Exam 2 Flashcards
1
Q
Bivariate Correlations
A
Associations that involves exactly 2 variables
3 types of associations
1. postive
2. Negative
3. zereo
2
Q
Correlation coefficient
A
r indicates direction and strength of the strength
- Direction (postive or negative)
-strength (how closely related the 2 variables are)
—-more closely related r=1.0 or -1.0
—-weaker = closer to 0
- good for two quantitative variables
3
Q
Scatterplot
A
- can be used for categorical variables
- values fall in one catagory or another
ex= martial satisfaction and online dating - meets spouse online of offline
- marital satisfaction is quantitative
- can use scatterplot
- one variable plotted on x axis other on y axis
- one person = dot
4
Q
Bar graph
A
- can also be used for categorical variable
- levels of the bar reflect the mean (average) within each group
- examine the difference between the groups average to see whether there is an association
5
Q
Correlational Study
A
a study is correlational if it has two measured variables
- the data can be plotted as a scatter plot or a bar graph
- the reported results may be a correlation coefficient or a difference between means
ex = meaningful conversation linked to happier people, dating apps are making marriages stronger
6
Q
interrogating association claims
A
2 MOST IMPORTANT WITH Association
1. construct validity - how well is each variable measured?
2. Statistical Validity - how well does the data support the conclusion? the extent to which statistical conclusions are precise, reasonable and replicable
7
Q
point estimate
A
the value that is the result of your analysis
8
Q
Effect size (statistical validity Question 1)
A
describes the strength of an association
- all else being equal, larger effect sizes are ore important, but small effect sizes can compound and also be important
(0.5. -0.5) - very small or very week
(.20, -.20) moderate
(0.40, -0.40) - unusually very large in psychology, either very powerful or possibly too good to be true
9
Q
Confidence Interval (statistical validity question 2)
A
- margin of error of the estimate
- how precise is the estimate?
- a range designed to include the true population value a high proportion of the time (usually 95%)
- does the confidence interval include zero?
- if it does not the relationship is statistically significant
- if it does include zero, the relationship is NOT statistically significant
- smaller sample sizes result in wider CIS (less precise)
- larger sample sizes result in narrower CIs (more precise)
10
Q
Has it been replicated? (statistical validity question 3)
A
- conducting the study again, making sure replications prove the same thing
11
Q
Could outliers be affecting the assoc.? ( Statistical validity question 4)
A
- outlier- an extreme score, single case that stands out from the rest of the data
- outliers can make correlations appear stronger
- more problematic when they have extreme values on both variables
12
Q
Is there a restriction of range (statistical validity question 5)
A
- when there is not a full range of scores on one of the variables
- can make the association appear weaker than it really is
- is there a relationship between
exercise and well being? - study people in a runner’s club to see whether time spent running associated with higher levels of happiness
- ppl in running club are going to run a certain amount per week
- relationship between SAT scores and college GPA
- already SAT scores restricted to get into college
13
Q
Is the association curvilinear? (Statistical validity question number 6)
A
- curvilinear association -correlation coefficient is close to zero, relationship between 2 variables is not a straight line, positive up to a point then negative
- ## can be detected using scatterplots
14
Q
Internal Validity and association claims
A
- not necessary to interrogate internal validity for an association claim, but we need to protect ourselves from temptation to make a causal inference
need 3 things for causal claim
1. covariance
2. temporal precedence
3. internal validity
15
Q
Potential third variables
A
- you may be able to identify serveral third variables that could potentially explain a bivariate association
- the third variable must correlate with both variables in the association
ex = association between height and hair lenght - gender may be impact, taller= boys, less hair
16
Q
Spurious association
A
- bivariate correlation is there but only because of some 3rd variable
17
Q
Causal Claim
A
- use powerful verbs like, makes influences, and effects, stating something about interventions and treatments
18
Q
Experiment
A
researchers manipulated at least one variable and measured another, can take place just about anywhere
19
Q
manipulated variable
A
variable that is controlled, researchers assign participants to random value or variable
- self reports, behavioral observation, psychological measures
20
Q
independent variable
A
- manipulated variable
21
Q
dependent variable
A
- measured variable, outcome variable, how partcipants are recorded on depent variable based on assigned indep.
22
Q
control variables
A
any variable held constant on purpose
23
Q
how do experiment support causal claims
A
- covariance - comaprison group vs control group
- temporal precedence - casual variable first
- internal validity - explored alternative explanations
- covariance - comaprison group vs control group
24
Q
confounds
A
- alternative explanations, threats to internal validity
25
design confound
experimenters mistake, occurs when a 2nd variable happens to vary alongside indep. variable
= poor internal validity, cannot support a causal claim
26
unsystematic variability
certain babies like music more, okay if not all in the same gorup systematically
27
selection effects
can happen when let participants choose groups, or when if experimenters assign one type of person to a specific group
28
Random assignment
used to avoid selection effects, a way of desystematizing types of participants that end up in each group
ex - match groups, top partcipants paired, 2 best paired, and so on x
29
independent group design (also called between subjects design or bw groups design)
- separate groups of participants are placed into different levels of the participants
2 basic forms of indep group design
1. posttest only design
2. pretest/post test design
30
within groups design/within subjects design
- each person is presented with all levels of the independent variable
1. repeated measures design
2. concurrent measures design
31
pretest/ post test design
- participants are randomly assigned to at least 2 groups, and are tested on the key dependent variables twice
- one before, once after exposure to the independent variable
- may use a pretest.post test design when they want to be sure random assignment made groups equal
- can be absolutely sure, no selection effects
- enables researchers to track people's change in performance over time
32
post test - only design
- participants are randomly assigned to independent group and are tested on the dependent variable once
33
repeated measures design
- participants are measured on a dependent variable more than once after each exposure to each level of the IV
- participants experience both levels
one group -> tast chocolate with the confederate -> rate chocolate-> taste chocolate alone -> rate chocolate
34
concurrent measures design
- participants are exposed to all levels of the IV, at roughly the same time, a single attitudinal behavioral preference is the DV
one group -> shown female face and male face -> track looking preference
35
Advantages of within groups design
- ensures participants in the 2 groups will be equivelant because same partcipants
- matched groups design can be treated as within groups
- requries fewer partcipants
36
Within groups design make casual claims?
1. covariance?
within groups design allow researchers to manipulate an IV and incorporate comparison groups
2. Temporal precedence?
researcher controls IV, can make sure it comes first
3. internal validity
don't have to worry about selection effects because participants exactly the same in the 2 conditions
- do need to avoid design confounds
- order effects
37
Order effects
- when exposure to one level of IV, influences responses to the next level
- behavior at later levels may be caused not by experimental manipulation but rather sequence
- can include practice effects and carryover effects
38
Practice effects
- long sequence may lead participants to get better at a task or to get bored or tired at the end
39
carryover effects
- some form of contamination carry over from one to another condition
40
Avoid order effects by counterbalancing
- present the levels of IV to participants in difference sequences, with counterbalancing any order effects should cancel each other out
ex - full counterbalancing, partial counterbalancing
41
full counterbalancing
- all possible conditions are represented, repeated measures design w/2 conditions is easy to counterbalance (A -> B) (B-> A)
- as number of conditions increase so does combo
42
Partial counterbalancing
which only some of the possible condition orders are represented
- present conditions in random order
- latin square
43
Disadvantage of within groups design
1. repeated measures have potential for order effects
- can threaten internal validity
- can usually control by counterbalancing
2. within groups design may not possible
ex - once taught a skill can't reteach the skill
3. when people see all levels of IV, and then change the way they would normally act
demand characteristic
44
Demand Characterisitc
- a cue that can lead participants to guess an experimenters hypothesis
45
pretest/post test vs repeated measures
- pretests vs posts test - participants see only one level of IV, not all levels
repeated measures- participants exposed to all levels of meaningful IV
- levels can be counterbalanced
46
interrogating causal claims w/4 big validities
1. construct validity
construct validity of dependent variable
- ask how well researcher measured the dependent variable
construct validity of indep. variables
- ask how well the researchers manipulated or operationalized them
- manipulation check
- pilot study
2. External validity - you ask whether the casual relationships can generalize to other, people, places, and times how were participants
how were participants recruited?Random sampling?
3. Statistical validity - asking about effect size, precision of the estimate, and replication
4. Internal Validity - often the priority, if a study had internal validity --> causal claim
if a study has confounds -> Not a casual claim
47
Manipulation check
extra dependent variable that researchers can insert into an experiment to convince them that their experiments manipulation worked
48
pilot study
- simple study, using a separate group of participants, that is completed before/after the primary test, to confirm effectiveness of manipulation
49
Internal Validity - Priority, how many threats?
12 internal validity threats
1. design confounds
2. selection effects
3. order effects
50
4. Maturation threats
- a change in behavior that occurs/emerges more or less spontaneously ver time
preventing: pretest/post test design, included age, comparison, etc
51
5. History Threats
- "historical" or external factors that systematically effect most members of the treatment group, at the same time as the treatment itself, making it unclear if change is caused by the treatment received
preventing - comparison groups can help
52
6. Regression threats
refers to a specific concept called regression to the mean, an unusually good performance is likely to regress down, towards the mean, unusually bad performance likely to regress up to the mean
- can occur only when a group is measured twice, and a group had an extreme score in the pre test
preventing - comparison groups, careful inspection of the pattern of results
53
7. Attrition threats
- pretest/post test, a threat that occurs when a systematic type of participant drops out of the study before it ends
- preventing = when participant drops out of a study, remove there pre - test score
54
8. Testing threat
- change in the participants as a result of taking a test more than once, people may have become more practiced with the test, or more bored/fatigued
- preventing= abandon a pretest all together, only do post test, or if they do a pre test use alternative forms of measurement
55
9. instrumentation threat
- occurs when a measuring instrument changes over time, ex = a person coding behavior changes, or use different forms of pretest/post test that are not equivalent
preventing = researchers switch to post test only design, or make sure pretest/post test is equivalent
56
combo threats - selection history threat
an outside event or factor only affects those at one level of the IV
57
combo threat - selection attrition threat
- only one of the experimental groups experiences attrition
58
10. observer bias
researchers expectations influence there interpretations of the results, can threaten internal validity and construct validity
solution = double blind study, or masked design
59
Three potential internal validity threats to ANY study even with clear comparison groups
1. observer bias
2. demand characteristics
3. placebo effect
60
double blind study
neither the participants or researcher who evaluates them knows who is the treatment group and who is the comparison group
61
masked design
participants know what group they are in, but the observer does not
62
placebo effect
- people receive a treatment and really believe they improve, but only because participants believe they are experiencing real treatment
solution = double blind placebo control study
63
double blind placebo control study
- neither people treating or participants, know whether they are in the real group or placebo
64
Null effects
finding that an independent variable, did not make a difference int eh dependent variable, no significant covariance between the 2
65
reasons for the null effects
- weak manipiulations
- insensitive measures
- ceiling effect
- floor effect
- noise
- measurement error
66
weak manipulations
not enough to matter, important that researcher operationalized the indep. variable well
67
insensitive measures
- sometimes researchers have not operationalized the dependent variable with enough sensitivity, important to have detailed, quantitiative increments
68
ceiling effect
- all the scores are squeezed together at the high end
69
floor effect
- all the scores are squeezed together at the low end
70
noise
unsystematic variability among the members of a group in an experiment
71
measurement error
- a human or instrument factor that can randomly inflate or deflate a persons true score
solution 1 = use reliable, precise tools
solution 2 = measure more instances
72
solutions to individual differences
- change the design
- add more participants
73
situation noise
- external distractions
- carefully control environment
- one of the reason lab are sterile and boring
74
power
- aspect of statistical validity
- likelihood that a study will return an accurate result when the independent variable really had an effect
