Lecture 15- Simple Regression

Question 1

When do you carry out a two way anova?

Answer 1

When you have multiple independent variables and therefore want to determine interaction effects as well as individual effects (tells you more info)

Question 2

How would you interpret the graph on lecture slide 5?

Answer 2

Looks like there is an interaction between the degree type and gender.

Within science degrees males have strongest bias to females in science. Within arts degree females have strongest bias to females in science.

Question 3

What does a p value of less than 0.5 indicate?

Answer 3

A real effect, not chance

Question 4

In a simple regression what 2 variables do we have? What are they typically represented by in a graph?

Answer 4

• 1 to be predicted variable (y)

- 1 predictor variable (x)

Question 5

What difference in situation describes when you would use a ANOVA as opposed to regression?

Answer 5

• Regression uses numerical variables

- In ANOVA variables are categories not numerical measures

Question 6

What types of relationships can not be described by linear regression or correlations?

Answer 6

-Non-linear graphs (curvilinear relationships)

Question 7

Where should a line of regression be in terms of the distribution of data?

Answer 7

In the middle i.e. equal number of points above and below

Question 8

What two things do you need to know to form an equation/ numerical description of a regression line?

Answer 8

• Y intercept
• Gradient

(think y=mx+c where y is the ‘to be predicted’ variable and x is the ‘predictor’ variable)

Question 9

In jamovi what is the label of the table that gives the values needed to form an equation for a regression line?

Answer 9

Model coefficients, only look at first value (estimate)

Question 10

What effect can an outlier have on a regression line? What is another factor that can result in this?

Answer 10

• Drag it either up or down so that it is no longer representative (not half way through data points).
• Restricted range

Question 11

What tells you how good our regression line is for estimating data?

Answer 11

P value at significance level of 0.05 (usually), this given in top table in jamovi

Question 12

Can regression be used to determine cause and effect?

Answer 12

No, random selection and assignment wasn’t used so it cannot be used to determine cause and effect. Correlation isn’t causation.