Calculate your model in SPSS. Does your model fit the data?

Question

The main sections of the output we need to answer this question are given below.

Answer 1

The first box, labelled ‘model fitting information’ provides us with a comparison between
the baseline model with no independent variables, called ‘intercept only’, with the model
with the three predictors which is called ‘final’. A Chi Square test was conducted to look
at improvement in prediction compared to the baseline model. If the test is significant this
indicates that our model fits better than the baseline model with no predictors. As we can
see, this is clearly the case here (sig is less than .05).
More measures of model fit are provided in the next box (Pearson and Deviance). These
two measures compare the actual results for each respondent (i.e. do they agree strongly,
agree somewhat, disagree somewhat or disagree strongly that school is fun) with the
outcome predicted by our model. Again we need to look at the significance level. Unlike
many of the measures we have looked at, in this case we want the difference between the
expected and actual results to be non-significant. This is because if our model fits well,
the observed and expected cell counts should be similar (i.e. respondents have given the
answer we predicted based on our model). This is the case here, which suggests that our
model fits the data.
The Pseudo R Squared statistics are given in the following box. When we look at Cox and
Snell and Nagelkerke’s measures (as in logistic regression), we find a good fit, with
Pseudo R squareds of over .5.