PSYC3042 Final: PSYC3042: Terms A - C

183 views5 pages
School
Department
Course
A - C
Attributes of
Relationships
Form.
e.g., linear.
e.g., curvilinear
e.g., cyclical (e.g., depression mood over time).
Direction.
e.g., positive (or forward) vs. negative (or inverse).
Strength.
e.g., effect size.
e.g., % of variance explained.
e.g., the slope of the change in the DV as a function of change in the IV.
Between-subjects
design
Condition (levels of IV) are varied between participants, so that number of
conditions = number of groups.
Advantage: No contamination
Disadvantage: Between Subject variability
Mitigating the disadvantages: Randomize the test condition assigned to
participants.
Benchmark Check
Also used to for sensitivity, but:
For construct validity, the issue is whether the phenomenon or DV is behaving as
we would expect if we have tapped the intended constructs (independent &
dependent).
Two main questions that can be addressed:
1. Does the task (IV) produce recognised (known) effects?
2. Is the DV responsive to variables previously established as
influential?
1. Does the IV task produce recognised effects?
e.g., Liebert & Baron: Replicated the previously-established effect of exposure to
violent TV (old IV) on aggressive play in the ‘benchmark’ free-play DV task (old
DV)
validating the IV manipulation & demonstrating the IV had expected effects.
2. Is the DV responsive to variables previously established as influential?
e.g., Liebert & Baron: TV violence manipulation (old IV) also affected button-
press DV (new DV), so the benchmark check helped to validate new DV as a
measure of aggression.
Causation
Causation can be inferred most easily when the suspected causal factor is
manipulated (i.e., a true experiment) and the influences of all other
contaminating factors or artifacts are excluded.
Checking Results for
a confound
General Strategy we’ll discuss for checking results for a confound:
Check data already available to rule in or out a possible confound.
Techniques involving Checking the Results for a Confound:
(a) Check Covariation of Confound with IV or DV:
If a suspected confound is not correlated with both the IV and the DV, it cannot
confound the results.
Example:
Good spellers may be better than poor spellers on a serial recall task.
Problem:
Unlock document

This preview shows pages 1-2 of the document.
Unlock all 5 pages and 3 million more documents.

Already have an account? Log in
Good spellers may also have higher IQ than bad spellers.
Solution: Show that participant’s IQ is not correlated with serial recall scores.
IQ cannot be a confound if it is not correlated with the DV.
(b) Assess Participants Awareness of Aims:
e.g., Use a post-experimental questionnaire to check whether the participants
worked out the hypothesis (e.g., Petty et al.).
Detection of a potential confound can be followed up by correction:
e.g. Remove data of suspicious Ps
e.g. Analyse Ps who figured out the hypothesis vs. not separately.
(c) Check for Observer Agreement & Validity:
Example: An anti-tantrum program reduces number of tantrums in children, as
shown by pre-/ post comparisons.
Problem: A systematic change in tantrum behaviour may be due to expectancy
effects on the part of the observers
E.g., reduction in number of tantrum resulting from raters gaining more
experience in classifying tantrums.
Solution:
Videotape the tantrums, and get ratings of tantrums from independent blind
raters, in non-chronological order.
Construct
A construct is a concept that is an element in psychological theorising (e.g., in the
hypothesis that head injury impairs working memory).
Often, constructs are abstract, and not directly observable.
Constructs can be ‘mental states’.
e.g., self-esteem, motivation, intelligence, hunger, aggression, knowledge,
empathy, etc..
BUT some constructs are not ‘mental states’ (cf. M&J).
e.g., diffusion of responsibility (e.g., groupthink; the bystander effect).
e.g., transparency of idiom, quantity of neurotransmitter at synapse.
Construct Validity
Definition: The degree to which an operation measures or manipulates the
targeted construct.
Precondition: Measure must have reliability.
Threats to construct validity:
Poor construct definition
Confounds
Reactivity
Demand characteristics
Experimenter bias
Floor and ceiling effects
In the psychometric context, validating focal constructs may involve:
[Face Validity.]
Criterion Validity.
Content Validity
Convergent Validity.
Discriminant Validity.
The last two are multi-trait and multi-method techniques.
Controls can apply to self-report measures, Likert scales, etc..
SEE INDIVIDUAL CELLS FOR AN EXPLANATION OF EACH KIND OF VALIDITY
Unlock document

This preview shows pages 1-2 of the document.
Unlock all 5 pages and 3 million more documents.

Already have an account? Log in

Document Summary

Form. e. g. , linear. e. g. , curvilinear e. g. , cyclical (e. g. , depression mood over time). Direction. e. g. , positive (or forward) vs. negative (or inverse). Strength. e. g. , effect size. e. g. , % of variance explained. e. g. , the slope of the change in the dv as a function of change in the iv. Condition (levels of iv) are varied between participants, so that number of conditions = number of groups. Mitigating the disadvantages: randomize the test condition assigned to participants. For construct validity, the issue is whether the phenomenon or dv is behaving as we would expect if we have tapped the intended constructs (independent & dependent). Causation can be inferred most easily when the suspected causal factor is manipulated (i. e. , a (cid:862)true experiment(cid:863)) and the influences of all other (cid:862)contaminating(cid:863) factors or artifacts are excluded. Ge(cid:374)eral trateg(cid:455) (cid:449)e(cid:859)ll dis(cid:272)uss for (cid:272)he(cid:272)ki(cid:374)g results for a (cid:272)o(cid:374)fou(cid:374)d: Check data already available to rule in or out a possible confound.

Get access

Grade+20% off
$8 USD/m$10 USD/m
Billed $96 USD annually
Grade+
Homework Help
Study Guides
Textbook Solutions
Class Notes
Textbook Notes
Booster Class
40 Verified Answers