Week 10-Regression

Question 1

What decimals should be used descriptives/inferentials/p values

Answer 1

D=2dp
I(P/T/F values, chi-square, p values etc.,)=2dp
P=2/3dp

Question 2

How do you report inferential stats using chi-squared as an example?

Answer 2

■Chi-square
■Test statistic symbol
■(degrees of freedom) + (sample size, for Chi-square only)
■=test statistic value
■P value
■Effect size
■χ2(1, N = 40) = 6.32, p = .01, φ = .39

Question 3

How do you report correlations differently?

Answer 3

■Test statistic/effect size symbol
■(degrees of freedom)
■=test statistic/effect size value
■P value
■r(123) = .49, p = .03

Question 4

What are regressions?

Answer 4

■Conceptually similar to correlation analyses
■But focuses on ‘predicting’ variance in an outcome sometimes called a criterion or response variable (DV) from predictors (IV)
■Regressions do this by creating a ‘statistical model’
■When we run a regression in SPSS we find out whether the model is a good ‘fit’ for our data, whether there is a significant association and the direction (positive or negative) of this association.
■This information is used to make predictions.
■For this it uses the line of best fit.

Question 5

Linear relationships: what’s the formula of Y = bX + a

Answer 5

–Y : criterion/response variable / DV (salary)
–b : the slope of the line (based on Pearson’s r)
–X : predictor variable / IV (years of experience)
–a : the constant or intercept (starting salary)

Question 6

What does a regression analysis do?

Answer 6

A regression analysis calculates a line of best fit for the observed data which can be used to make predictions for unobserved values.

Question 7

What are Bivariate/Simple Linear Regressions?

Answer 7

■Allows us to summarise and study relationships between the two variables.
■Bivariate (Bi= two, variate= variables).
■We can predict scores on one variable from the scores on a second variable.
■One variable: X, is the predictor variable (explanatory/independent variable)
■The other variable: Y, is the criterion variable (response/outcome/criterion/dependent variable)

Question 8

What’s the line of best fit/regression line and why is it not perfect?

Answer 8

It’s a statistical model AND not perfect as there is error. (How far the observed data is from the model.)

Question 9

What are the assumptions?

Answer 9

■Normally distributed continuous outcome
■Independent Data
■Interval/ratio predictors
■Nominal predictors with two categories (dichotomous)

Question 10

What’s the ANOVA result?

Answer 10

■ This ANOVA result is a measure of ‘model fit’ it tells us how well our statistical model (the regression)
■Whether or not the model is significant.

Question 11

What’s R squared/Adjusted R squared?

Answer 11

–How close the data are to the fitted regression line.
–The proportion of variance explained by the model.
–Presented as a percentage
–Coefficient of determination

Question 12

What’s Beta coefficient β?

Answer 12

Unstandardised B coefficient

Question 13

What’s the Beta value?

Answer 13

Represents predicted change in the DV (salary) for one-unit of change in the IV (YoE)

Question 14

What do we need to understand regression?

Answer 14

–To assess ‘model fit’ – F value
–Know how effective the model is – R squared value
–To know whether an association is significant and the direction- beta value.