Linear Regression Flashcards

1
Q

What is the main purpose of linear regression?

A

Causal associations.

Focuses on a mean group comparison.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the requirements for linear regression?

A

Outcome variable must be continuous.
Predictor variable can be anything.
Outcome should be approximately normally distributed.
Should be testing the linear relationship between two variables.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What does the ordinary least squares regression minimise?

A

The sum of the squared residuals.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is Bo in the equation?

A

The y-intercept.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is Ei in the equation?

A

The residual error (individual deviation).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is B1 in the equation?

A

The gradient.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What method should you use when conducting a LR in SPSS?

A

Forced-entry/enter.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Which assumptions must you check?

A

Normally distributed errors.
Homoscedasticity.
Linearity.
Influential cases.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is R-squared?

A

It is the amount of variance of the outcome variable which can be explained by the predictor.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is the F-value?

A

There ratio of the model MS divided by the residual MS.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the cut-off point for the F-value?

A

Has to be larger than 1.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

How do you check for normal distributed errors?

A

Histogram + P-P plot.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How do you check for homoscedasticity?

A

zResidual vs zPredicted plot.

Variance around the regression line should be the same for all values of the predictor variable.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

How do you check for linearity?

A

zResidual vs zPredicted plot.
Must add Loess line!
Line should be approximate to zero and linear.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

How do you check for influential cases?

A

Cook’s distance.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is Cook’s cut-off point?

A

Must be smaller than 1.

17
Q

What is an influential case?

A

A case that significantly alters the value of a regression coefficient when it is deleted from the analysis.

18
Q

In the SPSS output table, what is the intercept?

A

The unstandardised B (constant) value.

19
Q

In the SPSS output table, what is the gradient?

A

The unstandardised B (predictor) value.

20
Q

How do you report unstandardised results?

A

Results show that (predictor) is …. associated with (outcome) (B = … , p = …)) which supports our hypothesis.

21
Q

How do you report standardised results?

A

Results show that one unit higher in (predictor) is significantly associated with an … of … (outcome) (p = …) which supports our hypothesis.

22
Q

Which number should not be between the LCI and UCI?

A

0.

This means results could be non-significant as the regression weight in the population could be 0.

23
Q

Name one issue with linear regression?

A

Reverse causality.

24
Q

What can you do if the outcome variable is not approximately normally distributed?

A

Bootstrap.

25
Q

When is the F-value not useful?

A

When the predictor is non-significant.