Scribd
Upload
29 views

Control in Experimental Designs

Control groups are important in experimental designs to establish a baseline for comparison and reduce the effects of confounding variables. Strategies to establish control include matching subjects, random assignment, blinding of observers and subjects, placebos, using subjects as their own controls, and analyzing covariates. For example, when studying the effects of a stretching technique on hamstring length, local heat and acupuncture administered simultaneously would be confounding variables that require control, such as through matching or a control group.

Uploaded by

jasvindersinghsaggu
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
29 views

Control in Experimental Designs

Control groups are important in experimental designs to establish a baseline for comparison and reduce the effects of confounding variables. Strategies to establish control include matching subjects, random assignment, blinding of observers and subjects, placebos, using subjects as their own controls, and analyzing covariates. For example, when studying the effects of a stretching technique on hamstring length, local heat and acupuncture administered simultaneously would be confounding variables that require control, such as through matching or a control group.

Uploaded by

jasvindersinghsaggu
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 10

Control in Experimental Designs

Control
Key element of experimental and quasi-
experimental designs
Subjects (one group or several) on which no
variable (e.g., treatment) is applied
Hold a variable or condition constant
Establishes baseline to compare changes in
treatment groups
Attempts to reduce the effect of confounding
independent variables
Example - Confounding Variable
Does the contract-relax technique lead to an
increase in hamstring length?
Simultaneous treatments
Contract-relax
Local heat to hamstrings
Acupuncture
Local heat and acupressure would be
confounding variables
There is a need to control for these variables
Strategies for Gaining Control
Matching - narrow selection criteria (e.g.,
age, gender, similarities in a particular
condition) will limit potential for error when
comparing control vs. experimental groups
Sampling method - a key factor in
establishing control
Random assignment eliminates bias - makes
groups as comparable as possible
Strategies for Gaining Control
Double blind Blinding
Observer has no Observer and/or subjects
knowledge of subject knowledge of treatment
group may bias outcomes
Subject has no knowledge Blinding hides:
of placebo vs. treatment Observers knowledge of
subject assignment
Single blind (control vs. experimental)
Subject has no knowledge Subjects knowledge of
of treatment vs. placebo treatment (placebo vs.
experimental)
Strategies for Gaining Control
Placebo tends to make subjects feel they are
receiving a treatment or intervention being
studied thereby establishing control by
eliminating potential bias
Caution: Deception by using a placebo may be
challenged by the IRB
Strategies for Gaining Control
Subjects as their own control
Subjects are matched to themselves
Exposed to all levels of independent variables
Control treatment or condition
Experimental treatment of condition
Inherently a repeated measures design
Strategies for Gaining Control
Analysis of covariance (ANCOVA)
A traditional way to statistically gain control
Partitions extraneous confounding variables
Treated as a covariate
Controls for initial differences between groups
Effect is to adjust scores on the dependent variable
for pretest differences between groups
Statistically establishes equivalence
Example - Covariate
Mean pretest measures Measure maximal
between groups are hamstring length using a
likely to be different sit-and-reach test in a
Older subjects soft
community-sponsored
tissues often become fitness program before
less elastic with age and after a 10 weeks
exercise program
Treat subject age as a Groups
covariate
Children ages 8 - 16
Adults ages 30 - 50

You might also like