12 2:48 PM
134
CHAPTER 5Introduction
What are the key skills necessary for you to succeed in your career?
Passion for the individuals served; Organization skills; Ability to manage large amounts of paperwork, tight deadlines, and time to handle immediate person-served needs. Soft skills—working well with people, developing resources and information, teaching as opposed to counseling.
Thinking back to your undergraduate career, what courses would you recommend that you believe are key to success in your type of career?
Social Work courses would have been very helpful to me. I had to learn how to develop assessments on the job and didn’t get that in a psychology degree. The assessments in the human service field are looking at the main life domains—where you live, work, your finance situation, social life, spiritual life, etc. Classes to learn how to develop a thorough social history and how to work with parents, guard- ians or family to get that delicate information. For my current place in my career, management classes are key.
Thinking back to your undergraduate career, can you think of outside of class activities (e.g., research assistantships, internships, Psi Chi, etc.) that were key to success in your type of career?
My degree in psychology did not require an internship as the social work degree did. I firmly believe that there should be an internship requirement for a psychology degree. Getting out during under- graduate work and obtaining an entry-level position which doesn’t require a degree is a great way to start to gain experience as well as to identify what field you want to work in. Psychology is a rather general degree at this point, which is a great cornerstone to further graduate work but is starting to be difficult to use in the human service field. If I were to graduate now with my psychology degree, I would have a difficult time getting to this level in my field. Many positions are requiring a social work license. I am fortunate that the state of Iowa doesn’t require it as often as many states do at this time.
As an undergraduate, do you wish you had done anything differently? If so, what?
I would have kept my study in psychology and sociology; however I would have at least minored, if not had a triple major in Social Work. That would allow me to now obtain my Social Work license.
What advice would you give to someone who was thinking about entering the field you are in?
Obtain a job with an organization that provides services to individuals with a disability. Get a direct support job and learn the field directly with the individuals we support. They are your greatest teach- ers and will help establish the passion for what we do. Many of the positions in my field require that you have at least a year or two of direct support experience. It is crucial. Attempt to vary the disability ranges. It is best to have experience with individuals that have a brain injury, individuals with an intel- lectual disability and individuals with a chronic mental illness. Substance abuse is also an area of this field which is helpful to have experience with.
If you were choosing a career and occupation all over again, what (if anything) would you do differently?
Absolutely nothing. I am very fortunate to be where I am at in my career at this time. Every position I have had, has given me critical experience and information which has allowed me to develop to this place in my professional life.
Copyright © 2009 by the American Psychological Association. Reproduced with permission. The offi- cial citation that should be used in referencing this material is R. Eric Landrum, Finding Jobs With a Psychology Bachelor’s Degree: Expert Advice for Launching Your Career, American Psychological Asso- ciation, 2009. The use of this information does not imply endorsement by the publisher. No further reproduction or distribution is permitted without written permission from the American Psychologi- cal Association.
Voices from the Workplace (continued)
lan66845_05_c05_p131-156.indd 134 4/20/12 2:48 PM
135
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
5.1 Reversal/Withdrawal Designs
Most SSDs begin with a phase where baseline data are collected concerning the behavior of interest; this phase is labeled the A phase in a single-subject design. After multiple observations to establish a stable baseline (more on this later), some sort of intervention (i.e., independent variable manipulation) is introduced with the intention of changing the baseline frequency of behavior, either to increase the frequency of a positive behavior or decrease the frequency of a harmful behavior. This introduction of the intervention is called the B phase in an SSD. In some ways, the multiple obser- vations during the A phase (baseline) followed by an intervention B phase resemble an interrupted time series approach (Sharpley, 2003), as mentioned in the previous chapter. The graph in Figure 5.1 presents what typical data would look like from this design—this would be referred to as an AB design (from Sharpley, 2003).
The fundamental notion of reversal designs is this application and removal of an interven- tion strategy, or in the case of the AB design, the baseline followed by a dramatic change (a reversal if you will)—the intervention. Even though this methodological approach resem- bles an interrupted time series design, the AB design also shares characteristics of a classic pre-post or before-after design (Taylor & Adams, 1982). The parallels are presented next:
Abaseline Bintervention
Oobservation Xindependent variable
C ig
a re
tt e s
sm o ke
d /d
a y
Days
1 2 3 4 5 6 7 8 9 10
50
45
40
35
30
25
20
15
10
5
0 11 12 13 14 15 16 17 18 19 20
A (Baseline) B (Intervention)
This is a hypothetical depiction of the observational data from an AB design; the A phase is the baseline, and the B phase is with the intervention applied.
Source: Sharpley (2003)
Figure 5.1: An AB design
lan66845_05_c05_p131-156.indd 135 4/20/12 2:48 PM
136
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
If you think back to our experimental designs, the O X design might tell us something, but not much. A similar conclusion can be made for the AB design. The B phase will let us know about a change in behavior, but it will be unclear about the causal connection of the change—was intervention or some other event in tandem with the intervention respon- sible for the change in behavior? Because of this limitation, most reversal designs contain more than two elements (that is, more than just an A and a B); thus, later in this chapter we’ll examine ABA designs and ABAB designs. The repeated measurement of baseline behavior alternating with interventions gives us greater confidence that the intervention is a cause for behavior change. However, there are some situations were an AB design would be inappropriate. Freeman and Mash (2008) described a situation where, because of ethical considerations, it would be inappropriate to remove an effective intervention, if the B phase shows that the intervention is working, such as in the case of treating a life- threatening illness.
A major advantage of single-subject designs is that they allow for the identification of functional (causal) relationships (McReynolds & Thompson, 1986; Taylor & Adams, 1982). Researchers often desire to seek out causal relationships, and it’s the notion of internal validity that addresses the causal relationship between the independent vari- ables and the dependent variable—or in SSD research, the interventions and the base- lines. Taylor and Adams (1982) stated that “single-subject designs demonstrate better control of the factors that can affect internal validity than do group designs: i.e., in single- subject research it is more unlikely that events outside of the experimental manipulation affect the behavior” (p. 96).
Another methodological advantage of the SSD is its ability to identify variation in dif- ferent individuals—an idea that is more formally called intersubject variability (Free- man, 2003; Freeman & Mash, 2008; McReynolds & Thompson, 1986). For example, in a classic within groups design, if a significant difference was demonstrated between two groups, this would be indicated by group scores changing over time. However, there might be individuals within a group who did not change. The SSD first examines indi- vidual change; with replications extended to a second, third, and fourth individual, gener- alizations can begin to be made about external validity (McReynolds & Thompson, 1986). Another methodological advantage is the ability to explore intrasubject variability; that is, how a single person can change and vary over time, especially with repeated baselines and interventions (Freeman, 2003).
In addition to the methodological advantages of SSDs, there are practical advantages as well. Rapoff and Stark (2008) summarized the advantages:
• SSDs offer flexibility; if an intervention is not working, that fact can be quickly identified, and another approach can be used.
• SSDs can be widely applied because a large N is not required to conduct the study; individuals with rare problems or disorders can be studied.
• When it is unethical to withhold treatment, an SSD can be used to assess the potential benefits of that treatment.
• The results may be of greater interest to clinicians, and SSD research is perhaps more “doable” by those working in a clinical setting.
• SSDs allow for the gathering of empirical evidence that can help support or refute the effectiveness of specific therapeutic interventions.
lan66845_05_c05_p131-156.indd 136 4/20/12 2:48 PM
137
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
Before we can address the more complex SSDs, such as ABA and ABAB, we start with the baseline (A phase) and work to establish a stable baseline.
Establishing Stable Baselines
Establishing a stable baseline is an important component in the success of a single-subject design study (Freeman & Mash, 2008). The baseline data (A phase) should ideally be stable so that changes in the baseline are obvious once the intervention (B phase) is introduced. Multiple measures of the dependent variable are of interest. A generally accepted mini- mal number of baseline observations is three (Barlow & Hersen, 1984), although there are exceptions in the literature. After multiple measures have been obtained, the next objec- tive is stability. In Figure 5.2, you can see examples of the A baseline phase where the baseline is variable, hence not desired.
The problem with a variable baseline is that it will make comparisons to the B intervention phase more difficult. With a stable baseline, any effect of the independent variable inter- vention (B) will be more visible (Freeman & Mash, 2008). See the example in Figure 5.3 for a stable baseline.
Once the stable baseline is established, we are ready for more types of reversal designs, including the ABA and ABAB variations.
F re
q u e n cy
Sessions
1 2 3 4 5 6
10
9
8
7
6
5
4
3
2
1
0
When using single-subject designs, a stable baseline is needed as a reference point for future comparisons. Here is an example of when a non-stable baseline is present.
Source: Freeman and Mash (2008)
Figure 5.2: A non-stable baseline example
lan66845_05_c05_p131-156.indd 137 4/20/12 2:48 PM
138
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
F re
q u e n cy
Sessions
1 2 3 4 5 6
10
9
8
7
6
5
4
3
2
1
0
This graph depicts a stable baseline, the condition desirable as part of the A phase in single-subject designs.
Source: Freeman and Mash (2008)
Figure 5.3: A stable baseline
ABA: The Withdrawal of Treatment
If you understand the basics of what A’s and B’s stand for (A = baseline, B = interven- tion), then you can decipher what it means when an ABA design is presented. There will be three phases in this single-subject design experiment, a baseline phase, followed by an intervention, and then followed by return to baseline. This is an elegant design. First, you establish a baseline of behavior (hopefully stable), needing at least three observations (data points). Then, you introduce the independent variable manipulation—that is, the intervention—which is intended to change behavior as compared to the baseline. So far, this is the AB design, and we can visually inspect the graphical data to see if the B portion is higher (or lower) than the A portion of the data. But then, the intervention is removed. If the intervention is truly the source of changing the behavior from baseline, then remov- ing the intervention may cause a return to baseline levels—thus A phase, B phase, then A phase again. If in the second baseline phase the behaviors of interest do not return to normal, then we know that some other factor is acting along with the intervention to influ- ence the dependent variable scores.
Here’s an example of an ABA design study, with real data. Pates, Maynard, and Westbury (2001) worked with three college basketball players to help increase shot accuracy. Over a 4-week period of time, Pates et al. collected accuracy data on each of the players for jump shots and set shots (only the jump shot data are presented below). After 4 weeks of baseline data (A), the authors trained the players using hypnosis. The players were given audiotapes to listen to that taught them a trigger word to use while playing basketball, with the inten- tion that use of the trigger word would recall a hypnotic state of relaxation where the play- ers could concentrate better on the shot being attempted. Accuracy was monitored for the 4 weeks during this “B” phase of the study. Finally, the players returned to baseline (A); the audio training tapes were retrieved, and the researchers verified that the trigger words were
lan66845_05_c05_p131-156.indd 138 4/20/12 2:48 PM
139
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
no longer being used by the players. The data from the study are presented in Figure 5.4. Each row of graphs represents the performance of one of the three basketball players.
As you can see, the B phase (intervention) brought about marked increases in shooting accuracy, but when the intervention was removed (second A phase), accuracy fell back to baseline levels, especially for the first two players. When the dependent variable measure (in this case, shooting accuracy) returns to baseline after the removal of the intervention,
Baseline 1 Baseline 2Treatment
A cc
u ra
cy (
% )
70
60
55
50
45
40
35
65
70
60
55
50
45
40
65
30 1 2
Trials
Participant 1
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Baseline 1 Baseline 2Treatment
A cc
u ra
cy (
% )
90
80
75
70
65
85
60 1 2
Trials
Participant 2
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Baseline 1 Baseline 2Treatment
A cc
u ra
cy (
% )
1 2
Trials
Participant 3
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
This is an example of an ABA design study working with three college basketball players to help increase shot accuracy. What is significant about ABA design?
Source: Pates, Maynard, and Westbury (2001)
Figure 5.4: An example of an ABA design
lan66845_05_c05_p131-156.indd 139 4/20/12 2:48 PM
140
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
this pattern provides more confidence that the intervention is responsible for changes in behavior, because its delivery and removal is demonstrated by behavior change. Another example of the effective use of the ABA design is by Xu, Gelfer, Sileo, Filler, and Perkins (2008). In this study, the researchers studied children’s social interactions as the variable of interest. After the baseline phase, a classwide peer tutoring intervention was imple- mented and then removed. The results indicated that the classwide peer tutoring program was indeed effective in improving social interactions.
When you’re doing a study like these two, and the intervention proves to be effective, you may want to end the study on a B phase rather than an A phase. That is precisely what hap- pens with an ABAB design (which is next). But first, is there ever a BAB design? Yes. There may be occasions where an intervention needs to be the first course of action, such as some- one being admitted to the emergency room, or where it has already been started by another mental health professional (Freeman & Mash, 2008). Once any immediate danger has dissi- pated, it may be possible to remove the treatment and see if the presenting problem reoccurs. If it does, treatment can be reapplied (i.e., BAB). In fact, Freeman and Mash (2008) suggested that the BAB design may be superior because it ends on a treatment phase. But there are other SSDs that end on the treatment/intervention phase (B), such as the ABAB design.
ABAB: Repeating Treatments
The ABAB design is a versatile design and is widely applied in numerous areas where behavior change is of central interest. Figure 5.5 presents a graphic of what a typical ABAB design outcome looks like (from Taylor & Adams, 1982).
P e rc
e n t o f T
im e S
p e n t o n -T
a sk
Sessions
Baseline Intervention InterventionBaseline 100
50
0
This is a hypothetical example of what an ABAB or repeating treatments design would look like.
Source: Taylor and Adams (1982)
Figure 5.5: An example of ABAB repeating treatments design
lan66845_05_c05_p131-156.indd 140 4/20/12 2:48 PM
141
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
The first A is the baseline, followed by the B intervention, followed by withdrawal of the intervention (returning to baseline—A), and then followed by the re-administration of the intervention again (B). This repeated administration and withdrawal of the intervention (independent variable) allows for greater confidence that the B phase is causing a change from baseline.
There are good examples of where an ABAB design was successfully employed. For example, Hetzroni and Shrieber (2004) worked with three junior high students with writ- ing disabilities, and they used an ABAB design to look at how word processing on a computer might impact spelling errors, reading errors, and the number of words written per text (the dependent variables). Hetzroni and Shrieber found that by using (and with- drawing) word processing capabilities, during the B phases spelling mistakes and read- ing errors decreased, but the number of words written per text remained unchanged. In a different study using an ABAB design, Finley and Cowley (2005) worked with a 44-year old female to help reduce the amount of time between attempting to go to sleep and falling asleep (that time interval is called sleep onset). By using an intervention (B) that involved going to bed at a regular time each night, and by employing an ABAB design, Finley and Cowley found that the latency of sleep onset decreased substantially. In a third example, Woodard, Groden, Goodwin, and Bodfish (2007) used an ABAB design with autistic children in hopes of treating problem behaviors and core symptoms of autism. The B intervention was dextromethorphan, which is an active ingredient in many cough medications sold over the counter. Testing eight participants, Woodard et al. found that the use of dextromethorphan in an ABAB design was effective for three of them. This type of finding is important because it demonstrates the individual differ- ences that are detectable with an SSD. Had this been a group study, combining the data of all the autistic children may have resulted in failing to reject the null hypothesis. But by using the ABAB design, a more complicated picture emerges: This treatment modality works for some, but not for others.
The ABAB design, as versatile as it is, may not always yield a practical application. For example, when employing an ABAB design in sport psychology, if the first administra- tion of the treatment (B) is effective, a participant in the sports world may be reluctant to remove the treatment and return to baseline (A) (Bryan, 1987). Another concern is when the first treatment (B) involves training on some sort of skill. Say, for example, a discus thrower records baseline throws (A) and then is shown a new spin technique to be used right before release (B). With the success followed by the first B intervention, it may be impossible to “unlearn” or remove the training in an attempt to return to baseline. With the ABAB design ending on an intervention (B) phase, this hopefully facilitates the con- tinuation of any positive impact of the intervention.
These reversal designs demonstrate the basic versatility of single-subject designs, but as you can imagine, there are more complicated variations on this theme. Two of the themes we’ll explore next are the multiple baseline designs and the changing criterion (interac- tion) designs.
lan66845_05_c05_p131-156.indd 141 4/20/12 2:48 PM
142
CHAPTER 5Section 5.1 Reversal/Withdrawal Designs
Case Study: Hey, I’m an N = 1: Changing My Own Behavior
Sometimes in learning about theoretical concepts or behavioral principles we might forget that these principles not only apply to our understanding of others, but they also apply to our understanding of ourselves as well. The behavior change principles described in this chapter that can be used for indi- viduals (N = 1) or small-group (small N) research are principles that we could elect to use to change our own behavior, if we wanted to. By teaching SSD principles in a accelerated course format (6-week class meeting 4 times a week), David Morgan (2009) had his students learn about SSDs by doing an experi- ment—on themselves.
Students in the course were able to select a behav- ior that they would like to change about themselves, a behavior that was relatively easy to observe and record. You can probably imagine some of the behav- iors selected by students—amount of exercise, caf- feine consumption, dietary habits, and so forth. In the example highlighted here (with real data), a student’s targeted behavior was to reduce the amount of snack- ing after dinner, more formally known as post-meal consumption. To establish the baseline (A) phase, the student recorded the number of calories per day consumed after dinner for 14 days—the baseline condition presented in Figure 5.6. The solid horizon- tal line across the chart indicates the mean (average) level of calories, just about 500 calories of post- meal consumption per night. The dotted horizontal lines represent a 2 standard deviation distance away from the mean—both the higher distance and the lower distance. Using an SSD approach, if any subsequent observations (after the baseline) fall outside of the dotted range, that is interpreted as a statistically significant behavior change. As you can see from the data, this student reached that signifi- cant level of change on Day 20, and the change was in the desired direction (fewer calories post-meal), and the trend out to Day 29 provides some evidence that the intervention (B) phase was successful.
E ve
n in
g C
a lo
ri e s
C o n su
m e d
Successive Days
Baseline Intervention
Student Calorie Consumption
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
800
700
600
500
400
300
200
100
0
This graph shows the data from David Morgan’s experiment. What might be significant about the results of this experiment?
Source: Morgan (2009)
Figure 5.6: An applied example of statistical process control
Flirt/SuperStock
(continued)
lan66845_05_c05_p131-156.indd 142 4/20/12 2:48 PM
143
CHAPTER 5Section 5.2 The Multiple Baseline Approach
5.2 The Multiple Baseline Approach
One of the general concerns about SSDs (and about all experimental designs) is the ability to demonstrate internal validity—that is, that the manipulations of the inde-pendent variable caused changes to the dependent variable. In a single-subject design, there are so many factors not under the control of the researcher that it is difficult to have high confidence in a direct causal link, even if the graphical data support that conclu- sion. One methodology available in SSDs is the multiple baseline design. There are at least three different types of multiple baseline designs: (a) multiple baselines across participants, (b) multiple baselines across variables, and (c) multiple baselines across situations (Dattilo & Nelson, 1986; Gliner, Morgan, & Harmon, 2000; Horn & Heerboth, 1982; Robison, Mor- ran, & Hulse-Killacky, 1989). A common example is the multiple baselines across partici- pants. In the hypothetical example from Dattilo and Nelson (1986) in Figure 5.7, all three participants begin in a baseline condition (the A phrase). The intervention (the B phase) is only introduced to one of the participants, while the remaining two stay in the A phase. If
These techniques are used in more than just SSD research. This type of approach, including the dotted horizontal lines indicating a distance of 2 standard deviations, is also used in the manufacturing world for the analysis and interpretation of data—it’s called statistical process control (SPC; Morgan, 2009). SPC procedures use charting techniques such as those used here in manufacturing as a means of measuring quality control, such as the quality of product coming off an assembly line. This approach provides a data analytic strategy that is not so dependent on the subjective judgment of interpreting a graph, yet the inclusion of a graph tells a powerful story about whether the behavior change approach works or not.
Critical Thinking Questions
1. The research projects that Morgan (2009) described are part of an intense course in Research Methods—and the student project itself took 29 days to complete. If you were enrolled in a 5-week course, for example, do you think you could do this type of project? What modifications would you need to make for it to be successful? For instance, a 5-week course spans 35 days— could you make a significant behavior change in a 35-day span of time? Would that behavior change be long-lasting? How would you measure the long-lasting impact of the behavior change?
2. Small-N designs are intended to help change the behavior of one person or a small number of individuals. If you were to embark on a behavior change project intended to change one of your behaviors, what behavior would you select? How would you make that behavior easy to observe and easy to record? Can you think of practical applications of SSD principles in your daily life? Do you have a son or daughter, another loved one, a neighbor, or a coworker who you would like to see change a particular behavior? What behaviors would you target, and how would you make those behaviors observable and recordable—that is, measurable?
3. Think about the applications of statistical process control and quality assurance. If you have heard about TQM (total quality management) before, then you may be familiar with SPC. How might you use these SSD principles to improve product manufacturing? What would be measured, and what would those graphs look like? In the manufacture of automobile axles, for example, if SPC were used to record the number of faulty axles produced per day (that is, errors), two standard deviations away from the mean might be an acceptable error rate. But what if your company were manufacturing artificial heart valves used in heart bypass surgery or the artificial bone and socket used in hip replacement surgery? How might we change the level of acceptable errors in regard to a single subject design/SPC approach?
Case Study: Hey, I’m an N = 1: Changing My Own Behavior (continued)
lan66845_05_c05_p131-156.indd 143 4/20/12 2:48 PM
144
CHAPTER 5Section 5.2 The Multiple Baseline Approach
the intervention is effective, in theory only the first participant should be affected, and the other participants see a change in behavior only after the later, staggered introduction of the intervention/B phase.
Although this example uses different individuals to receive the interventions, there could be different variables studied in each of the three conditions with one individual (Robi- son et al., 1989). For example, if the multiple baseline design is being used to help a child focus on math skills, reading speed, and positive playground behavior, then three base- lines would be established. Then the intervention would be applied to the math skills first (while not trying to affect reading speed or positive playground behavior). If each behavior improves with the onset of the intervention, this outcome lends greater support to the tentative conclusion that the intervention is causing the change in behavior. Simi- larly, there could be three different scenarios or environments where behavior change is desired, such as home, school, and playground. Using the multiple baseline approach, the intervention would be introduced one environment at a time to determine if the interven- tion alone is responsible for any observed behavior change.
B e h a vi
o r
S co
re B
e h a vi
o r
S co
re B
e h a vi
o r
S co
re
Observation Period
Baseline
Subject A
Subject B
Subject C
Intervention
5 10 15 20
30
20
10
0
5 10 15 20
30
20
10
0
5 10 15 20
30
20
10
0
A hypothetical example of a multiple baseline approach from Dattilo and Nelson.
Source: Dattilo and Nelson (1986)
Figure 5.7: An example of a multiple baseline approach
lan66845_05_c05_p131-156.indd 144 4/20/12 2:48 PM
145
CHAPTER 5Section 5.3 The Changing Criterion Design
The multiple baseline design is essentially the AB design mentioned earlier in reversal designs, but with variations on the theme. One advantage of the true AB design is that if an intervention (B) is found to be effective, it is not withdrawn (Gliner et al., 2000). How- ever, the general drawback to the AB design is that it may be unclear if the B intervention alone is responsible for behavior change; the multiple baselines do strengthen that argu- ment, however. Horn and Heerboth (1982) suggested that, if possible, the combination of the multiple baseline design and the ABA design would lead to very strong conclusions, but if the intervention (B phase) is effective, depending on the situation, it may not be ethi- cally prudent to withdraw an intervention or treatment that works.
5.3 The Changing Criterion Design
The multiple baselines design is essentially an AB design, with the twist being that there are multiple baselines for comparison (different participants, settings, or behaviors). The changing criterion design is also an AB design, except in this case there are not multiple baselines, but the B phase is manipulated such that criteria for opti- mal performance are gradually changed over time (Foster, Watson, Meeks, & Young, 2002; Freeman, 2003; Rapoff & Stark, 2008), which is reminiscent of shaping in operant condi- tioning. So, in the example shown in Figure 5.8 from Foster et al. (2002), baseline data are collected in the A phase as usual. The goal of this particular changing criterion design was to improve the amount of classroom homework that a student was completing. With the baseline around 20-30%, it would be a daunting task for the student to leap from 30% completion to 100% completion, so the intervention is introduced gradually, over time, but adjusting the criterion for success.
P e
rc e
n ta
g e
o f
W o
rk C
o m
p le
te d
Days
Changing-Criterion Design
Baseline Criterion 1 Criterion 2 Criterion 3100
90
80
70
60
40
30
20
10
50
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
In this example the goal is to improve the amount of classroom homework that a student is completing. What impact do you think the gradual introduction of the intervention might have had on the data?
Source: Foster et al. (2002)
Figure 5.8: A changing criterion design example
lan66845_05_c05_p131-156.indd 145 4/20/12 2:48 PM
146
CHAPTER 5Section 5.3 The Changing Criterion Design
So as you can see, in the first B (intervention phase), the goal is to get the criterion of class- room homework completed up to 50%; once there and stable, the criterion is increased again. This gradual increase in the criterion over time helps to make the desired goal more attainable.
The changing criterion design can also be used to decrease the frequency of an undesir- able behavior, such as cigarette smoking. In the graphic shown in Figure 5.9, Taylor and Adams (1982) utilized a changing criterion design to gradually decrease the number of cigarettes smoked. As you may be aware, it is difficult for smokers to go “cold turkey” and completely quit and maintain that behavior. The changing criterion design is similar to operant conditioning and shaping, where the goal behavior is gradually acquired over time by changing the targeted amount of the criterion variable.
In the changing criterion design, the internal validity (causality) argument is strengthened by the occurrence of two events—a change from baseline to intervention and a continuing change in the intervention (B phase) as behavior changes (Freeman, 2003).
N u m
b e r
o f C
ig a re
tt e s
S m
o ke
d
Sessions
Baseline Intervention40
30
20
10
0
In this example, a changing criterion design is used to gradually decrease the number of cigarettes smoked. How is the changing criterion design similar to operant conditioning and shaping?
Source: Taylor and Adams (1982)
Figure 5.9: Another example of a changing criterion design
lan66845_05_c05_p131-156.indd 146 4/20/12 2:48 PM
147
CHAPTER 5Section 5.4 Data Analysis and Evaluating Change
5.4 Data Analysis and Evaluating Change
As in any other research approach, the definition and measurement of the depen-dent variables in SSDs is key. Horner et al. (2005) summarized the features needed of dependent variables in single-subject designs: • The dependent variables are operationally defined such that valid and reliable
measurements are possible as well as replication. • The dependent variables are repeatedly measured within the relevant phases (A
baseline phase, B intervention phase) such that stable patterns emerge and are recorded.
• The definition of the dependent variables remains constant over the course of the study, so that there is no measurement drift in the recording of scores.
• The researcher selects variables based on their social significance—that is, the desired change in behavior leads to an improvement of the human condition for the person under study.
The measurement of the dependent variable is then followed by an appropriate level of analysis. A long tradition in the analysis of SSD data is by visual inspection (Ottenbacher, 1990; Tankersly, McGoey, Dalton, Rumrill, & Balan, 2006), such as the graphical displays presented at various points in this chapter. Statistical analyses have emerged that go beyond the visual inspection of the data, and these analyses tend to fall into one of two cat- egories: non-regression methods and regression methods (Campbell, 2004; Jenson, Clark, Kircher, & Kristjansson, 2007; Olive & Smith, 2005). An in-depth presentation of each of
these approaches is not possible here. A typical approach of the regression method is to calculate a regression line for the baseline data, calculate another regression line for the intervention data, and then make a specific comparison between the two lines (Olive & Smith, 2005). Examples of effect size calculations that do not rely on the regression approach include the standard mean difference, per- centage of non-overlapping data, mean baseline reduction, and per- centage of zero data (Campbell, 2004; Olive & Smith, 2005).
Before we leave the topic of ana- lyzing single-subject design data, two additional points are relevant. In addition to statistical vigor in determining the impact of the results from an SSD, “replicability is the final arbiter of whether an effect is likely to occur by chance” (Crosbie, 1999, p. 105). A distinct
Statistical vigor and replicability are important factors in single-subject designs. Why do you think these are particularly important to SSDs?
PR Newswire/Associated Press
lan66845_05_c05_p131-156.indd 147 4/20/12 2:48 PM
148
CHAPTER 5Section 5.5 Applications and Limitations of Single-Subject Designs
advantage of the ABAB design, for example, is the built-in replication of the effect of the treatment and the return to baseline. If confidence in the effectiveness of the interven- tion is desired, repeat the intervention—either within the design or with another similar participant. The validity of the results is established over a series of studies. Finally, the applied researcher using an SSD has to consider clinical significance in addition to statistical significance (Tankersley et al., 2006). That is, helping a person who is trying to reduce the frequency of cigarette smoking, and analyzing those efforts via some of the options described here, may not yield a statistically significant finding (reducing the number of cigarettes smoked per day from 15 to 11), but if the person is helped to some degree, that outcome may have clinical significance to the therapist and the client.
5.5 Applications and Limitations of Single-Subject Designs
In some ways, it’s all about having the right tool for the right job. Single-subject design methodology is extremely effective when used properly in the appropriate situation. In fact, SSDs have been used in many different situations, including school settings, recycling and anti-littering programs in communities, occupational safety, workplace quality and productivity, a variety of clinical disorders, and the treatment of individuals with a developmental disability (Carr & Austin, 1997). There is elegance to the logic of the single-subject design; although sharing some components of a traditional experimental design, the SSD achieves validity and reliability through alternative mechanisms. Many of the examples used throughout this chapter have been of hypothesized data; as you can imagine, actual data from an SSD can be much more complicated. Just to give you a realistic taste, Figure 5.10 presents data from Hume and Odom (2007) where the desired outcome was to enhance work and play systems of three students with autism. To quote from the abstract, “observational data indicated that all students showed increases in on-task behavior, increases in the number of tasks completed or play materials utilized, and reduction of teacher prompts. The results were maintained through the 1-month follow-up” (Hume & Odom, 2007, p. 1166). Not only were the results verified through an ABAB design, but they were also replicated across 3 students, and the benefits of training were still evident after a 1-month follow-up. It is clear that the SSD can be an invaluable tool for detecting change in behavior over time, especially when the focus is on the individual.
lan66845_05_c05_p131-156.indd 148 4/20/12 2:48 PM
149
CHAPTER 5Section 5.5 Applications and Limitations of Single-Subject Designs
P e rc
e n t o f In
te rv
a ls
Mark
Session
Baseline 1 Intervention 1
Training criteria met
On Task Prompted
Intervention 2 MaintenanceBaseline 2
100
90
80
70
60
40
30
20
10
50
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 Month
P e rc
e n t o f In
te rv
a ls
Scott
Session
Baseline 1 Intervention 1
Training criteria met
On Task Prompted
Intervention 2 MaintenanceBaseline 2 100
90
80
70
60
40
30
20
10
50
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 Month
P e rc
e n t o f In
te rv
a ls
Chris
Session
Baseline 1 Intervention 1
Training criteria met
On Task Prompted
Intervention 2 MaintenanceBaseline 2 100
90
80
70
60
40
30
20
10
50
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 1 Month
More examples of real data from SSD studies. Note how it is obvious from visual observation alone that a procedure is effective. This feature makes SSD approaches attractive to a large number of researchers.
Source: Hume and Odom (2004)
5.10: Observational value of single-subject design data
Newman and Wong (2004) summarized some of the chief limitations of SSD methodology. First, in order for researchers to be confident of behavior change, stable baselines must be achieved, and in a clinical setting it may not be prudent to wait for this to happen. Second, as part of a reversal design, a treatment is withdrawn to ascertain its effective- ness. Although this is important for the SSD researcher, this may not be a desired turn of
lan66845_05_c05_p131-156.indd 149 4/20/12 2:48 PM
150
CHAPTER 5Section 5.5 Applications and Limitations of Single-Subject Designs
events for the patient or client who is being helped by the intervention. A third concern rests with the analysis of the data (Newman & Wong, 2004). Although more sophisticated types of analyses are now available (regression- and non-regression-based approaches), the clinician may not be overly motivated (or trained) to pursue these higher-level analy- ses beyond visual inspection of the graphical data. Ultimately, the more these statistical approaches are emphasized during the graduate school training of clinicians and practi- tioners, the more frequently these data analytic techniques will be employed. When clear, the graphical picture tells a compelling story; when the picture is foggy, more sophisti- cated data analysis and interpretation techniques are highly valued.
Classic Studies in Psychology: Helping Hyperactive Children (Ayllon, Layman, & Kandel, 1975)
Since the mid-1970s, amphetamines such as Ritalin have been used to help control hyperactivity in children. Hyperactivity is typically defined as excessive movement, behaving unpredictably, a lack of awareness of the consequences of behavior, the inability to focus, and poor academic performance (Ayllon, Layman, & Kandel, 1975). In the United States in the mid-1970s, about 200,000 children were taking prescription amphetamines for hyperactivity. Using a single- subject multiple baseline design, these researchers were interested in measuring the hyperactivity levels on and off the prescription drug, and the impact that would have on math and reading per- formance. In addition, Ayllon et al. (1975) utilized a reinforcement program during one phase of the design to see if reading and math improvements could be evidenced in hyperactive children while not taking a prescription medication such as Ritalin.
The SSD comprised four phases:
Phase 1: On medication—three students (Crystal, Paul, and Dudley) who were hyperactive and on medication were studied to establish baseline levels of both hyperactivity and academic performance. Even with medication, hyperactivity is not eliminated; each student exhibited measured levels of still behaving in a hyperactive manner about 20% of the time. This phase established the baseline.
Phase 2: With parental and school consent, the prescription medications were discontinued for 3 days to allow the drugs to “wash out” of the students’ systems. As you can see by the graphs below, hyper- activity skyrocketed, with academic performance remaining at its previously low level.
Phase 3: Although the children continued without their medication, a reinforcement program (similar to a token economy) was introduced for math instruction only.
Phase 4: The three children were still off their medication, but in this phase reading instruction was added to the reinforcement program already in use with math instruction.
Study Figure 5.11 and see what happened to student performance over the course of the study. The solid lines connecting filled dots represent hyperactivity. The patterns are roughly the same for all three children. The medication was effective in controlling the hyperactivity, but academic perfor- mance remained low. When the medication was removed, we see the spikes in the solid lines—hyper- activity increased when the drug used to control hyperactivity was removed. But then look what happened when the reinforcement program for math started in Phase 3—hyperactivity dropped, but academic performance improved, and this improvement was sustained in Phase 4 when the reinforce- ment program was extended to reading instruction.
Associated Press
(continued)
lan66845_05_c05_p131-156.indd 150 4/20/12 2:48 PM
151
CHAPTER 5Section 5.5 Applications and Limitations of Single-Subject Designs
So what was so effective about this reinforcement program? The teachers used a token economy. In the classroom, when a child performed a desired behavior, the teacher placed a checkmark on an index card for that child. At the end of the day, students could trade in their index cards with the checkmarks (ranging from 1 checkmark to 75 checkmarks) for items determined to be positive rein- forcers, such as candy, school supplies, free time, eating lunch in the classroom (rather than the caf- eteria), and picnics in the park.
The results of the Ayllon et al. study (1975) have lasting implications (Brown & Borden, 1986; Moeller, Barratt, Dougherty, Schmitz, & Swann, 2001; Pelham, Wheeler, & Chronis, 1998) and are meaningful today. One of the conclusions drawn by Ayllon et al. (1975) was that “the present results suggest that the continued use of Ritalin and possibly other drugs to control hyperactivity may result in compliant but academically incompetent students” (p. 144). It is the ingenuity and cleverness of the researchers using a single-subject design that allows for such meaningful conclusions. In fact, Ayllon et al. (1975) concluded their article with this advice:
On the basis of these findings, it would seem appropriate to recommend that hyperactive children under medication periodically be given the opportunity to be drug-free, to mini- mize drug dependence and to facilitate change through alternative behavioral techniques. While this study focused on behavioral alternatives to Ritalin for the control of hyperactiv- ity, it is possible that another drug or a combination of medication and a behavioral pro- gram may also be helpful (p. 145).
Critical Thinking Questions
1. Thinking about the nature of individual differences observed in this classic study, and then think- ing about a typical grade-school teacher that might have 28 pupils in the classroom,
M a
th (
% )
100
50
25
75
0
R e
a d
in g
( %
)
100
50
25
75
0 1 2
Medication
Days (in Blocks of 3)
No Medication
Reinforcement Math Plus
Reading Math
3 4 5 6
Crystal Hyperactivity
Academic Performance
7 8 9 10 11
M a
th (
% )
100
50
25
75
0
R e
a d
in g
( %
) 100
50
25
75
0 1 2
Medication
Days (in Blocks of 3)
No Medication
Reinforcement Math Plus
Reading Math
3 4 5 6
Paul Hyperactivity
Academic Performance
7 8 9 10 11
M a
th (
% )
100
50
25
75
0
R e
a d
in g
( %
)
100
50
25
75
0 1 2
Medication
Days (in Blocks of 3)
No Medication
Reinforcement Math Plus
Reading Math
3 4 5 6
Dudley Hyperactivity
Academic Performance
7 8 9 10 11
Here are real data from Ayllon et al. (1975). What is the significance of these results? Do you agree with the advice given by Ayllon et al. after their study?
Source: Ayllon, Layman, and Kandel (1975)
Figure 5.11: The classic SSD design from Ayllon, Layman, and Kandel (1975)
Classic Studies in Psychology: Helping Hyperactive Children (Ayllon, Layman, & Kandel, 1975) (continued)
(continued)
lan66845_05_c05_p131-156.indd 151 4/20/12 2:48 PM
152
CHAPTER 5Chapter Summary
The single subject design is effective at helping to understand the behavior of individuals, which has high utility in psychology. But when the attitudes and opinions of large num- bers of individuals is of interest, surveys and questionnaires can also be effective tools for answering particular research questions of interest; this is the topic of the following chapter.
Chapter Summary
The single-subject design approach is so different from, yet so similar to, previ-ous research approaches presented in this book. It is different from much of the traditional research in psychology because it utilizes an idiographic approach to studying single individuals or small numbers of individuals (sometimes called small- N studies). It is similar in that some of the same experimental design principles, such as pretest-posttest designs and time series designs are incorporated into SSD research (but the processes are labeled differently). The ability to make an experimental effect appear and reappear with predictability lends support to cause-and-effect conclusions, which may be extremely important to an individual undergoing treatment but of little generalizability beyond the person studied. However, multiple SSD studies over time can contribute to our cumulative knowledge of effective behavior change. Single-subject designs are quite versatile and are helpful to the practitioner seeing a client or patient, with the goal of affecting change (and being able to graphically represent that effective- ness). Often, with special needs children, the SSD approach is the only type of research approach available for effective use.
how might an understanding of SSDs apply to teaching? Does a one-size-fits-all strategy appear likely to work based on an examination of individual differences? Which concepts emerging here might be helpful to curb undesired behaviors in the grade-school classroom?
2. Imagine you were in the midst of this classic study, and following the A (baseline) phase and the B (intervention) phase you discovered a wildly successful approach for helping children with ADD. What would be the ethical considerations of stopping the study there (AB), or continuing on with an ABA or ABAB design? Why would you want to remove an intervention (the B phase) after you had evidence that the targeted behavior change was working?
3. Think about the dynamics of research in a grade-school classroom. What ethical issues might emerge if a teacher who is informed about SSDs attempts behavior change with kids? Is all teach- ing about behavior change, or should some attempts at behavior change be monitored and approved by school officials and parents? What types of bad habits might children pick up in the course of being trained toward good habits (for example, paying junior high school students $10 for every A on a semester report card)?
Classic Studies in Psychology: Helping Hyperactive Children (Ayllon, Layman, & Kandel, 1975) (continued)
lan66845_05_c05_p131-156.indd 152 4/20/12 2:48 PM
153
CHAPTER 5Concept Check
Research designs from this chapter
Type of Study Design Name
Symbolic Representation
A = observation/ data collection
B = intervention
Brief Features
Single-Subject Withdrawal of Treatments
A B A
With this design, cause-and-effect conclusions are possible when a stable baseline is established. This design is functionally equivalent to a pretest-posttest design, but only with one individual (or just a few individuals, studied separately).
Single-Subject Repeating Treatments
A B A B
The repeating treatments design is similar to ABA, but the intervention is then applied for a second time. This allows for confirmation (replication) that the treatment/ intervention (B) is related to changes in the outcome (A).
Single-Subject Multiple Baseline
N/A
This design involves the AB pairing of events in multiple individuals, but the start of the intervention (B) phase is staggered so that the direct A-B behavior change is visible at different times across multiple individuals.
Single-Subject Changing Criterion
N/A
This design builds on the basic AB sequence of single-subject designs, but the level of B outcomes is expected to change over time. That is, there is a sliding scale of expectations about the intended increases or decreases in targeted behaviors.
Concept Check
1. A case study differs from a single-subject design in that the single-subject design
A. uses only one participant or limited unit. B. randomly selects the participant(s). C. includes the administration of a treatment. D. can be used with observations or interviews.
lan66845_05_c05_p131-156.indd 153 4/20/12 2:48 PM
154
CHAPTER 5Questions for Critical Thinking
2. Ms. Turner has her tennis team practice for a week while she records their per- formance. She then has them practice with feedback from a professional tennis player each day and records their performance for a week. Ms. Turner then has the team return to regular practice while recorded for another week. This single- subject design would be symbolized as
A. BAB. B. OXO. C. XOX. D. ABA.
3. The horizontal dotted lines in graphs usually represent
A. the introduction of the treatment. B. the averages of the data points. C. 2 standard deviations from the mean. D. where the behavior has changed.
4. The dependent variable in a single-subject design should
A. measure between select treatment phases. B. be measured until a “drift” pattern emerges. C. select variables based on their research simplicity. D. use variables that improve the human condition.
5. In the United States in the mid-1970s, about how many children were on pre- scription amphetamines for hyperactivity?
A. 20,000 B. 2000 C. 200 D. 200,000
Answers 1. C. Includes the administration of a treatment. The answer can be found in the Introduction.
2. B. ABA. The answer can be found in Section 5.1.
3. C. Two standard deviations from the mean. The answer can be found in Section 5.1.
4. D. Use variables that improve the human condition. The answer can be found in Section 5.4.
5. D. 200,000. The answer can be found in Section 5.5.
Questions for Critical Thinking
1. How successful would it be to implement a behavior change program following SSD principles on yourself? Ever kept a tally of number of text messages per day, number of diet soft drinks consumed, number of stairs climbed per day, etc.? How might you use the baseline data on yourself to implement an SSD behavior change program on yourself? What approach would be most effective, in your opinion, and why? If you were in the midst of an ABAB design, and after the first intervention (B), you found that you were seeing substantial improvement in
lan66845_05_c05_p131-156.indd 154 4/20/12 2:48 PM
155
CHAPTER 5Key Terms to Remember
behavior areas you selected, would you be willing to return to baseline to exam- ine the potential cause-and-effect relationship? What about the placebo effect?
2. Think about how you might implement an SSD program for your child, a co- worker, or perhaps a neighbor. What behavior would you strive to change? How would you obtain baseline observations on that behavior? How long should the baseline proceed? What types of ethical implications may be in play when observing others prior to the implementation of an SSD research project?
3. Why is it so important to measure change? As a psychology major, will your goal always be to change behavior? What others goals might you pursue? To measure a change in behavior, you have to be able to measure. What types of behaviors would you be interested in changing, either your own or others’ around you? How would you measure that change? Thinking beyond asking questions on a survey (which is the topic of the next chapter), how can we measure behaviors (as opposed to attitudes, opinions, or perceptions)?
Key Terms to Remember
A phase A single-subject design where baseline data are collected concerning the behavior of interest.
ABA design A single-subject design experiment with a baseline phase, fol- lowed by an intervention, and then fol- lowed by return to baseline.
ABAB design A single-subject design experiment where the first A is the base- line, followed by the B intervention, followed by withdrawal of the interven- tion (returning to baseline—A), and then followed by the re-administration of the intervention again (B). This repeated administration and withdrawal of the intervention (independent variable) allows for greater confidence that the B phase is causing a change from baseline.
B phase The introduction of an interven- tion in a single-subject design.
baseline Data that are collected at the beginning of an experiment to determine a starting point in the data collection process.
changing criterion design A study design in which different participants, settings, or behaviors change gradually over time.
intersubject variability The ability of a single-subject design to identify variation in different individuals.
multiple baseline design The study design in which there are multiple base- lines for comparison.
reversal designs A study design involving the application and removal of an inter- vention strategy.
single-subject design A study design in which cause-and-effect conclusions can be approximated with limited generalizability since the data are typically based on one participant, who serves as his or her own control.
stable baseline When a participant’s behavior is consistent previous to par- ticipation in a study to ensure accurate measurement once an intervention is administered.
lan66845_05_c05_p131-156.indd 155 4/20/12 2:48 PM
156
CHAPTER 5Web Resources
Web Resources
Discussion of the five characteristics of single-subject design, providing examples with visual representations of design methodology. http://www.practicalpress.net/updatenov05/SingleSubject.html
Explanation of details of ABAB single-subject design with visual and situational exam- ples. This enables researchers to conceptualize the differences between this research design and other more commonly used formats. http://allpsych.com/researchmethods/ababdesign.html
Single-subject design examples with a focus on ABAB withdrawal design, multiple-base- line design, and alternating treatment design. http://winginstitute.org/Graphs/Mindmap/Single-Subject-Design-Examples/
National Center for Technology Innovation, where students can learn the elements of single-subject research, the importance of single-subject research, and real-life application. http://www.nationaltechcenter.org/index.php/products/at-research-matters/ single-subject-research/
lan66845_05_c05_p131-156.indd 156 4/20/12 2:48 PM