Lecture 23: single-subect designs

when do you use single-subject designs? (4)

• rare cases/difficult to recruit (ex: HM)

• when each individual has different baseline (ex: someone who smoked for 10 years VS someone who smoked for 5 months)

• when different treatment or conditions must be compared within an individual (ex: medical treatments)

• when internal validity is important so the participant need to be their own control condition (ex: personal preferences)

define “single-subject designs” AKA “single-case designs”

designs that use the results from only one participant to establish a cause-and-effect relationship between the conditions

what do single-case designs need in order to be qualified as experiments? (2)

• manipulation of an IV

• control of extraneous variables to prevent other explanations

how can you control for extraneous variables in single-case designs? (3)

• baseline

• repeated observations

• replication

*in other experimental designs, it’s usually manipulation and control

how do you observe statistical significance in a graph?

you should compare them level (value of Y) and the trend (slope) between A (before the treatment) and B (during the treatment)

why can’t we only use a graph to see if a treatment caused a change in behaviour? (2)

• no control over extraneous variables that change across time

• differences could be caused by chance

what are the possible single-case experimental design? (4)

• AB (simplest)

• ABA (reversal design)

• ABAB (reversal design)

• multiple baseline (ABAC): two baselines at the same time, one starts the treatment earlier

define “phase”

series of observations of the same person under the same condition (A and B are different phases)

define “baseline phase”

series of baseline observation/observations before the treatment

*identified with A

define “treatment phase”

series of treatment observations

*identified with B

define “consistent level” and how we can observe it in a graph

• series of measurement that are approximately the same magnitude

• graph: the points cluster around a horizontal line

define “consistent trend” and how we can observe it in a graph

• differences from one measurement and the next are in the same direction and nearly the same magnitude

• graph: the series of data points are clustered around the slope

define “stability”

degree to which observations show a consistent pattern

true or false: level and trend can only be stable

false: they can be stable or unstable which creates overall stability

what’s the difference between consistent level, consistent trend and stability?

• consistent level: series of measurement with the same magnitude

• consistent trend: differences from one measurement to the next with the same direction and nearly the same magnitude

• stability: degree to which observations show a consistent pattern (overall level and trend)

what should you do if your data is unstable? (3)

• wait until the data stabilizes before continuing measuring

• average a set of two or more observations

• look for patterns in the inconsistency (maybe there’s something you can control)

explain how looking for patterns with inconsistency can help with unstable data

maybe there’s something you can control (ex: time of day causes inconsistency)

explain how averaging your data could help with unstable data

• left: original unstable data

• right: average over 2 days which makes the data more stable

define “phase change” and how you can do it (3)

changing the conditions/manipulating the IV by

• administering a new treatment

• stopping a treatment

• changing a treatment

*”phase change” because this initiates a new phase (ex: go from A to B)

when should you change phases? (3)

→ consider the participant’s response

• participant is improvement without the treatment, don’t implement the treatment

• participant is deteriorating quickly without the treatment, implement treatment right away

• participant produces deterioration, stop treatment immediately

true or false: in a single-case design, you run the experiment and then observe the results to determine if they should receive a treatment

false: you observe as the experiment is running to adjust the treatment

how many observations should you have within a phase to determine a pattern (level, trend, stability)?

at least 3, usually we need 5-6 (depending on the participant)

what are the visual characteristics that help us determine whether there is a meaningful change between phases? (4)

• change in average level

• immediate change in level

• change in trend

• latency of change (how long it takes to generate a pattern)

explain what we can see in these graphs (individually)

• first: clear difference between phases, difference between last point in A and first point in B → behaviour changed when treatment was introduced

• middle: A = no trend/consistent level → B = increasing trend

• bottom: A = increasing trend → B = decreasing trend (reversal)

what’s the difference between these two graphs? what does the dotted line mean?

• dotted line represents variability: instability of the data

• top graph: treatment works because it goes outside of the variance

• bottom graph: larger variance, but treatment doesn’t seem to have an effect

what’s the advantage of an ABA design?

• it allows the scientist to infer that B was the cause of change in A

• without the second A, there are other explanations that could explain the AB change

• graph: baseline, treatment, baseline → did the B really have an effect on A? let’s go back to A to see

what are the limits of the ABA reversal design? (2)

• you don’t know whether future treatments (B) would generate the same outcome

• if the two baselines (A) aren’t the same, you don’t know if B would have the same effect with A1 than with A2

what’s the goal of an ABAB design?

to demonstrate that the treatment consistently causes changes in the participants behaviour (which is a weakness in an ABA design)

true or false: in an ABAB design, the changes from each baseline to treatment should be the same

true

what’s the advantage of an ABAB reversal design?

because you return to baseline after the first treatment, this can confirm a causal relationship

what are the limitations of the ABAB design? (3)

• can’t always fully test the treatment condition

• can’t always return to baseline (ex: if the participant is cured)

• ethical problem if the treatment if effective (not ethical to take away)

true or false: with an ABAB design, you can create some complex variations

true: you can add a treatment or modify the sequence (ex: ABBACCA, ABCABC)

define “multiple-baseline design”

two multiple baselines phase: one treatment phase is imitated for one baseline while the other baseline continues to be observed

what are the possible multiple-baseline designs? (3)

• across subjects: initial baseline phases are the same behaviour for two participants

• across behaviours: initial baseline phases are two separate behaviours for the same participant

• across situation: initial baseline phases are the same behaviour in two different situations

define “component-analysis design”

series of phases in which each phase adds or subtracts one component of a complex treatment to determine how each component contributes to the overall treatment

*it’s considered as a multiple baseline design

explain this graph

• ABAC/multiple baseline design

• both A baseline conditions have similar level

• we see that C is more efficient than B

what’s the logic between multiple-baseline design?

• it’s the same as for a ABAB reversal design: you want to show that there is a clear and immediate change when you go from baseline to treatment

• the difference is that you have two different treatments

what are the characteristics of a single-case design? (6)

• one participant

• DV is measured repeatedly, at least 3 times

• IV is actively manipulated

• baseline must have a predictable pattern (level, trend, stability) to demonstrate if treatment worked

• experimental control is achieved through change in phase (experimenter decides phase change and order)

• visual analysis confirms level, tend and stability

what are the advantages of single-case designs? (3)

• allows researcher to establish cause-and-effect relationship with one participant

• flexibility: researcher is free to modify the treatment if participant doesn’t respond

• no need to standardize treatment across groups, there’s only one participant

what are the disadvantages of single-subject designs? (3)

• threat to external validity: relationship among variables is only for one person

• multiple and continuous observations are required:

  • no statistical controls

  • reliance on graphs to display data

AB, ABA, ABAB or multiple design: evaluate whether using a nicotine patch reduces smoking

AB of ABA: there’s only one type of nicotine patch tested

AB, ABA, ABAB or multiple design: evaluate whether meditation classes reduce social anxiety

ABAB: see if it’s really meditation that causes reduction

AB, ABA, ABAB or multiple design: evaluate whether an omnivore or plant-based diet leads to the same level of muscle mass

multiple baseline (ABAC): normal, omnivore, normal, plant-based