Lecture 3 (Three claims, Four Validities) Flashcards
(50 cards)
1
Q
variable
A
- an attibute that varies
- must have at least two levels
- sex—> female or male
- age—> 18, 19, 10
2
Q
Level or a variable vs variable
A
DO NOT CONFUSE A VARIABLE AND ITS LEVEL
- depression meaured on two levels—> not depressed or depressed
3
Q
measure
A
- observe and record values
- ex: height weight, sex, stress levels, IQ
4
Q
manipulate
A
- researcher assigns participant to different levels of variable
- participants do not choose
ex: medication dose, treatment group
5
Q
Why can some variable be measured only?
A
they are difficult, impossible, or unethical to manipulate
6
Q
can some variable be _______ and ________
A
manipulated and measured
7
Q
Variable can be described in two different ways
A
- conceptual variable
- operational definition of a variable
8
Q
conceptual variable
A
-concept of interest, sometimes abstract
9
Q
operational definition of variable
A
-specific concrete way in which a concept is measured or manipulated in a study
10
Q
conceptual variable
A
sex, intelligence, study habits, depression
11
Q
operation definition
A
-indicating Mor F, self reported number of minute spent studying, score on 21 item Beck Depression Inventory
12
Q
Three Claims
A
- frequency claims
- association claims
- causal claims
- claims will ne evaluated in terms of four big validities
13
Q
four validities
A
- construct validity
- statistical validity
- external validity
- internal validity
14
Q
frequency claim example
A
- 8 million Americans Consider Suicide Each Year
- at times children Play with the impossible
- deadliest Day for suicides is wednesday
15
Q
Association claim example
A
- eating disorder resk higher in educated families
- swet or Dry? wine shoice tied to personality
- sexual orientation linked to handedness
16
Q
causal claims example
A
- summer sun may trigger suicidal thoughts
- loneliness makes you cold
- collaboration gives recall lift to elderly
17
Q
frequency claims
A
- describe a particular rate or level of something
- how frequent or common something it
- ONE VARIABLE
- variable is MEASURED
- not manipulated
18
Q
Association Claims
A
- argues that one level of a variable is associated with a particuar level of another varibale
- variable correlate, or covary
- when one variable changes, the other also changes in a predictable way
- TWO VARIABLES
19
Q
Four Types of Assocation Claims
A
- Positive
- Negative
- Zero
- Curvalinear
20
Q
Scatterplot
A
- can represent assocations with a scatterplot
- one variable plotted on X axis(horizontal) the other plotted on Y-axis (vertical)
- each dot represents one individual measured on two variables
21
Q
positive assocation
A
- two variables tend to change in same direction
- as X goes up, Y goes up
- as X goes dow, Y goes down
22
Q
negative assocation
A
- two variables tend to change in opposite direction
- as X goes up, Y goes down
- as X goes down, Y goes up
23
Q
zero association
A
no association between variables
-points scattered randomly
24
Q
curvilinear assocation
A
-points tend to cluster around a curved line
25
Direct Relationship
- line with a slope
- positive slope = positive assocation
- negative slope = negative assocation
26
Making Prediction with associated variables
- if two variables are associated, knowling level of one allows for prediction of other
- ex: height and weight
- prediction not perfect (stronger association = more accurate prediction)
27
without knowing association or if zero association.....
prediction mean is best bet
28
causal claims
argues that one varibale is responsible for changing the other
-two variables
-sugguests that variables covary
-uses causal language to suggest one variable causes the other
-note: not all causal claims are warranted!
ONE VARIABLE IS MANIPULATED, OTHER MEASURED
29
Assocation Language
- is linked to
- are more/ less likely
- predicts
- is tied to
- is at risk for
30
Causal Language
- affects
- curbs
- leads to
- helps
* * softening words can be added like could, may, might, suggest
31
Three Criteria of Causal Study
- must establish that two variables are correlated (also true for associations) to support hypothesis
- must show that causal variable came first and outcome variable came later (temporal precedence)
- must establish that no other explanation exists for theis relationship (interal validity)
32
construct validity
-do the meausres of the variable actually measure what they are supposed to
33
statistical validity
-are the statistical conclusions accurate and reasonable
34
external validity
-do the findings generalize to other populations, settings, and situations
35
internal validity
are we able to rule out alternative explanations for a causal relationship?
36
Interogating Frequency Claims
- construct validity
- how was the variable operationalized?
- how well was it measured?
- external validity
- can we generalize these results?
- how were the participants chosen?
- how well does the sample represent the population?
37
Interrogating Association Claims
- construct validity
- how well were each of the variable operationalized?
- external validity?
- can association generalize to other individuals or situations?
- statistical validity
- statistical conclusion errors? effect size? significant association?
38
Good Statistical validity
-minimizes type II and ! errors
39
Type I error
- flase alarm
| - conclude there is an assocation when there really isn't
40
Type II error
- -miss
| - conclude there is no association when there really is
41
How strong is the association (effect size?)
stronger assocations lead to better predictions
| x: height vs shoe height is a better assocation than height vs income
42
Is the assocation statistically significant?
assocation not likely due to chance alone?
43
Interrogating Causal Claims
-make sure 3 criteria for causal claims met
to establish A--> B
1. covariance, as A changes, so does B
2.temporal precedence: A come first in the, before B
3. internal validity: A is only explanation for chage in B, no confounding variable C (third variable rule)
44
experiments to support causal claim
- sell designed experiment required
- one variable manipulated (independent variable)
- another variable measured (dependent variable)
45
by manipulating the independent variable....
- one ensures tempral precedence
- control for alternate explanations
- random assignment is essential
- ensure participants in each group are similar
46
Bad Causal Claim: "Debt stress causes health problems poll finds"
```
1. covariance
yes high stress, more health problems
2. temporal precedence
no debt stress ---> health problems
health problems---> debt stress
WHICH ONE???
3. Internal Validity
No. OSme third varibale (e.g. conscientiousness) can be contributing to both
```
revise claim: Debt stress linked to health problems
47
construct validity
- How (well) was DV measured?
| - How (well) was IV manipulated?
48
external validity
-can we generalize results to other populations or settings?
49
statistical validity
- how strong is the effect (how big is the difference between groups)?
- is it significant or just due to chance?
50
Internal Validity
-is a priority for causal claims!
