Midterm I Flashcards

Question

Can self-reports be trusted? Why?

Answer 1

Yes, people can be trusted, just need good questionnaires. Think about the questions and potential limitations.

Answer 2

``` Good Design o Captured construct well o Appropriate number of items Good answers o Order effects o Response sets (similar answers= to identify response sets) o Faking ```

Answer 3

1. Exactly two variables 2. Both variables are measured 3. The minimum value is -1 4. The maximum value is +1 5. A value of 0 indicates no association between the variables

Answer 4

1. As low as ± .10 shows a small or weak association 2. As low as ± .30 shows a medium or moderate association 3. As low as ± .50 shows a large or strong association

Answer 5

1. Covariance of cause and effect: Results must show a correlation, or association, between the cause variable and effect variable 2. Temporal precedence: The method must ensure that the cause variable preceded the effect variable; must come first in time 3. Internal validity: No other plausible alternative explanations for the relationship between the 2 variables.

Answer 6

Outliers are unusual observations or scores that are different from the rest. Yes, but justification is needed to prevent attrition and regression to the mean. It is often not needed in a big sample.

Answer 7

1. Design confounds: Does another variable vary systematically with the IV? (Need for control variables) 2. Selection effects: Are different kinds of participants in the groups? (Independent-groups designs) 3. Order Effects: Are later responses systematically affected by earlier ones? (Within-groups designs)

Answer 8

- Random Assignment: Everyone has an equal chance to be in whatever group - Matched groups: Participants who are similar on some measured variables then randomly assigned to experimental conditions.

Answer 9

By counterbalancing which is to present the levels of IV to participants in different sequences

Answer 10

Independent: No contamination across IV levels Within: Requires fewer people, Individuals are their own controls (power!)

Answer 11

Independent: Requires more people (control group!) Within: Order/Demand effects (participants recognize what you're looking for), may not be possible

Answer 12

If the confidence interval does not include 0

Answer 13

Good=60 | Bad=40

Answer 14

Isolated experiences

Answer 15

That your manipulation or measured IV is not effective at producing a change in the outcome, but it seems like it is so Type I error • Could be the case that manipulation seemed only somewhat effective or not at all, yet it should have been • Threat to internal validity of our claim

Answer 16

No but can recognize their impacts

Answer 17

Comparison groups are the best way to identify these threats to validity to do something about it or at least recognize it

Answer 18

Special case of order effects, where the testing process itself changes the outcome (such as response fatigue or guessing)

Answer 19

Some unrelated event occurs during the study, affecting the entire study or an entire subgroup (not design cofound), can’t be planned, difference between groups

Answer 20

Natural or spontaneous change occurring, affecting the entire study or entire subgroup (difference internal to the individual that changes results, natural changes within the person. Often in long-term research)

Answer 21

Outliers dropped out of the study, normalizing the group (not include scores from people that did not finish the study, attrition problem is solved except when have a lot of outliers because could mean no effect or that participants didn’t like the study or what was being tested) Both pre-test groups have similar results and have a different post-test result. Look equivalent as means but not as dots. Would look like a good effect but no effect when outliers are no longer in the study.

Answer 22

Extreme scores at T1 have normalized at T2 (comparison groups, random shit happens, think there was an effect) • If significantly different from comparison group/ the mean, might indicate there was an effect. • Want to make sure starting point= equivalent (otherwise there might be something unusual causing regression to the mean)

Answer 23

Decay of equipment accuracy (or coding method) (think of machinery such as computer or if person such as coder, important to train them a lot). Would only reduce reliability if random fluctuations

Answer 24

Rating or data become adjusted to fit hypothesis. To match what you want to see (could be intentional or not) Solution: Don’t tell hypothesis to research assistant and participants

Answer 25

(Demand Characteristics) Participants attempt to produce the hypothesized outcome (if figure out hypothesis, want to be good participants) Solution: Asking participants what they believe the hypothesis of the study was. Diminished by using a between-subject design

Answer 26

Participants will themselves to change (improve). Especially when a drug is involved Solution: Was there a placebo group?, see similar effect, have a 3rd comparison group

Answer 27

Within-subject: Need to look at the stats, can’t trust figure Between-subject: Can trust the figure

Answer 28

o Which part of the hypothesis is being addressed? o What was the outcome? o Is the outcome consistent with the hypothesis?

Answer 29

Can explain if there's an alternative explanation o Rules out potential selection effect o Control group makes you confident that you can rule out alternatives.

Answer 30

Regression to the Mean: One group as more extreme scores at the beginning than the other one but at post-test, there is no difference (=regression to the mean) Attrition: Both pre-test groups have similar results and have a different post-test result. Look equivalent as means but not as dots. Would look like a good effect but no effect when outliers are no longer in the study.

Answer 31

Regression to the mean, Testing and Demand effects are very likely (demand effects especially in letter of informed consent when stating hypothesis)

Answer 32

History, Attrition, and Instrumentation Effects are pretty unlikely

Answer 33

In the scenario where it is involved, yes!

Answer 34

Helps us to avoid accepting bad arguments as valid ones | Helps us recognize ways to improve for follow-up studies

Answer 35

When there are plausible alternative explanations that can come in form of alternative causal variables Can hold them constant as control variables in an experiment (see differences in IV and DV, and verify by how much it deviates= identify 3rd variable)

Answer 36

o Multiple Regression | o Partial Correlation (can deal with more 3rd variables at once)

Answer 37

Answer questions about an association (or set of associations) while controlling (correcting or taking into account) some other association

Answer 38

a. Can take into account a potential confound= improve internal validity b. If determine preferred causal variable has stronger influence than another cause= improve internal validity

Answer 39

Means that there is no association between IV and DV

Answer 40

Overall, how well can we predict the outcome can be explained by all the variables combined. Betas is a better predictor.

Answer 41

- Mediation | - Moderation

Answer 42

Occurs when other variables help us better understand the nature of the (causal) association between 2 variables.

Answer 43

No! It’s the mediator that helps us better understand the nature of the relation between IV and DV

Answer 44

Something that is somewhat affecting the outcome (on IV and DV)

Answer 45

Mediation: Mediation occurs when some other variable is an intermediary in a causal association Moderation: Moderation occurs when some other variable controls whether your association does or does not appear, involves causal association

Answer 46

Longitudinal designs help address the problem of temporal precedence with correlations

Answer 47

o Follow participants over longer periods of time (which can be hard though) o Lot more work than cross-sectional study designs (all done at same time) o If 2 variables, use of correlations (across variables and when know not manipulating something) o In longitudinal, still have cross-sectional designs because holding it constant every time by being measured at the same time

Answer 48

Correlation with itself, correlation with the same variable overtime, self-correlations

Answer 49

Correlating 2 variables across time

Answer 50

Degree of simplicity in a theory Occam’s Razor states that all else being equal, the solution that makes the fewest assumptions is usually the correct one

Answer 51

Demonstrating a series of different findings (preferably different studies) converges on the same solution

Answer 52

If multiple studies show support for the same, simple conclusion it’s much more likely that your proposed causal association is correct