Transcript Single-Subject/Small-n Research and Designs

Single-Subject/Small-n
Research and Designs
Definition
• the hallmark of this style of research is that the
individual’s data are not averaged with
those of other participants
• the intensive study of the individual
• “Idiographic” research
Idiographic Research
• study of the individual to identify what is
unique
• the unique individual can never be described
by the “average” value
• non-statistical, no large sets of numbers
• look for consistency and reliability
• This is in contrast to “nomothetic” research
Nomothetic Research
• study of groups to identify what is typical
• Establish broad generalizations and general
laws that apply to a diverse population
• The advent of nomothetic research was
actually fairly recent
• Most early research in psychology was
idiographic in nature
“Brief” History of Psychology
• Wundt (1879) and the “introspectionists”:
looked at the individual’s “mental experience”
• Ebbinghaus (1885): studied the way in which
associations are formed in memory, used only 1
subject, himself
• Thorndike (1898): search for intelligence in
animals
• Fisher (1920)- work in agriculture led to the
discovery of the field of statistics
• It became generally accepted that the
nomothetic approach (statistics) was necessary
to scientific research
• Problem: nomothetic method takes individual
differences and defines them as error variance
• Skinner(1930s)-experimental analysis of
behavior-a swing back to the study of the
individual
• Looked for consistency of behavior across
individuals
• Allport (1961)- need both idiographic and
nomothetic approach to truly understand
behavior
Two types of single-subject designs:
Descriptive versus Experimental
•
•
•
•
Descriptive
describe the effect on an individual of a
specific treatment or simply describe a
particular individual
Can be a few pages or a book
May compare several case studies within one
piece of research
Can be from any area of psychology
Experimental
Applied Behavior Analysis
• introduce a particular factor (treatment) and
measure its effect on some aspect of the
individual’s behavior
• systematic application of treatment
Case Study Approach:
a descriptive method
There are three methods of gathering data for
a case study:
• Observation- direct observation of the subject
• Interview- written or recorded responses to
questions from an interviewer
• Archival records- stored documents such as
test scores, medical records, academic records
Case Study Example
Ulric Neisser (1981) Cognition
• case study of John Dean’s memory for the
events & conversations surrounding Watergate
• Dean gave testimony in court about events and
meetings prior to Watergate
• Dean seemed to be possess an amazing
memory of events during that time
• As in all court proceedings, Dean’s testimony
was recorded
• Later on, it was revealed that Nixon had
secretly taped these meetings
• Dean did not know about these tapes
• Neisser took these tapes and compared them to
the court documents of Dean’s testimony
What method of data collection did Neisser
use??
Applied Behavior Analysis:
Experimental single-subject design
• The application of principles of operant
conditioning discovered in the lab to real life
problems (B.F. Skinner)
• Each subject’s data is studied separately, not
combined with any other subject’s data
• No means, variance, standard deviations, or
NHST
Steps in a simple AB Behavior
Analysis
• Stage 1: Baseline (A)- observe (measure) how
the behavior is occurring before
intervention/treatment. Rate of behavior
before treatment
• Stage2: Treatment (B)- introduce
treatment/intervention and continue to monitor
behavior, looking for change
Example
• shy child in a preschool environment. Not
interacting with peers or teachers, non-verbal
• want to encourage child to talk
• give this child a gold star to stick on a poster
each time he speaks
• First monitor speech without treatment:
baseline before treatment (A)
• Then begin to give gold stars as treatment (B)
and monitor speech, see if speech increases
A (baseline)…………..............B(treatment)
16
14
number of utterances/hour
12
10
8
6
4
2
0
1
2
3
4
5
6
7
8
Days
9
10
11
12
13
14
• Did treatment (gold stars) increase rate of
vocalizations?
• Could there be other reasons for an increase in
speech?
• Think about potential threats to internal
validity.
• Other designs are attempts to rule out these
threats
ABAB Reversal Design
• monitor baseline (A)
• apply treatment (B)
• withdraw treatment and return to baseline (A)
• reapply treatment (B)
• hence, “ABAB” design
Gold Star Example (again)
Baseline (A).....Treatment (B)....Baseline (A)....Treatment (B)
18
16
number of utterances/hour
14
12
10
8
6
4
2
0
1
2
3
4
5
6
7
8
9
10
11
Days
12
13
14
15
16
17
18
19
20
Potential problems in an ABAB design
• ABAB rules out History and Maturation
confounds
• But you have two potential issues
surrounding the recovery of baseline during
the second “A” phase
1. Failure to recover baseline in second “A”
period: The behavior may not revert to
baseline. Other factors in the environment
may support the behavior change
2. It may be unethical to recover baseline if the
behavior is dangerous to either the subject or
others (self-injury behavior for example)
Multiple Baseline Designs
• argue against History and Maturation threats
• do not require recovery of baseline
Three Multiple baseline designs:
1. Multiple baseline across individuals (or
“subjects”)
2. Multiple baseline across behaviors
3. Multiple baseline across situations
Multiple baseline across individuals
• compare several people (subjects)
• get a baseline for each person (subject)
independently
• introduce treatment at different points in
time for each person (subject)
• This allows an argument against History and
Maturation
Example:
Allison and Ayllon (1980)
• Traditional versus behavioral coaching
• football skills
• Traditional: verbal instructions, some
modeling, and if not correct then yelling,
berating, punishment
• Behavioral: systematic verbal feedback,
positive and negative reinforcement with
verbal reinforcement
For all three multiple baseline designs:
• the change in behavior for each
person/behavior/situation should occur at
the point in time that treatment is
introduced for that person/behavior/situation
• This allows an argument against history or
maturation as a “plausible alternative
explanation”
Multiple baseline across behaviors
• use only one subject
• monitor several behaviors
• Switch from baseline to treatment at different
points in time for each behavior
Example: Social skills training
• Michel Hersen & Alan Bellack, 1976
• patient in a psychiatric hospital, chronic
schizophrenic
• monitored several social behaviors/skills
• eye contact, speech duration, number of
requests
• all different aspects of being socially
appropriate
Treatment
• Four weeks of “social skills training” (treatment)
• Introduced training for each skill separately: first
eye contact, then duration of speech, and then
number of requests
• Showed video of social situation
• Subject responds to situation
• Researcher provides: Feedback on response and
instructions on how to improve response (skills
training) as “treatment”
• Example:
• Working on eye contact
• Subject makes response to video situation
• Researcher gives feedback
• “That was better, but you still looked away at
the end”
Data from one subject
Multiple baseline across situations
• look at one subject
• monitor one behavior
• do so in a variety of settings
Example:Kay, Harchik, and Luiselli (2006)
Problem
• George, a high school student
• autism with developmental delay
• drooling often/constantly, puddles on school
work
• experiencing social rejection by other student
Treatment
Monitored drooling in three situations:
1. classroom (3 h/day)
2. vocational setting (2 h/day)
3. cooking class (1 h/day)
• Differential Reinforcement of Other Behavior
(DRO)
• gave him small edible treat for each 5 min
period with no drool
Achieving a Stable Baseline
• Baseline issues- Applied Behavior Analysis
bases the argument for an effective treatment
by looking for a change in behavior from some
baseline level
• what if you cannot establish a “stable”
baseline?
Problem Baselines
Unstable Baseline
50
40
30
20
10
0
1
2
3
4
5
6
7
8
Ascending baseline
9
10
Descending baseline
30
25
25
20
20
15
15
10
10
5
5
0
0
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
• All of these are problematic. How can you
achieve a stable baseline?
• Can look for the cause of the instability and
control it.
• Can try to “wait it out” and it may become
more stable over time
• Can average it out