2. PRE-EXPERIMENTAL CLASS
1. One-shot Case Study
2. One-group Pre-test Post-test Design
3. Intact Groups Design
QUASI-EXPERIMENTAL CLASS
1. Non-equivalent Control (or Comparison) Groups
Design
2. Time Series Design
3. Equivalent Time Samples Design
MAJOR DESIGNS
AND CLASSES OF
RESEARCH DESIGNS
IN EXPERIMENTAL
RESEARCH
EX POST FACTO CLASS
TRUE EXPERIMENTAL CLASS
1. Criterion Group Design*
1. Post-test Only Control Group Design
2. Correlation Design
2. Pre-test Post-test Control Group Design*
(after Shavelson, 1981)
3. Design
Does the design
involve more
than one group
of subjects?
Does the design
involve
administering a
treatment?
Does the design
involve a randomly
selected, randomly
assigned control
group to compare
with the
experimental
group(s)?
Does the design
involve some
other kind of
comparison
group?
Is a pre-test
given?
Is random
selection used to
constitute the
sample?
Is random
assignment used
to constitute the
groups?
One-shot Case
Study
No
Yes
No
No
No
No
No
One-group Pretest Post-test
Design
No
Yes
No
No
Yes
No
No
Intact Groups
Design
Yes
Yes
No
Yes
No
No
No
Non-equivalent
Control (or
Comparison)
Groups Design
Yes
Yes
No
Yes
Yes
No
No
Time Series
Design
No
Yes
No
No
Yes
No
No
Equivalent Time
Samples Design
No
Yes
No
No
Yes
No
No
Criterion Group
Design*
Yes
No
No
Yes
No
Yes
No
Correlation
Design
No
No
No
No
No
Yes
No
Post-test Only
Control Group
Design
Yes
Yes
Yes
No
No
Yes
Yes
Pre-test Post-test
Control Group
Design*
Yes
Yes
Yes
No
Yes
Yes
Yes
6. SAMPLE
- a portion of a population, serving as a basis for
estimates of the attribute of the whole population
POPULATION
- a finite or infinite collection of items under
consideration
7. Descriptive
Statistics
provides methods concerned
with summarizing and
describing numerical data.
Uses tables, graphs, and
charts that display data.
Mean
Standard Deviation
Frequency Polygon
Inferential
Statistics
concerns with generalizing
information of making an
inference about a population
based only on a sample or
portion observed
9. difficulty of controlling all the possible confounding variables associated
with research on human subjects
Classroom research “entails a very large human and institutional factors that can
affect research design and outcomes in an unforeseen and unforeseeable ways”.
(Rounds and Schachter, 1996)
more philosophical problem
the issue of subjectivity and objectivity
10. a well-documented, highly codified approach to conducting a research with
well-developed quality control procedures
an internationally recognized method of conducting a research
there are clear criteria for interpreting the outcomes
possibility of making comparisons across studies conducted at different sites
may enable researchers obtain grant money or get their reports published
12. The experimental method has often been used in
language classroom research, but with varied success,
since it is so difficult to control all the possible
confounding variables that can arise in the research
with real people. It is, however, a valuable approach to
understanding teaching and learning, and it has
influenced many other approaches.