# 2.8 Using SPSS to Perform a Simple Linear Regression Part 2 - Interpreting the Output

We've been given a quite a lot of output but don’t feel overwhelmed: picking out the important statistics and interpreting their meaning is much easier than it may appear at first (you can follow this on our video demonstration ). The first couple of tables (
The The next three tables (
The The The
The table below (
It also provides standardised versions of both of these summaries. You will also note that you have a new variable in your data set: We requested some graphs from the
We have also generated a P-P plot to check that our residuals are normally distributed (
As you will see, when using real world data such imperfections are to be expected! There are some issues with the distribution of the residuals and we may need to make a judgement call about whether to remove outliers or alter our model to fix this problem. In this case we decided that the benefits of keeping all cases (including the students with low grades!) in our sample outweighed the issues regarding the ambiguous interpretation of whether or not our residuals were normally distributed. Finally, the scatterplot (
Generally there is a problem with a large range of scores at the lowest end of the predicted value (X-axis). This is due to 'floor' effects in the data. It is often worth considering identifying outlying cases and maybe removing them from the analysis but in this example such cases are too interesting to remove! Despite our floor effect, overall the residuals at each predicted value do not appear to vary differently with the exception of a few outliers so it looks as if we have met the assumption. Note that scatterplots can look like giant ink blobs when datasets are as large as this one and this can make interpreting them tricky. SPSS/PASW has a facility called ‘binning’ that is not as rubbish as it sounds (sorry) and can help us here. We discuss binning in
Overall our regression model provides us with a good method of predicting age 14 exam scores by using age 11 scores. We could report this in the following way:
We highly recommend you check the style guide for your university or target audience before writing up. Different institutions work under different criteria and often require very specific styles and formatting. Note that it is important not to simply cut and paste SPSS output into your report - it looks untidy and, as you know, it is full of unnecessary detail! Perhaps you have been introduced to a few too many new ideas in this module and you need a little lie down. Rest assured that if you can grasp the basics of simple linear regression then you are off to a flying start. Our next module on multiple linear regression simply expands on the ideas you have already bravely survived here.
We recommend that you take our quiz and work through our exercises to consolidate your knowledge. Then move on to the next module! |