Small-N Designs

nomothetic

compare across individuals

drawback - misses out on individual differences, unique features/experiences

idiographic

look intensely within one individual

allows for more examination of individual responses, experiences

idiographic approaches

narrative case study

systematic case study

single case experimental design

multiple baseline design

changing criteria design

time series design

narrative case study

intensive study of a single person

• includes richly detailed information (qual)

commonly used to establish initial basis for developing new theory and treatments / examining exceptions to establish theory and treatments

ex.

HM- anterograde amnesia

phineas gage

freud’s case studies

narrative case study benefits

unique and valuable information

source of ideas and hypothesis

source for developing therapy techniques

possible to study rare phenomena

inform clinicians about side effects of treatment

narrative case study limitations

alternative explanations often available

relies heavily on anecdotal information

often limited generalizability

non-representative cases are often presented

many threats to validity

systematic case study

intensive study of a single person > retains richly detailed info (qual)

but adds quantitative measures/data, systematic approach to evaluation

provides increased confidence that intervention is cause of change

systematic case study - demonstrating change occurred

use simple focused, standardized measure - symptom measure

use an individualized measure - therapy goals

additional general standardized measure - quality of like measure; general distress

more assessment points - mid-treatment; specific session intervals

use qualitative approach - interviews about therapy outcomes; feasibility

systematic case study - demonstrating change is due to intervention

self-report measures about therapy outcome

evidence of reliable change over time via standardized measures

examine intervention process - therapy notes; measures of therapeutic alliance

qualitative info about intervention processes directly prior to change

within-case correlations linking intervention processes and change

evaluate possible alternative explanations

general principles of single-case designs

similar in general philosophy to repeated measures group designs - goal is still assessing intervention outcomes

but only one person participating - so comparison is between different conditions presented to that single participant

continuous assessment

observations made several times before and during intervention

enables meaningful comparisons

multiple phases

multiple time periods for measurment of each condition

inferences are drawn based on patterns across phases

baseline assessment

necessary to provide info about behavior of interest prior to intervention

must include several observations

single case experimental design

quasi experimental design

low interval validity

• may be other explanations for effect (maturation, history, regression to the mean)

low external validity

A-B (baseline-treatment)

A-B-A (baseline-treatment-baseline)

A-B-A-B (baseline-intervention-base-intervention)

single case experimental design - example

studying intervention to decrease cocaine use in folks completing methadone treatment

• measured initial desire of cocaine

• introduced reinforcement for cocaine-free urine sample (escalating reinforcement: $2.50 for 1st + 2.96 for each additional - max $1950) > withdrew reinforcement and measured desire for cocaine

A-B-A

single case experimental design - limitations

not all treatment-related behaviors may be reversible

ethical concerns associated with the withdrawal of treatment

switching the treatment “on and off” may have undesirable consequences

multiple baseline design

similar to A-B designs - goal is still to assess impact of introducing a treatment

but you select multiple targets / contexts and introduce treatment for each one

• introduce treatment to each target sequentially > measures how treatment impacts different behaviors + provides more confidence treatment is what impacts behavior

A-B-C-D

A- Baseline treatment A

B- Baseline Baseline Treatment B

C- Baseline Baseline Baseline Treatment C

D- Baseline Baseline Baseline Baseline Treatment D

multiple baseline design - limitations

behaviors cannot be highly interrelated

expects an a priori way of determining a treatment response, which is not always the case

assumes the treatment effect is consistent across problems, behavior, situations, or even people

Changing criterion design

experimental control demonstrated through successive replications of change in same target behavior / problem

• take initial baseline measure

• apply treatment over series of trails

• in each successive trial the criterion threshold for defining a treatment response is set a little higher

incremental change in target must be possible

identify “just noticeable difference”

• amount of change is large enough to be observed but small enough to be considered incremental

treatment trials must be long enough for new behavior to emerge and stabilize

variation - measuring across intervention sub-phases

general challenges with single-case designs

instability - for some targets it may be hard to get stable baseline - variability in baseline data makes it hard to know when baseline trend has stabilized

ambiguity - interpretation of treatment efficacy can be ambiguous without statistics

reactivity -behavior may be reactive to measurement, which would make it impossible to obtain a valid measurement

data analysis in single-case designs

goal - identification of effects that are reliable and unlikely due to chance

generally not statistical tests

most common: visual inspection

criteria for visual inspection

change in the average score across phases (mean)

change in the direction of trend line during different phases (slope)

difference between last day of previous phases and the first day of new phase (level)

period of time between onset or termination of a phase and change in scores

(latency of change)

time-series analysis

evaluates statistical significance of change across A-B phases for each participant in clinical replication series

helpful when visual inspection is difficult or impossible and need to analyze change in slope and/or trend

time series analysis - limits

requires sufficient number of data points

lots happening “behind the scenes” that is important to understand

risk of mis-estimation in data model

difficulties and considerations

lack of clear decision-making tools

need for big effects

attrition

participants may remain in study but lose interest / engagement

requires specific pattern at baseline

treatment phase-variations may be difficult to interpret

stability

purpose of single case designs is to show control IV has over DV

this means some variability is expected when initially changing the iV, BUT order/stability should follow or can’t say IV is controlling DV

run conditions until data have stabilized

one strategy is comparing average 1st half of points and average of 2nd half

must be willing to run several sessions before adding or removing IV (especially at baseline)