Suggested answer to exercise: Readmission to hospital for asthmatic children using Stata 9, 5

Question 5: Which two of these predictor variables would you expect to be related to one another, whatever the disease in question?

Suggested answer

I would expect age and number of previous admissions to be related. The older the subject is, the more opportunity there will have been for previous admissions.

In fact, they appear to be related, as the scatter diagram shows:

 
. scatter admiss age

There should be 1034 points on this graph and clearly some are superimposed. We can add a bit of random noise with the jitter, to shake them up a bit:

. scatter admiss age, jitter(5)

This graph is just to guide me in my analysis. I would not use it for publication, so I have not improved it.

There certainly appears to be a relationship, but the data do not look suitable for regression or correlation. Because the number of previous admissions is highly skew, I chose Kendall's rank correlation coefficient, tau b:

. ktau  age admiss

  Number of obs =    1029
Kendall's tau-a =       0.1680
Kendall's tau-b =       0.2032
Kendall's score =   88860
    SE of score =   10496.947   (corrected for ties)

Test of Ho: age and admiss are independent
     Prob > |z| =       0.0000  (continuity corrected)

Hence tau b = 0.20, P<0.001. We have strong evidence that older children tend to have more previous admissions.

Back to Exercise: Readmission to hospital for asthmatic children.

To Biostatistics in Research Practice index.

To Martin Bland's M.Sc. index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 6 February, 2008.