Calculate your model in SPSS. What does it tell you about the coefficients?

Question

Answer 1

In this box we can see the parameters for the individual variables. The ‘threshold’
statistics refer to the dependent variable, and are less important for us at this stage. The
statistics for the independent variables are given under ‘location’. What we can see here
is that we have statistics not for the variables as a whole, but for each category of the
variable. We can see this most clearly when we look at the variable ‘The teachers think
I’m good at English’ (engsc4). Firstly, we can see that there is one category, 4, which
corresponds to an ‘agree strongly’ response, that does not have a significance level
calculated (see column labelled ‘sig’) and for which the coefficient is 0 (see column
labelled ‘estimates’). This is because this is the reference category, to which we compare
all the others, like we did when we used nominal predictors in multiple linear regression.
We can see here that responding 1 (disagree strongly) is significantly (column labelled
‘sig’) related to responses on the dependent variable. When we look at the estimates, we
can see that the coefficient is -6.96, which means that respondents who disagree strongly
that teachers think they are good at maths are less likely to agree they get good marks in
English than respondents who agree strongly that teachers think they are good at maths
(our reference category 4). Category 2 (this corresponds to a response of disagree
somewhat) is also related significantly to responses on the independent variable. The
coefficient is -5.74, so respondents who disagree somewhat that teachers think they are
good at maths are less likely to agree that they get good marks in maths than respondents
who agree strongly that teachers think they are good at maths, though this is less the case
than it was for respondents who disagreed strongly (as the coefficient here was -6.96).
The third category, which corresponds to agree somewhat is also significant, though the
coefficient, at -2.756 is lower still. Overall we would say that there is therefore a
relationship between ‘the teacher thinks I’m good at maths’ and ‘I get good marks in
maths’, and that the relationship is neatly ordered.
When we look at the other two variables, gender and ‘I like going to school’ (attsc7), we
can again see that one of the categories is the reference category. For gender this is
category 2 (girl), for ‘I like going to school’ this is category 4 (agree strongly). When we
look at the estimates we can see that boys (coefficient -.337) are somewhat less likely to
agree that they get good marks in English. None of the categories for ‘I like going to
school’ are significant.