Correlation and Regression (wk 5)

Question 1

How do you describe the relationship between two datasets?

Answer 1

-Two datasets (e.g. height, weight) may be related to each other
-When you look at them on a graph, you may see this relationship
-The first statistics invented were for analysing these co-relationships

Question 2

What are the rules for data when describing relationships between two datasets?

Answer 2

-Rules for data
1. Look at data to identify if there are any relationships which can be observed
2. If the data shows a perfectly straight line, it’s probably a mistake
3. If there’s one or more datapoints a long way away from the others, it’s probably a mistake
4. If there’s no relationship at all between things you really expect to be related, it might be worth checking for mistakes
5. Before you do any statistical tests, check your data for mistakes

Question 3

What is a correlation and what is the equation?

Answer 3

-Correlation finds the best fit line by minimising the difference between the data and the line
-Correlation does this through adjusting the line manually to get the best fit;

Question 4

What does the r-value tell you in a correlation?

Answer 4

-The r-value tells you;
1. Which direction is your correlation? -> if r is above zero, correlation is positive and is r is below zero, correlation is negative
2. How strong is your correlation? -> if r is close to one, your correlation is strong and if r is close to zero, your correlation is weak

Question 5

How much does the correlation explain in with the r-value?

Answer 5

-How much does the correlation explain? -> The r-squared value tells you:
1. How much of the variance is explained by your correlation -> If r2 is close to 1, your correlation explains a lot of variance and if r2 is close to 0, your correlation explains only a little variance. Sometimes r2 is called the coefficient of determination.

Question 6

What are the 6 principles of correlation?

Answer 6

• Describes single relationship
• Direction of relationship
• Strength of relationship
• X and Y are inter-changeable
• R, r2
• Does not allow prediction

Question 7

What are the 5 principles of regression:

Answer 7

-Can describe multiple relationships
-Directions of relationships
-Strengths of relationships
-X and Y are NOT inter-changeable
-R, R2, F, t, SE allows prediction

Question 8

How does Jamovi help to interpret outputs with correlations?

Answer 8

-Jamovi allows us to explore multiple relationships in one go, using a correlation matrix

Question 9

What is a multiple regression?

Answer 9

-Multiple regression -> A single outcome variable (y), multiple predictor variables find the best-fitting surface, Residuals are distance from the surface. Predictors can be almost anything (continuous, ordinal, discrete, normally distributed-or not, linear/ non linear).

Question 10

Non-linear relationships in correlations and regressions and how to solve it:

Answer 10

-Non-linear relationships can cause problems in correlation and regression
1. Look at the data
2. Check for mistakes
3. Transform the data -> quadratic, cubic and logarithmic etc
-Correlation doesn’t equal causation