10.0 Systematic Reviews for Single Subject Designs

1
10.0Systematic Reviews for Single Subject
Designs
2
Source of Effect Size For SSD

• 3 Classes Typically Reported
• Standard Mean Difference
• Percentage of Non-Overlapping Data (PND)
• Regression

3
Baseline (A)
Intervention (B)
4
Simulation Dataset
5
(No Transcript)
6
Sources of Effect Size for SSD

• Standard Mean Difference
• Subtract the mean of the baseline phase from the
  mean of the intervention phase
• Divided by standard deviation of baseline

Xtx- Xb d sdb
7
Sources of Effect Size for SSD
Xtx- Xb d sdb
6.71-2.12 d
6.20 .74
8
SMDall 7.23
SMD3 8.25
9
d 6.20
10
Interpretation for SMD

• A 6 standard deviation advantage for the treated
  participants
• .8 standard deviationslarge effect
• The SMD for SSD is not the same metric as the SMD
  for group comparisons

11
Sources of Effect Size For SSD

• 2. Percentage of Non-Overlapping Data (PND)
• Identify the highest baseline point
• Count the number of intervention points that
  exceed the highest baseline point
  (non-overlapping).
• Calculate the proportion of non-overlapping to
  total number of intervention points (cant use if
  baseline has a zero point)

12
PND 7/10 70
13
Interpretation for PND

• PND scale
• 90 Highly Effective
• 70-90 Moderately Effective
• 50-70 Minimally Effective
• gt50 Ineffective

14
Sources of Effect Size SSD

• Mean Baseline Reduction (MBR)
• Subtract the average of last 3 Tx points from the
  average of the last 3 baseline points
• Divide by the average of the last 3 baseline
  points
• Multiply by 100 for percent of baseline reduction
  (MBR)

15
Simulation Dataset
16
1.77 - 6.84 MBR
x 100 286.0
1.77 (mean percent reduction)
17
Interpretation for MBR

• Relative effect 200 is greater than 100
• ???? There is no standard for magnitude of
  effect.