L02 Design

Question 1

Types of study methods

Answer 1

Experimental
Quasi-experimental
Correlational

Question 2

Types of study designs

Answer 2

Between subject
Within subject
Matched subject

Question 3

Types of variables

Answer 3

IV Independent variable/ DV Dependent variable

Or predictor and outcome variables

Question 4

Hypothesis

Answer 4

Alternative/Experimental hypothesis H1

Null hypothesis H0

Question 5

Experimental method

Answer 5

Manipulation of IV to see effects on DV
Extraneous variables can be controlled
Causality can be inferred

Question 6

Criteria for causality

Answer 6

1. Contiguity: cause (IV) precedes effect (DV)
2. Correlation - must co-occur
3. Absence of a tertium quid or third factor or confound that might affect results

Question 7

Quasi-experimental method

Answer 7

There is no real manipulation of IV - cannot randomly assign participants to various conditions
More extraneous variables than experimental
Contiguity: yes

Question 8

Correlational method

Answer 8

How variables behave naturally - no manipulation
Simultaneous measurement
No inference of causation
Extraneous variables are present

Question 9

Independent variable

Answer 9

Variable that is manipulated by experimenter
Thought to be the cause of some effect
Exp -> one or more IV -> each 2 or more LEVELS or conditions

Question 10

Dependent variable

Answer 10

What is measured (see levels of measurement)
Thought to be affected by changes in IV
Exp -> one or more DV

Question 11

Levels of a variable

Answer 11

Different conditions of IV

Question 12

Levels of measurement

Answer 12

Categorical
- Nominal (includes binary variables)
- Ordinal
Continuous
- Interval
- Ratio

Question 13

Nominal scale

Answer 13

Categories
Where numbers assigned are just labels - no meaning
eg: gender, marital status

Question 14

Ordinal scale

Answer 14

Categories ordered according to a criterion

eg: Mineral hardness scale, position in a race

Question 15

Interval scale

Answer 15

Continuous

Equal differences in scale represent equal differences in measure; No meaningful zero or nothing (°C)

Question 16

Ratio scale

Answer 16

Continuous
Proportional intervals between values
0 or origin = absence of attribute

Question 17

Extraneous variables

Answer 17

Aka nuisance variable

Any factor other than IV that may impact results and cause variation in DV

Question 18

Confounding variables

Answer 18

Type of extraneous variable
Some aspect of experiment SYSTEMATICALLY varies with the IV
Experimental design should eliminate potential confounds

Question 19

Between subjects design

Answer 19

Independent/Between groups design
Different conditions -> Different groups of subjects
Sometimes the only option (for gender)
Random assignation to minimise individual differences

Question 20

Within subjects design

Answer 20

Aka Repeated Measures Design
All subjects perform all conditions
Individual differences are controlled

Question 21

Potential confounds for within subject design

Answer 21

Order effects
Practice effects
Fatigue effects

Question 22

How to minimise order effects

Answer 22

Counterbalancing blocks of trials - randomize order of presentation of conditions

Question 23

Matched subjects

Answer 23

Recruit samples in pairs matched to minimize individual differences
Statistically treated like within subjects - one pair = one record/individual tested on 2 different levels

Question 24

Correlational design

Answer 24

When IV cannot be manipulated (ethics, time limitation)

Measure pre-existing variables to see how co-related or co-varying

Question 25

Alternative hypothesis

Answer 25

Aka Experimental Hypothesis (H1)

Hypothesis formed from the theory - presence of an effect or differences between the two conditions

Question 26

Null hypothesis

Answer 26

H0
No effect or differences between two conditions - samples come from the same population
The baseline
Default until strong evidence to the contrary

Question 27

When do we reject H0

Answer 27

When probability of observing effects if null hypothesis were true (p)
is less than pre-set criterion (α)

Question 28

p value

Answer 28

How likely it is (probability) that we would observe a test statistic as big (or as large an effect) if the null hypothesis was true
Determined from the test statistic

Question 29

α Alpha value

Answer 29

Criterion level that we set - p has to be less than this value for us to think effects were not due to chance
i.e, max accepted level of type I error/ false positive
set as 0.05 or 5%

Question 30

p-hacking

Answer 30

Selectively reporting only significant p values

like trying multiple analyses and reporting only the ones that yield significant results

Question 31

Statistical significance

Answer 31

Unlikely to occur by chance - reliable
p<0.05 Statistically significant finding
p>α or p>0.05 -> NON Significant finding = failed to reject the null hypothesis

Question 32

Two types of errors

Answer 32

Type I (α) - False positive
Type II (β) - False negative

Question 33

Type I error

Answer 33

alpha
We believe/claim there is an effect when there is none
Accepting alternate hypothesis in error
Could be due to confounds
(FP)

Question 34

Type II error

Answer 34

beta
We believe/claim there is no effect when, in reality, there was.
Accept null hypothesis in error
Possible reasons: noise, small effect size, low power of test
(FN)

Question 35

Power of a statistical test

Answer 35

Probability that the test will find an effect when one exists