2. The Purpose
Of Educational
Research
To provide evidence
To help us develop
to help people decide
better ways to think
which opinions are
about the field of
correct –at least
education.
more correct.
2
3. Exercise: Indicate, on a scale of 1-5, the extent to which you
think research has demonstrated the truth of each statement.
research
Clearly refuted by
research
refuted by
Somewhat
inconclusive
research
refuted by
Somewhat
by research
Clearly supported
1 2 3 4 5 1. The more time beginning readers spent on
phonics, the better readers they become. Answer
1 2 3 4 5 2. Bilingual education for nonnative speakers
impairs their academic proficiency.
Answer
1 2 3 4 5 3. Increased contact with handicapped people
results in a more positive attitude toward them.
Answer
1 2 3 4 5 4. Boys are better in math; girls are better in
languages.
Answer
3
4. Exercise: Indicate, on a scale of 1-5, the extent to which
you think research has demonstrated the
truth of each statement. (Continued)
research
Clearly refuted by
by research
Somewhat refuted
inconclusive
by research
Somewhat refuted
by research
Clearly supported
1 2 3 4 5 5. Requiring students who do not like one
another to work together on a project results in
an increase in their liking for each other. Answer
1 2 3 4 5 6. Students who take moral education courses
behave more ethically than students who do not
take such courses. Answer
1 2 3 4 5 7. The use of manipulatives in the elementary
grades results in improved achievement in
mathematics. Answer
1 2 3 4 5 8. Behavior modification is an effective way of
teaching skills to very slow learners.
Answer
4
5. Exercise: Indicate, on a scale of 1-5, the extent to which
you think research has demonstrated the
truth of each statement. (Continued)
research
Clearly refuted by
by research
Somewhat refuted
inconclusive
by research
Somewhat refuted
by research
Clearly supported
1 2 3 4 5 9. Classroom discussion of real-life sexual
issues and problems results in increased
promiscuity among teenagers. Answer
1 2 3 4 5 10. Among children who become deaf before
languages has developed, those with hearing
parents become better readers than those with
deaf parents. Answer
1 2 3 4 5 11. The more teachers know about a specific
subject matter, the better they teach it.
Answer
5
7. Statement 1 is rated at 3.
Despite a great deal of research on the topic, this
statement can be neither clearly supported nor
refuted. It is clear that phonics instruction is an
important ingredient; what is not clear how much
time should be devoted to it.
7
8. Statement 2 is rated at 2.
Evidence is unclear as to whether or not
bilingual methods are superior to English-
only instruction, but several studies
indicate no impairment of academic skills.
8
9. Statement 3 is rated at 2.
Evidence indicate that a more positive
attitude results only if the nature of
the contact is structured beforehand.
9
10. Statement 4 is rated at 3.
There is a considerable amount of
evidence that these gender differences
exist, though the reasons are not
clear.
10
11. Statement 5 is rated at 3.
The evidence here is quite clear that
the outcome depends on whether the
students involved see one another as
necessary to achieving success.
11
12. Statement 6 is rated at 3.
There is relatively little
research on ethical
behavior.
12
13. Statement 7 is rated at 4.
The evidence is quite
supportive of this method of
teaching mathematics.
13
14. Statement 8 is rated at 5.
There is a great deal of
evidence to support the
statement
14
15. Statement 9 is rated at 3.
Not much evidence exists
and the evidence that does
not exist is inconclusive.
15
16. Statement 10 is rated at
1.
The findings of many studies
refute the statement.
16
17. Statement 11 is rated at
3.
The evidence is inconclusive despite
the seemingly obvious fact that the
teacher must know more than the
students.
17
19. Basic Research vs. Applied Research
Basic Applied
Results apply to a great Results are applicable only
many people and situations. to a specific group of people
in a particular situation.
Result are related to Results are not necessary
general theory or to a related to a broader field of
general field of knowledge. knowledge .
Results need not have Results must have
immediate or even clear immediate and clear
implications for practice. implications for practice.
19
21. Examples of Research Questions
(with an appropriate methodology)
Does client-centered therapy produce more satisfaction in clients
than does traditional therapy? (experimental research)
Are the descriptions of people in social studies in textbooks biased?
(content-analysis research)
What goes on in an elementary school classroom during an
average week? (ethnographic research)
Do teachers behave differently toward students of different
genders? (causal-comparative research)
How can we predict which students might have trouble learning
certain kinds of subject matter? (correlational research)
How do parents feel about the school counseling program? (survey
research)
How can a principal improve faculty morale? (interview research)
21
23. Question 1 and 2 do not suggest a
relationship.
Question 1 asks for no more than a description regarding the
current usage of manipulative materials in a particular school
district. Similarly, question 2 asks only for a survey of
administrative opinions. Investigations of such questions may
be extremely useful in their own right, but they do not extend
our knowledge as to why such conditions exist.
23
24. Question 1 and 2 indicate a
relationship.
Question 3 seeks to investigate a possible relationship between eating
disorders and sexual abuse. If a history of sexual abuse is related to eating
disorders, this suggests (although it does not prove) that such abuse may be a
cause of such disorders. It also suggests that counseling which addresses
patient history may be helpful. Question 4 seeks to investigate a possible
relationship between the type of language instruction and fluency in the
language taught. If the language laboratory method is shown to be more
effective than classroom instruction by individual teachers, this has clear
implications for improving language learning.
24
25. RELATIONSHIP and VARIABLE
A variable is any
characteristic that is not
always the same—that
is, any characteristic
that varies.
Examples of variables include
gender, eye
color, achievement, motivation, and
running speed.
25
26. Exercise: What are the variables in
this research question?
Answer: the variables are age and level of
anxiety in mathematics courses.
26
27. Quantitative vs. Qualitative
Variables
Measured/ Quantitative Variables
• ~exist in some degree rather than all or none
• are measured along a continuum from ―less‖ to ―more‖
• assigned numbers to different individuals or objects
• An example would be height.
Categorical/ Qualitative Variables
• ~not vary in degree, amount, or quantity, but are
qualitatively different
• e.g. eye color, gender, religious
preference, occupation, position on a baseball
team, and most kinds of ―treatments‖ or ―methods‖
27
28. Independent vs. Dependent
Variables
Independent Variables
• are those the investigator chooses to study (and
often manipulate) in order to assess their
possible effect(s) on one or more other variables
• are those the investigator chooses to study (and
often manipulate) in order to assess their
possible effect(s) on one or more other variables
Dependent Variable
• is the variable which the independent variable is
presumed to affect
• All outcome variables are dependent variables. 28
29. Exercise: What are the independent and
the dependent variables in this question?
29
31. Extraneous Variables and Constants
Extraneous Variables are
independent variables that
have not been controlled
Constants are potential
variables that are not
allowed to change
31
32. Ethics and Research
Every researcher should consider:
the protection of participants from
harms
the ensuring of confidentiality of
research data
the knowing deception of research
subjects
32
33. Hypotheses
A hypothesis is, simply put, a prediction of
some sort regarding the possible outcomes of
a study.
A research question is often restated as a
hypothesis.
• Question: ―Do individuals who see themselves as
socially attractive want their romantic partners also to
be socially attractive?‖
• Hypothesis: ―Individuals who see themselves as socially
attractive will want their romantic partners also to be
(as judged by others) socially attractive.‖
33
34. Directional vs. Nondirectional
Hypotheses
A Directional Hypothesis is one that indicates
the specific direction
(e.g., higher, lower, more, less) that a
researcher expects to emerge in a
relationship.
Nondirectional Hypothesis does not make a
specific prediction about what direction the
outcome of a study will take.
34
35. Reviewing the Literature:
General References: the
sources a researcher refers to
first.
Secondary Sources: publications
in which authors describe the
work of others.
Primary Sources: publications in
which investigators report the
results of their studies.
35
36. Steps Involved in a Literature Review
• Define the research problem as precisely as possible.
1
• Skim through some relevant secondary sources.
2
• Peruse one or two general reference works.
3
•Formulate search terms (key words or phrases) that are
4 pertinent to your research question.
• Search the general references for relevant primary sources.
5
• Read the relevant primary sources.
6
• Take notes and summarize the key points in the sources.
7
36
37. A Computer Search of the Literature
Define the problem as precisely as
possible.
Decide on the extent of the search.
Decide on the Database. (e.g.,ERIC)
Select descriptors.
Conduct the search.
Broaden or narrow the search.
Obtain a printout of the desired
references.
37
38. Writing Your Summaries
1. Try to locate at least five recent primary
sources that are pertinent to your topic.
At least three of these be should be
research reports that present data of
some kind (scores on a test, responses
to a questionnaire, and so on). The other
two may be the viewpoint or ideas of
someone as expressed in an article (that
is, merely an opinion piece that does not
present data).
38
39. Writing Your Summaries
2. Limit your summary to approximately
one-half page (200 words).
3. Be sure to describe what the author did
and what the author’s conclusions were.
4. If the reference you are summarizing
pertains to a research study, you should
briefly describe the method of the
researcher used. Be sure that you also
note how the author arrived at his/her
conclusions.
39
40. An Example of a Summary
Walberg, H. J., and Thomas, S. C. 1972. An operational definition and
validation in Great Britain and the United States . American educational
research journal, 9:197-216.
The purpose of this article is to describe the development of an
observation scale and a teacher questionnaire for assessing the degree
of “openness” of a given elementary school classroom. Items were
written within each of eight “themes” obtained from available literature
and reviewed by a panel of authorities.
The resulting instruments were used in approximately 20 classrooms for
each of three types: British open, American open, and American
traditional. The classrooms were identified by reputation and personal
knowledge. Approximately equal numbers of lower and middle
socioeconomic-level classrooms were included.
Results showed that overall assessments obtained with the two
different instruments (observation scale and questionnaire) agreed
quite highly. Differences between the open and traditional classrooms
were much greater than those between socioeconomic levels or
between countries. 40
42. Examples of populations
All of the high school principals in the
United States.
All of the elementary school counselors in
the state of California.
All of the students attending Central High
School in Omaha, Nebraska, during the
academic year 1987-1988.
All of the students in Mrs. Browns’ third-
grade class at Wharton Elementary
School.
42
43. Examples of samples
A researcher is interested in studying the effects of diet
on the attention span of third-grade students in a large
city. There are 1500 third graders attending the
elementary schools in the city. The researcher selects
150 of these third graders, 30 each in five different
schools, to study.
The principal of an elementary school district wants to
investigate the effectiveness of a new U.S. history
textbook being used by some of teachers in her
district. Out of a total 22 teachers who are using the
text, she selects 6, comparing the achievement of
students in the classes of these 6 teachers with those
of another 6 teachers who are not using the text.
43
44. Simple Random Sampling
Stratified Random Sampling
Probability
Sampling Random Cluster Sampling
Two Stage Random sampling
Sampling
Procedures
Convenience Sampling
Nonprobability
Sampling Purposive Sampling
Systematic Sampling
44
45. Simple Random Sampling (SRS)
In SRS every member of the population
has an equal and independent chance of
being selected for the sample.
Example:" We interviewed a sample of 41
mothers of eight graders from one middle
school. These mothers were randomly
selected from a list of 129 mothers
provided by the principal of the school.‖
(Baker and Stevenson, 1986, p.157).
45
46. Simple Random B G
A
E C F H
I
Q D
J
Population L
O
R K
P M
S N
V Z U
W T
X Y
D Y
Sample
N
P
L
H 46
47. Stratified Random Sampling
Stratified sampling is a process whereby
certain subgroups, or strata, are selected
for the sample in the same proportion as
they exist in the population.
Example: ‖From a pool of all children who
returned a parental permission form
(more than 80% return rate) 24 first
graders (10 girls, 14 boys; mean age, 6
years, 6 months), and 24 third graders
(13 girls, 11 boys; mean age, 8 years, 8
months) were randomly selected.‖ (Clements
and Nastasi, 1988, p.93)
47
48. Stratified Random
ABCDE
25%
Population FGHIJ
KLMNO
50%
PQRST
25%
B D
25%
Sample FMOJ
50%
PS
25%
48
49. Random Cluster Sampling
When it is not possible to select a
sample of individuals from a
population--for example, a list of all
members of the population of
interest is not available—cluster
sampling is used. It involves the
random selection of naturally
occurring groups or areas and then
the selection of individual elements
from the chosen groups or areas.
49
50. Cluster Random
AB CD
QR NOP
Population LM
EFG JK
STU
HI
QR CD
Sample
EFG
50
51. Two-Stage Random Sampling
It is often useful to combine cluster
sampling with individual sampling.
Rather than randomly selecting 200
students from a population of 3000
ninth graders located in 100
classes, the researcher might
decide to select 25 classes
randomly from the population of
100 classes and then randomly
select 8 students from each class.
51
52. AB CD
Two-Stage Random
QR NOP LM
EFG JK
Population HI
STU
CD LM
Sample of clusters
STU
Sample of
individuals
Sample C,L,T
52
53. Convenience Sampling
A convenience sample is a group of
individuals who (conveniently) are
available for study.
Example:" A high school counselor
interviews all of the students who
come to her for counseling about
their carrier plans.‖
53
54. Convenience B G
A F
E C H
K
D O
P J
Population R S Z
N
V X M
Q L
I U
Y
T
W
Easily Accessible
Sample Q Y
X L
I
54
55. Purposive Sampling
In purposive sampling the researcher
selects particular elements from the
population that will be representative or
informative about the topic.
Purposive sampling is different from
convenience sampling in that the
researcher does not simply study whoever
is available, but uses his or her judgment
to select the sample for a specific
purpose.
55
56. Purposive B G
A
E C F H
I
Q D
J
Population L
O
R K
P M
S N
V Z U
W T
X Y
B F
Sample N
V
L
56
57. Example of Purposive Sampling
―Introductory psychology students
(N=210) volunteered to take the
Dogmatism Scale (Form E) for
experimental credit. From the upper
and lower quartiles on the
Dogmatism Scale, 44 high and 44
low dogmatic subjects were
selected for the experiment.‖ (Rickards
and Slife, 1987, pp.636-637)
57
58. Systematic Sampling
In systematic sampling every nth
element is selected from a list of all
elements in the population.
58
59. Systematic A B C D E
F G H I J
Population K L M N O
P Q R S T
B G L
Sample Q
59
60. Measurement
Measures are specific techniques or
instruments used for measurements and
generally refer to quantitative devices.
These are often tests and questionnaires
that provide objective and quantifiable
data.
Measurement is an essential component
of quantitative research because it
provides a standard format for recording
observations, performance, or other
responses of subjects and because it
allows a quantitative summary of the
results from many subjects.
60
61. The Purpose of Measurement
~To provide information about the variables
that are being studied.
In an experiment, the dependent variable
is measured.
In correlational research each variable is
measured.
In practice, the variable is defined by how
it is measured (operational definition),
not by how it is labeled or defined by the
researcher.
61
62. Instrument vs. Instrumentation
An instrument is a device or procedure for
systematically collecting information.
Common types of instruments include
tests, questionnaires, rating
scales, checklists, and observation forms.
Instrumentation refers not only to the
instrument itself but also to the conditions
under which it is used, when it is to be
used, and by whom it is to be used.
62
63. • Validity refers to the extent to
which an instrument gives us the
information we want.
validity
• Validity is a judgment of the
appropriateness of a measure for
the specific inferences or decisions
that result from the scores
validity generated by the measure.
63
64. Types of Evidence for Judging Validity
• refers to the nature of the content included within
Content-
the instrument, and the specifications the researcher
related used to formulate the content
evidence
• refers to the relationship between scores obtained
using the instrument and scores obtained using one
Criterion- or more other instruments or measures (often called
related
evidence
criteria)
• refers to the nature of psychological construct or
Construct characteristic being measured by the instrument
-related
evidence
64
66. Validity and Reliability Coefficients
• expresses the relationship which
exists between scores of the same
individuals on two different
A validity instruments
coefficient
• expresses a relationship between
scores of the same individuals on the
same instrument at two different
A reliability times, or between two forms of the
coefficient same instrument
66
67. Methods of
Estimating Reliability
Require two Administrations Require One Administration
The Test-Retest Method Internal Consistency Methods
The Equivalent Forms Method
Split-Half Testing
The Kuder-Richardson Approaches
KR20 KR21
67
68. RESEARCH DESIGN
Nonexperimental Research Experimental Research
Weak Experimental Designs:
• The One-Shot Case Study Design
Descriptive Studies • The One-Group Pretest-Posttest Design
• The Static-Group Comparison Design
Relationship Studies True Experimental Design
•The Randomized Posttest-Only Control
e.g. Simple Correlational Group Design
•The Randomized Pretest-Posttest Control
Studies, and Prediction Studies Group Design
•The Randomized Solomon Four-Group Design
Causal-Comparative Studies Quasi-Experimental Design
• The Matching Only Posttest-Only Control
Group Design
• The Matching Only Pretest-Posttest Control
Group Design 68
True Experimental Designs in Suter (1998)
69. Common Statistical Tests
The t Test To compare two means
The F Test
(ANOVA) To test two or more means
Test for r To test the significance of a
correlation coefficient
Chi-square To test for relationships
Test involving frequency data in
the form of tallies or
percentages 69
70. Descriptive Studies
A descriptive study simply describes a
phenomenon.
Example: ―Their initials attributions were
primarily task attributions (46% to 58% said
the words were easy). Their own effort was
the next most common cause of their
success (40% of the responses). When
asked for a second response, the subjects
evenly divided their answers among the four
types of attributions.‖ (Cauley and Murray, 1982, p.476)
Back to
research 70
designs
71. Criteria for Evaluating Descriptive Studies
1. Conclusions about the relationships and
causal relationships should not be made.
2. Subjects and instrumentation should be
well described.
3. Graphic presentations should not distort the
results.
(McMillan, 1992: 146)
71
72. Relationship Studies
Relationship studies investigate the degree to
which variations or differences in one variable are
related to variations or differences in another
variable.
Examples:
1. Correlational Studies indicate relationships by
obtaining two scores from each subject.
2. A predictive study shows how one variable can
predict what the value will be on a second variable
at a later time.
Back to
research 72
designs
73. Example: Relationship Study
of Differences Among Groups
―Advanced level students were more
internally responsible for their intellectual-
academic failures than general level
students. Surprisingly, neither general nor
advanced level students were internally
responsible for their intellectual-academic
failures than the basic level students. (p.320)
(McMillan, 1992: 149)
73
74. Example: Predictive Research
―Our final three hypotheses dealt with classroom
environment factors…In elementary schools we find
that where teachers perceive class size as
manageable, the reported level of career
dissatisfaction is lower than in elementary schools in
which teachers perceive class size as less
manageable…. In secondary schools, only the
perceived absence of student learning
problems…and the perceived absence of student
behavior problems…emerged as predictors of
teacher career dissatisfactions.‖ (p.72)
(McMillan, 1992: 153) 74
75. Criteria for Evaluating Correlational Studies
1. Causation should not be inferred from correlation.
2. The reported correlation should not be higher or
lower than the actual correlation.
3. Practical significance should not be confused with
―statistical‖ significance.
4. The size of the correlation should be sufficient for
the use of the results.
5. Prediction studies should report accuracy of
prediction for new subjects.
6. Procedures for collecting data should be clearly
indicated.
(McMillan, 1992: 153-156)
75
76. Using Surveys in Descriptive
and Relationship Studies
In a survey, the researcher selects a group of
respondents, collects information (by asking
them a number of questions), and then
analyzes the information to answer the
research questions.
In a Cross-Sectional Survey, information is
collected from one or more samples or
populations at one time.
In a Longitudinal Survey the same group of
subjects is studied over a specified length of
time. 76
77. Causal-Comparative Study
Ex Post facto Research
In Ex Post facto Research the investigators
decide whether one or more preexisting
conditions have caused subsequent
differences between subjects who
experienced different types of conditions
(the phrase ex post facto means ―after the
fact‖).
Back to
research 77
designs
78. Ex post facto vs. experimental
and correlational designs
Ex Post facto designs have some similarities with
both experimental and correlational designs. Like an
experiment, there is typically a ―treatment‖ and/or
―comparison‖ group, and the results are analyzed
with the same statistical procedures. Of course in Ex
Post facto Research there is no manipulation of the
independent variable because it has already
occurred, but the comparison of group differences
on the dependent variable is the same. Like
correlation studies, no manipulation of the
independent variable, so that technically the study is
nonexperimental. However, in a correlation two or
more measures are taken from each subject,
whereas in ex post facto research each subject is
measured on the dependent variable. 78
79. Causal-Comparative Study
Correlational Research
Correlational research, like causal-comparative
research, is an example of what is sometimes called
associational research.
In associational research, the relationships among
two or more variables are studied without any
attempt to influence them.
In their simplest form, correlational studies
investigate the possibility of relationships between
only two variables, although investigations of more
than two variables are common.
A correlational study describes the degree to which
two or more quantitative variables are related, and it
does so by use of a correlation coefficient.
79
80. Similarities and Differences between
Causal-Comparative and Correlational
Research
Similarities. Both causal-comparative and
correlational studies are examples of
associational research, that is, researchers
who conduct them seek to explore
relationships among variables. Both attempt
to explain phenomena of interest. Both seek
to identify variables that are worthy of later
exploration through experimental
research, and both often provide guidance for
subsequent experimental studies.
However, neither permits the manipulation of
variables by the researcher.
80
81. Similarities and Differences between
Causal-Comparative and Correlational
Research
Differences. Causal-comparative studies
typically compare two or more groups of
subjects, while correlational studies require
two (or more) scores on each variable for
each subject. Correlational studies investigate
two (or more) quantitative variables, whereas
causal-comparative studies involve at least
one categorical variable (group membership).
Correlational studies analyze data using
scatterplots and/or correlation
coefficient, while causal-comparative studies
compare averages or use crossbreak tables.
81
82. Similarities and Differences between
Causal-Comparative and Experimental
Research
Similarities. Both causal-comparative and experimental
studies typically require at least one categorical variable
(group membership). Both compare group performances
(average scores) to determine relationships. Both
typically compare separate group of subjects.
Differences. In experimental research, the independent
variable is manipulated; in causal-comparative
research, no manipulation takes place. Causal-
comparative studies provide much weaker evidence for
causation than do experimental studies. In experimental
research, the researcher can sometimes assign subjects
to treatment groups; in causal-comparative research, the
groups are already formed—the researcher must locate
them. In experimental studies, the researcher has much
greater flexibility in formulating the structure of the
design. 82
83. Criteria for Evaluating Causal-Comparative
Research
The primary purpose of the research should be to
investigate causal relationships when an experiment is
not possible.
The presumed causal condition should have already
occurred.
Potential extraneous variables should be recognized and
considered.
Differences between groups being compared should be
controlled.
Causal conclusions should be made with caution.
(McMillan, 1992: 161-162)
83
84. True Experimental
Designs according to
Suter (1998: 196-203)
Randomized
Randomized Randomized
pretest-
posttest matched Randomized
posttest
control control factorial
control
group group design
group
design design
design
84
85. Survey Research
A common form of research involving
researchers asking a number of
questions about a particular topic or
issue (often prepared in the form of a
written questionnaire or ability test) to
a large number of individuals (either
by mail, by telephone, etc.).
85
86. Survey
Research
Cross-
Longitudinal
sectional
collects information
from a sample that has Collects information at
been drawn from a different points in time
predetermined in order to study
population at just point changes over time
in time
86
87. Longitudinal
Survey Research
Changes in a subpopulation group
identified by a common
characteristic over time
Changes in the
same people Trends in the
over time same population
over time
87
88. Cross-sectional
Survey
Research
Community needs
National Attitudes and
assessment Practices
Program evaluation Group
Comparisons
88
89. Weak Experimental Designs
These designs are referred to as
―weak‖ because they do not have
built-in controls for threats to
internal validity.
Any researcher who uses one of
these designs has difficulty
assessing the effectiveness of
the independent variable.
89
90. Weak Experimental Designs
1. The One-Shot Case Study: a single group is
exposed to a treatment or event, and a
dependent is subsequently observed
(measured) in order to assess the effect of
the treatment.
X O
treatment Observation
(dependent variable)
90
91. Weak Experimental Designs
2. The One-Group Pretest-Posttest Design: a
single group is measured or observed, not
only after being exposed to a treatment of
some sort, but also before.
O X O
Pretest treatment Posttest
91
92. Weak Experimental Designs
3. The Static-Group Comparison Design: Two
already existing, or intact, are used.
Comparisons are made between groups
receiving different treatments.
X1 O
X2 O
Note:
------ : already formed, not randomly assigned
X1 and X2: different treatments
Os : placed vertically to each other, occurs at the same time92
93. True Experimental Designs
Subjects are randomly
assigned to treatment groups
for controlling the subject
characteristics threat to
internal validity.
93
94. True Experimental Designs
1. The Randomized Posttest-Only Control Group
Design: involves two groups, one receives the
experimental treatment while the other does not.
Treatment Group R X1 O
Control Group R X2 O
R: random assignment
X1 = T = Treatment
X2 = No treatment
O = test 94
95. True Experimental Designs
2. The Randomized Pretest-Posttest Control Group
Design: both groups are measured twice, the first
measurement serves as the pretest, the second as
the posttest.
Treatment Group R O X1 O
Control Group R O X2 O
95
96. True Experimental Designs
3. The Randomized Solomon Four-Group Design:
involves random assignment of subjects to four groups, with two of the groups
being pretested and two not. One of the pretested groups and one of the
unpretested groups is exposed to the experimental treatment. All four groups
are then posttested.
Treatment Group R O X1 O
Control Group R O X2 O
Treatment Group R X1 O
Control Group R X2 O
96
97. True Experimental Designs
3. The Randomized Matched Control Group Design:
It is similar to the randomized posttest control group design, but it is
distinguished by the use of matching prior to random assignment. This design is
used if the sample size is too small (perhaps less than 40 per group) to
reasonably assure group comparability after random assignment. Subjects are
first rank ordered on a variable closely related to the posttest. Then one of the
two highest – forming matched pair – is randomly assigned to T or C, with the
remaining one being assigned to the other. The next highest matched pair is
similarly assigned, and this until the lowest two matched subjects are assigned
randomly.
Treatment Group M R X1 O
Control Group M R X2 O
97
98. Quasi-Experimental Designs
Do not include the use of
random assignment.
Researchers who employ these
design rely instead on other
techniques to control (or at
least reduce) threats to internal
validity.
98
99. Quasi-Experimental Designs
A. The Matching Only Design:
The researcher still matches the subjects in the
experimental and control groups on certain
variables, but he/she has no assurance that they
are equivalent on others since subjects are not
randomly assigned to groups.
The two groups are intact (they are already existed
before the intervention) and so are probably not
comparable.
An illustration of Matched Control Group Design 99
100. Quasi-Experimental Designs
1. The Matching Only Posttest-Only
Control Group Design
Treatment Group M X1 O
Control Group M X2 O
M = Matched
100
101. Quasi-Experimental Designs
2. The Matching Only Pretest-Posttest
Control Group Design
Treatment Group O M X1 O
Control Group O M X2 O
101
102. Quasi-Experimental Designs
B. Counterbalanced Designs:
Represent another technique for equating experimental and control groups.
Each group is exposed to all treatments, however many there are, but in a different order.
Any number of treatments may be involved.
Researchers determine the effectiveness of the various treatments simply by comparing
the average scores for all groups on the posttest for each treatment.
Example: A Three-Treatment Counterbalanced Design
Group One X1 O X2 O X3 O
Group Two X2 O X3 O X1 O
Group Three X3 O X1 O X2 O
102
103. Quasi-Experimental Designs
C. Time-Series Designs:
involves repeated measurements or
observations over a period of time both
before and after treatment.
O1 O2 O3 O4 X O5 O6 O7 O8
103
104. Quasi-Experimental Designs
D. Factorial Design:
extend the number of relationships that may be examined in an
experimental study
allows a researcher to study the interaction of an independent variable
with one or more other variables, sometimes called moderator variables
Treatment Group R O X1 Y1 O
Control Group R O X2 Y1 O
Treatment Group R O X1 Y2 O
Control Group R O X2 Y2 O
104
105. Threats to Internal Validity
Subject Characteristics
Mortality
Location
Instrumentation
Testing
History
Maturation
Attitude of Subject
Regression
Implementation 105
106. Suggested Readings
Butler, Christopher. 1985. Statistics in Linguistics.
New York: Basil Blackwell.
Fraenkel, Jack R. and Norman E. Wallen.
1990.How to Design and Evaluate Research in
Education. New York: McGraw-Hill, Inc.
McMillan, James H. 1992. Educational Research:
Fundamentals for the Consumer. New York:
HarperCollinsPublishers.
Suter, W.Newton. 1991.Primer of Educational
Research. Boston: Allyn and Bacon.
Singleton, Royce and Bruce Straits. 1999.
Approaches to Social Research (3rd Edition).
Oxford: Oxford University Press.
Wallen, Norman E. and Jack R.Fraenkel. 1991.
Educational Research: A Guide to the Process.
New York: McGraw-Hill, Inc.
106