Ch.9, Independent Groups

Question 1

Random Groups Design: t

Answer 1

Simplest independent groups design; there are two groups and participants are randomly assigned to one of these two groups/conditions

Question 2

Random Assignment:

Answer 2

occurs when each participant in the sample has an equal chance of being assigned to each of the two groups; each participant is only assigned to one of the two groups

Question 3

Experimental/Treatment Group:

Answer 3

group that receives the intervention

Question 4

Control Group:

Answer 4

does not receive the intervention

Question 5

Advantages of Experiments over Correlation

Answer 5

Experiments can begin to make causal statements
Correlation is ideal when you can’t manipulate the IV
“Affect, impact, and cause” can be statements made with experiments

Question 6

Statistical Power:

Answer 6

a researcher’s ability to find an effect size of interest if it actually exists

Question 7

Null Hypothesis and underpowered studies

Answer 7

with underpowered studies (studies that have not used enough participants), it is likely that no group differences will be produced ‘

Question 8

Data-Peeking:

Answer 8

occurs when researchers run statistical analyses after collecting a smaller subset of their data to get a feel for whether there is already statistically significant results (if they believe they have, they may choose to terminate data collection early, or they may choose to continue the data peeking until they reach p=0.05)

Question 9

Single-Blind Design:

Answer 9

participants don’t know which group they are assigned to (ideal because participants are less likely to have preconceived notions about how the experimenter expects them to behave in the study)

Question 10

Double Blind Design

Answer 10

neither the experimenter nor participants know which condition they have been assigned to

Question 11

Matched Groups

Answer 11

Researchers may choose to match their experimental groups on some measure to ensure their groups do not differ drastically (age, socioeconomic class)

Question 12

Statistical Significance

Answer 12

A difference in a group is significant if there is a small likelihood of observing results as more extreme as the current results if the null hypothesis is true
When analysis reaches the p<.05 level, the result is statistically significant
PROVES THAT CONTROL AND TREATMENT GROUPS DIFFER FROM ONE ANOTHER

Question 13

Practical Significance

Answer 13

When something has real world impact, it is practical significance with a high external validity
Practical significance is when the group differences translate to real world and impactful differences

Question 14

Two Independent Group Design

Answer 14

Used when a participant could be randomly assigned to one of your groups and not the other

Question 15

Inferential Statistics vs. Descriptive Stats

Answer 15

better than descriptive stats because they do not only describe, the allow for prediction over a population

Question 16

l

Answer 16

l

Question 17

What is the problem with running multiple separate t-tests?

Answer 17

The more separate t-tests that are run, the more likely you are to find false positives in data

Question 18

Post-Hoc Analyses:

Answer 18

comparisons after the fact that the researcher did not plan on

Question 19

HARKing:

Answer 19

Hypothesize after the results are known; problematic when the researchers do not disclose that they came up with their hypothesis after running some data analyses and claim they knew it all along

Question 20

Over-trimming outliers

Answer 20

is a p-hacking technique so that data appears more homogenous

Question 21

Golden Rules for Experiments:

Answer 21

random assignment to condition, every member of a group should have an equal chance on being assigned to group, MAKE SURE EVERYTHING IS IDENTICAL BETWEEN CONDITIONS O

Question 22

Why should groups be matched in smaller sample sizes?

Answer 22

Interested in cause but the population of interest is not that big
Deciding which group is experimental or control can no longer be randomied in this case: therefore people need to be matched based on some variable of importance (age/gender) and then split into either experimental or control condition
Need to do this in order to MAKE CAUSAL CLAIMS; IF GROUPS ARE NOT EQUAL IN THIS REGARD THERE IS DECREASED INTERNAL VALIDITY

Question 23

Power Analysis and its purpose

Answer 23

done ahead of time so that we know whether a study is underpowered or whether there is just a small effect size (know whether to increase amount of participants)