Lecture 14- Correlation

Question 1

What is the difference between a controlled experimental study and a correlation study?

Answer 1

In controlled experimental study the experimenter directly manipulates an IV to have an effect on the DV, therefore causality can be inferred. This is not the case for correlation study where we rely on pre-existing associations between variables.

Question 2

Where is it common to do a correlation study?

Answer 2

During exploratory research: come up with the theory (generate new ideas). Can then use experimental research to test this theory later. Also often used when ethics don’t allow us to directly manipulate the characteristics of individuals.

Question 3

What type of graph is used to show correlations?

Answer 3

Scattergram

Question 4

What is Pearson’s ‘r’ correlation coefficient? How do we interpret it?

Answer 4

It quantifies the strength and direction of a correlation. The largest values you can have is 1 or -1 which would be a perfect positive correlation and a perfect negative correlation respectively.

Question 5

Can a correlation work for non-linear relationships?

Answer 5

No, it only describes straight line: therefore doesn’t work for curvilinear relationships.

Question 6

How many variables do correlations involve?

Answer 6

Only 2

Question 7

What is a useful function of correlations?

Answer 7

Being able to predict how changes in one variable will effect the other.

Question 8

How do you differentiate graphically a strong versus weak correlation?

Answer 8

• Strong= all dots are clustered closely to the trend line

- Weak= dots are more spread out with trend less clear/defined

Question 9

What is the difference between a positive and negative correlation?

Answer 9

• Positive= as one variable increases so does the other one

- Negative= as one variable increases the other one decreases

Question 10

What is the top line in the Pearson’s correlation coefficient?
What is the bottom line?

Answer 10

• Top line= Variation in scores associated with apparent linear relationship
• Bottom line= Variation in scores associated with apparent linear relationship plus variation In scores associated with everything else (random).

Question 11

How do we frame the null and alternative hypothesis in terms of R values?

Answer 11

• Null= true R is 0, there is no relationship

- Alternative= True R is not O, there is a relationship

Question 12

In jamovi what type does your variable need to be to perform a correlation?

Answer 12

Switch from nominal to scale (symbol=ruler)

Question 13

What is range restriction in correlation, what is the result?

Answer 13

When scores are clustered in one are due to only taking a limit number/range of measurements. This makes it hard to accurately determine the overall trend. For example, with the correlation PSYC210 homework scores and exam grades because you can’t get higher than 10 in homework mark and 100 in exam the top of the correlation is flat, the relationship would have been portrayed a lot stronger if could extend further.

Question 14

What is correlation not…. Why?

Answer 14

• Causation

- Could be a third factor influencing both (confounder)