L2

Question 1

When can we make causal claims?

Answer 1

When we use an experimental design (manipulate the IV and measure DV)

Question 2

What is a correlation design?

Answer 2

Relationships among variables that are observed and measured

Question 3

What is operationalisation?

Answer 3

Strictly defining variables so they can be measured

Question 4

Why do we need to be careful when thinking about correlation vs causation?

Answer 4

Sometimes there are other variables which are related to both correlated variables that we are unaware of that is causing the difference

Question 5

What is the third variable problem?

Answer 5

A confound in correlational studies: where another ‘lurking’ or unobserved variable can explain the relationship between the observed variable.

Question 6

Wine being correlated with weight loss is an example of?

Answer 6

The third variable problem

• E.g. wine is usually drank while socialising and being social is more likely to make you want to lose weight (the third variable)*
• Knowing there is a relationship between two variables does not mean there is causation*

Question 7

What do you need to do with your study to infer causation?

Answer 7

Have a randomised controlled experiment

• Where experimenters control the independent variables, randomly assign participants and materials and help manage other sources of variation.*
• Allows us to say X increases Y, X decreases Y, X results in Y etc.*

Question 8

What are the two methods we can use to compute correlations/measure relationships?

Answer 8

Pearsons correlation coefficient (r)

Spearmans rank correlation coefficient (rs)

Question 9

Describe the features of Pearsons correlation coefficient

Answer 9

Question 10

Describe the features of Spearman’s rank correlation coefficient

Answer 10

Question 11

What does a monotonic relationship mean?

Answer 11

as x goes up, y goes up

does not go up and down

Question 12

Which correlation analysis is best when the data is assumed to be linear?

Answer 12

Pearson’s

Question 13

How can we measure if X predicts Y? (rather than a causation or correlation)

Answer 13

Simple Linear Regression

Question 14

Describe the features of a simple linear regression (R²)

Answer 14

Question 15

When should we use multiple regression?

Answer 15

When there is more than one X variable

Question 16

Describe the features of a multiple regression (R²)

Answer 16

Question 17

Does a significant relationship between variables imply causation?

Answer 17

No

Question 18

What two factors creates variations in our data?

Answer 18

1. Systematic factors (genuine data)
2. The variations look systematic but it is only chance (chance explanations)
   * E.g. sampling error, when data looks legitimate but it is flawed in some way*

Question 19

Detecting Signals

What is Signal Detection Theory (hypothesis testing)?

Answer 19

Question 20

How many ways of being right and how many ways of being wrong are there in signal detection theory?

Answer 20

2 x 2

Question 21

How is signal detection theory similar to null hypothesis testing?

Answer 21

There are two ways of being right and two ways of being wrong

Question 22

A false alarm in signal detection theory is equivalent to what in null hypothesis significance testing?

Answer 22

Type 1 Error

Question 23

If you reject the null hypothesis but the effect was absent, what type of error is this?

Answer 23

Type 1 error (false positive)

Question 24

If p<.05 can the effect still be due to chance?

Answer 24

Yes, 1 in 20 cases will be due to chance as there is still a 5% probability of being due to chance

Question 25

Describe where the decisions are made on a graph using signal detection theory

Answer 25

Question 26

Getting better evidence and reducing chance factors reduces the degree of ___ in signal detection theory

Answer 26

**Overlap** (chance 1 and 2 errors)

Question 27

Is psychology good at operationalising and having low type 2 and type 1 error rates?

Answer 27

No, our experiments are typically ‘fuzzy’ in terms of clarity

Question 28

If variables aren’t controlled properly this can lead to…

Answer 28

Type 1 and type 2 errors

Question 29

What are the two types of **response bias** in decision making in null hypothesis significance testing?

Answer 29

Liberal and Conservative

Question 30

What is a liberal response bias?

Answer 30

When experimenters tend to err in favour of saying a relationship exists as opposed to saying a relationship doesn’t exist (i.e. higher type 1 error rate, lower type 2 error rate)

Question 31

What is a conservative response bias?

Answer 31

Where you lean towards accepting the null hypothesis instead of rejecting the null

Question 32

What is the trade-off of having a conservative response bias?

Answer 32

You reduce the number of type 1 errors however you increase the number of type 2 errors

Question 33

What are the positive aspects of a conservative response bias?

Answer 33

Increased number of correct rejections

Question 34

What is the scientific communities standard for response bias?

Answer 34

p \< .05 (5%)

Question 35

What else can we vary in signal detection?

Answer 35

The base rate (size of the distributions) *However in most experiments we normalise each condition*

Question 36

Baggage screeners at an airport are an example of

Answer 36

Signal Detection Theory with a liberal response bias ## Footnote *They have a large ‘noise’ as they correctly reject a lot of bags but they have adjusted the decision making criteria*

Question 37

What does having a noise curve that is highly variable mean (high variance of the curves)?

Answer 37

How variable our curves are (how different the things are that make up the curve)

Question 38

Response criteria (response bias) can be...

Answer 38

Liberal or conservative