SPSS outputs

Question 1

When asked to look at a scatterplot what are some things which can be mentioned?

Answer 1

• Can a linear relationship be seen? (important in deciding the later stats test)
• is this relationship positive or negative?

Question 2

How to write up a Pearson’s correlation?

Answer 2

We test H0: no association between A and B from the table we see (n= C)
10%|5%|1%
D | E | F
we see P < or > 0.05/1 and so reject/ fail to reject H0 at the 5/1% level.
we have strong/no evidence of an association between A and B

Question 3

What are the assumptions made by Pearson’s?

Answer 3

• the data is an interval scale
• the population is normal
• there is a non-linear trend

Question 4

what are the assumptions of a simple regression?

Answer 4

• the data is linear
• the errors are:
  –> independent
  –> normally distributed
  –> common variance
• residuals should show a random scatter and normal distribution of values
• no obvious pattern to data

Question 5

How would you go about explaining the output of a multiple regression from SPSS?

Answer 5

• have correlation table and regression table
• correlation table is helpful to understand whether there is a relationship between any of the variables helping to decide whether they are useful to predict one another
• is there a significant correlation?
• is this positive or negative?
• in the second table the R squared column is important this shows the variability of Y explained by X1 and X2

Question 6

the write up of a multiple regression

Answer 6

There are strong/weak positive/negative correlations of A and B between Y and X1 and X2 respectively. Both are significant/unsignificant, at C level. There is also a strong/weak positive/negative correlation between X1 and X2 this is significant/unsignificant.
R2 = D so E% of the variation of Y can be explained by X1 and X2

Question 7

How to write up the regression equation

Answer 7

• uses the next table along
  (all numbers from b column)
• Y=(the constant row)+(FxX1)+(G x X2)
• From both X1 and X2 p </> H and so both variables are significant/not significant b = or fancy no equals 0. Both should/shouldnt be kept in model

Question 8

should a normality test be significant?

Answer 8

no

Question 9

how would you write up causal steps mediation

Answer 9

3 output therefore 3 conclusions
- write down regression equation for each
- at what level are they significant?
- is this strong evidence of a prediction
the order theses will occur
- X to Y
- X to M
- X to M to Y
need last two to be significant to say the process is mediated
- X to Y doesnt have to be significant
X to M to Y equation bit diff
Y = constant + C x X + D x M
- remeber variance of each

Question 10

how would you write up the output of a confidence interval?

Answer 10

Output 1
- Rsq = varitation
- coeffients - used to write equation
- P of X - probability
- M using X
- write regression equation
output 2
- this is the output of Y on X and M
- Y = constant + E x X + f x M
Output 3
- total effect = simple regression of X on Y is this significant
- direct = multiple regression of X and M on Y significant?
- indirect = X on Y via M this is estimated use Confidence intervals (final two columns) does contain 0?

Question 11

How to write up a moderation analysis?

Answer 11

• regression equation (same as before)
• for each term we test H0: b =0, H1: b (not equal sign) both are