Experimental Design

1
Experimental Design

• Suzanne L. Groah, MD, MSPH
• NRH RRTC on Secondary Conditions After SCI

2
Non-Experimental Design

• Historical, developmental, qualitative
• Weakest type of study in respect to establishing
  a cause-effect relationship
• One of the most common forms of research
• Strong design for descriptive research questions

3
Developmental Research

• Longitudinal studies
• Assess changes over extended period of time
• Craig Hospital RRTC on Aging
• Cross sectional studies
• Assess different demographic characteristics at
  the same time
• RRTC R1 - Cardiovascular disease risk
  stratification across injury levels after SCI
  Assessment of need for intervention and its
  predictors
• Cross sequential studies
• Combine longitudinal and cross sectional methods

4
True Experiment

• Important because it has the power to be the most
  rigid type of research
• Dependent on feasibility
• Goals
• Control for as many confounders as possible in
  order to reduce errors in assumptions
• Minimize threats to internal and external validity

5
True Experiment

• Strengths
• Most rigorous of all research designs
• Gold standard
• Strongest internal validity
• Critical to cause and effect relationship

• Weaknesses
• External validity limited
• Human subjects (confound any study)
• Time restraints
• Situational limitations
• Intrusive
• Attrition

6
Types of Experimental Designs

• Pre-experimental
• Quasi-experiment
• True experiment

7
Pre-Experimental

• Experimental design without the control group
• Single group without comparison to non-treatment
  group
• Case Series
• One group pretest posttest study
• Static group comparison study

X O1
O1 X O2
X O2 O2
8
Quasi-Experimental

• Control group allows for comparisons
• Lack of randomization
• Pretest Posttest Nonequivalent group
• Time Series Designs

O1 X O2 O1 X O2
O1 O1 X O2 O2
9
Experimental

• Only research method that can adequately measure
  cause and effect relationship
• Utilize control group
• To measure change that occurs in both groups
• Attempts to control for confounding

10
True Experiment

• Randomized Posttest Study
• Randomization and control/experimental groups
• Weakness is lack of pretest
• Pretest Posttest Randomized Study
• Most effective at demonstrating cause-effect
  relationship
• Most difficult to perform

11
Experimental Design

• Randomized Posttest Study
• R X O2
• R O2
• Pretest Posttest Randomized Study
• R O1 X O2
• R O1 O2

12
NRH RRTC R2
R O1 X O2 O3 R O1 O2 O3

• Randomization - People with acute SCI randomly
  assigned to control or intervention group
• Pretest - Osteoporosis and bone turnover are
  measured initially using DEXA, CT, blood and
  urine tests
• Intervention routine inpatient rehabilitation
  FES to the lower extremities
• Posttest DEXA, CT, blood and urine markers
  checked immediately post-intervention and 3
  months later

13
NRH RRTC R3
R O1 X O2 O3 R O1 O2 O3

• Randomization - People with subacute SCI randomly
  assigned to control or intervention group
• Pretest - Osteoporosis and bone turnover are
  measured initially using DEXA, blood and urine
  tests
• Intervention routine rehabilitation
  ambulatory training
• Posttest DEXA, blood and urine markers checked
  immediately post-intervention and 3 months later

14
Factorial Design

• Focus is almost entirely on the components/setup
  of the treatment
• Typically examine several different variations of
  a treatment
• NRH RRTC R4
• Study of exercise program(s), examining location
  of exercise and the effect of peer mentoring

15
NRH RRTC Project R4 Design
Influence of Peers
Setting
16
NRH RRTC R4 Factorial Design

• R X1 1 O R XPeer Mentor/Home O
• R X1 2 O R XPeer Mentor/Gym O
• R X2 1 O R XNo Peer/Home O
• R X2 2 O R XNo Peer/Gym O
• R O R O

17
RRTC R4 Mixed Methods

• O1 R O2 XPeer Mentor/Home O3 O4
• O1 R O2 XPeer Mentor/Gym O3 O4
• O1 R O2 XNo Peer/Home O3 O4
• O1 R O2 XNo Peer/Gym O3 O4
• O1 R O3 O4

6 months
1 year
6 months
Longitudinal component
18
More Mixed Methods!