Transcript Theory consists of

Nature of Research
Pt 2
Definitions:
CONSTRUCT:
• a fabrication of mind,
invisible, unmeasurable.
THEORY:
• built of constructs,
• showing relationships
between constructs,
• a systematic view of
phenomena to explain
and/or predict,
THEORY:
• a network of related
constructs,
• abstract.
Theory consists of:
1. Definitions, explication of
constructs.
2.
Operational Definitions.
Theory consists of:
• what we wish to measure,
observe or manipulate.
Theory consists of:
• criteria: must exhaust the
definition of the constructs, must
be exclusive (not include
additional things).
Theory consists of:
•Explication of constructs.
•Operational Definitions.
•What we wish to measure,
observe or manipulate.
•Criteria
VARIABLE:
Any quality that has more
than one value.
HYPOTHESIS: Pertaining to the
expected relationships among
variables.
HYPOTHESIS
•
Must be specific expression
of expected answer about
relationships among variables.
• Interaction of operational
definitions, concrete, observable.
HYPOTHESIS
From testing of the hypothesis
one draws inferences which may
allow generalization based on
the theory.
VARIABLES
• can be added to sharpen
the prediction,
provide more explanation.
VARIABLES
Independent Variable: cause.
Dependent Variable: effect.
(depends on independent
variable)
“If [I.V.] then [D.V.].”
VARIABLES
Moderator Variable:
moderates the effect of
I.V. on the D.V.
More than one variable
effects outcome.
VARIABLES
Mediated Variable: one
that intervenes, allows one
variable to act upon
another.
VALIDITY: Internal
• experiment is well
controlled, clean.
• clearly measures what is
concluded.
• clearly tests the question
asked, the target question.
• with no confounding
variables
VALIDITY: External
• artificiality of created
situation.
• is the conclusion
generalizable outside of
experimental situation?
VALIDITY: External
Ex. Piaget studied his own children.
Can this be generalizable?
A problem of Sample Validity.
VALIDITY: External
Ex. Study of violent tv watching.
Experimental Group
watches 3 hrs. of violent tv.
VALIDITY: External
Ex. Study of violent tv watching.
Experimental Group
watches 3 hrs. of violent tv.
No Control Group.
No Choices given.
A problem of Extreme Treatment.
Measurement
Is behavior observed a typical
expression of what is to be
measured?
Ex. What does “time-on-task”
measure?
CAUSAL RELATIONSHIP
• must show statistical
relationship
• must precede event
• must rule out plausible
rival hypotheses (PRH)
CONTROL GROUP
• Ask, What is it designed
to control for?
• Must be equivalent to
experimental group.
• Function: to rule out
Plausible Rival Hypothesis.
Threats to Internal Validity
• History
If the study has multiple
phases, there may be an
influence by the early
phase on the later phases.
Threats to Internal Validity
• Testing
All participants must be
sensitized to the same
extent.
Threats to Internal Validity
• Mortality
If dropouts are not equally
distributed (randomly)
there may be systematic
bias.
Threats to Internal Validity
• Maturation
Participants would improve
anyway by growing up.
HAWTHORNE EFFECT:
Effect of attention paid to
participants.
Threats to Internal Validity are
less important in correlational
studies than in functional studies.
(A correlational study does not
show causal relationship.)
Regression Toward the Mean
Performance varies, and
over time values will
gravitate to the mean.
Highs become lower, and
lows become higher.
Random Assignment
(Random Selection)
Allows for comparability between
experimental group and control group.
Each individual has the same chance to
be in either group.
Otherwise there may be bias.
(RA)
(RS)
Sampling
Science of choosing
representationally from
each ethnic, racial, age,
class, etc., group.
Allows validity with only
1100 polled of 260 million.
Precision
Difference between groups
=
_______________________
Difference between individuals
within group
Precision
DB
_____
DW
=
Probability (p)
Standard Deviation
Size of variability of distribution
of individual differences.
Mean (x) yields some information,
not necessarily meaningful.
Range tells more of the story.
Standard Deviation
Shoe Sizes
A
B
8
Mean (x)
8.5 10 9.5
11.5 6
12 6.5
9.0
9.0
Standard Deviation
Shoe size Mean Deviation SS
(x)
11.5
6
12
6.5
9
9
9
9
2.5
-3
3
-2.5
6.25
9
9
6.25
Variation Std Dev
(s2)
()
1.56
2.25
2.25
1.56
2.76
(x)2
SS/N
Standard Deviation
X




6 7 8 9 10 11 12
Standard Deviation
X







6 7 8 9 10 11 12
Standard Deviation
X







6 7 8 9 10 11 12
Standard Deviation
X







6 7 8 9 10 11 12
Standard Deviation
X







6 7 8 9 10 11 12
Statistical Significance
“Significance of difference”
or, meaningfulness.
Statistical Significance
Ex.
Grp #1–Individual Instruction
Grp #2–Group Instruction
n=30 in each group
(could be 3000)
RA, RS
Grp. #1 does 3 pts. better on posttest.
Is that Significant?
Statistical Significance
Ex.
Grp #1–Individual Instruction
Grp #2–Group Instruction
n=30 in each group
(could be 3000)
RA, RS
Grp. #1 does 3 pts. better on posttest.
Is that Significant?
What is the probability that observed
differences between the Experimental
group and the Control group is due to
chance?
Probability (p)
p = (probability that difference is
due to chance)[+ –] .05
95% sure of significance
N.B.
Level of significance is not level
of importance.
Probability (p)
p = >.05 is the agreed upon standard
>.07 is acceptable
End part 2