Week 6-8 Content

Question 1

Correlational designs

Answer 1

We could just measure the IV and DV, and see if there’s an association between them. This is called a correlational study.

The higher someone’s score on one variable, the higher their score on another variable.

The higher someone’s score on one variable, the lower their score on another variable.

The two variables have no association (null hypothesis).

“Experience of Hotter Temperatures are associated with increased belief in climate change”

There are variations that use nominal variables, but these are less common.

Question 2

Causation from Correlation

Answer 2

We can’t draw any conclusions about causation from a correlational study.

Direction of causation
Maybe people who are more worried about climate change move to colder regions – climate change belief causes the temperature of that person’s environment to go up or down, rather than vice versa

“Third variable” problem
Maybe there’s different education priorities in higher latitudes, so people are learning different things about climate change
Maybe the weather fluctuates between extremes more in higher latitudes
Either way the distance is not necessarily the cause of the belief

Question 3

IV/DV in correlational designs

Answer 3

No specific variables

Question 4

What are correlations looking at?

Answer 4

Rather than comparing groups, we’re looking at a collection of individual cases and seeing if there’s an overall trend.

Question 5

Experimental designs

Answer 5

We often talk about groups or conditions in experimental designs – these are levels of the IV.

Control conditions are special conditions where the IV is pretty much left alone

Zero-dosage conditions, placebo conditions, etc.
In experimental conditions, we set the levels of the IV.
We can then use statistical methods to compare conditions and see if the IV has made a difference

Question 6

Critical elements of a true experiment

Answer 6

Manipulation of the IV (directly changing it, rather than just observing it: we place participants in hot or cold rooms)

Random assignment to conditions (participants don’t get to choose which room they go into)

Controlling possible confounding variables (things that could affect the relationship between IV/DV)

Question 7

Quasi-experimental designs

Answer 7

These are like experiments in that they compare two or more groups of participants to one another.

However, they use pre-existing groups. Since there is no random assignment (though there may be manipulation), causation can’t be inferred.

Question 8

Observational studies

Answer 8

These are pretty much correlational or quasi-experimental studies where the researcher monitors participants’ behaviour.

Participant observation: the researcher takes an active part in the goings-on

Covert observation: the participants don’t know that the researcher is there

Structured observation: the researcher has a coding scheme, like a checklist of behaviours

Common in developmental psychology

Question 9

Between-Subjects

Answer 9

participants are assigned to different conditions

Question 10

double-blind study design

Answer 10

The participant does not know whether they are in the experimental or control group

The experimenter also does not know whether the participant is in the experimental or control group

Question 11

Type I and II Errors

Answer 11

Observe a statistically significant result when there is NO real relationship between our variables to be found – known as a Type I error

Not observe a statistically significant result when there IS a real relationship between our variables to be found – known as a Type II error

Question 12

Z-Scores

Answer 12

A Z-score is a standardized score

It tells us how a certain observation (i.e. participant) scores relative to others in our sample.

The score is expressed in standard deviations.

A positive Z-score means the raw score is above the mean of the sample

A negative Z-score means the raw score is below the mean of the sample

If you assume a normal distribution, you can determine what percentage of observations are above or below a certain observation

Question 13

Different types of correlations

Answer 13

Correlations in statistics are denoted with r

r is short for Pearson’s Correlation Coefficient

r can range between -1 and 1

Question 14

Positive correlation:

Answer 14

When you go up on one variable, you go up on the other variable

Question 15

Negative correlation:

Answer 15

When you go up on one variable, you go down on the other variable

Question 16

No correlation:

Answer 16

One variable gives no information on the score on an other variable

Question 17

Testing a correlation

Answer 17

Simple rules:
If p < .05, reject the null hypothesis, and consider the relationship between your variables to be statistically significant
If p = .05 or p > .05, retain the null hypothesis – you haven’t found evidence for a relationship

Question 18

Each of these assumptions must be satisfied in order to use Pearson r

Answer 18

Pearson r is not appropriate to use for every data set – it makes certain assumptions about the data

The association is linear (straight line), not curvilinear (curved line)

The variables are measured at the interval or ratio level

No outliers

Question 19

Linearity

Answer 19

Pearson r is for linear associations only– that go in a straight line
Some relationships are curvilinear. Pearson r cannot detect these, even if obvious!

Question 20

Outliers

Answer 20

Outliers are a particular problem in correlation – an extreme outlier can mask a genuine relationship, or create one where none really exists

Question 21

Other Measurements

Answer 21

The most common “non-parametric” equivalent of Pearson’s r is Spearman’s rho (ρ)
This transforms the data into ranks, then correlates the ranks with one another
Same rules for effect size apply