Age–period–cohort models of chronic disease rates. II: graphical approaches by C. Robertson and P. Boyle, Statistics in Medicine, 17, 1325–1340 (1998)
✍ Scribed by Niels Keiding
- Publisher
- John Wiley and Sons
- Year
- 1999
- Tongue
- English
- Weight
- 43 KB
- Volume
- 18
- Category
- Article
- ISSN
- 0277-6715
No coin nor oath required. For personal study only.
✦ Synopsis
In the above paper the authors give a well-balanced set of recommendations concerning graphical analysis of age}period}cohort variation of disease (or other) rates. For the historical record it may be pointed out that the usefulness of graphical approaches to age}period}cohort variations of mortality rates was discussed at length in the German statistical literature of the 1870s by Knapp, Zeuner and particularly Lexis, see Keiding for a survey. It is notable that Lexis, besides the rectangular (cohort, age)-diagram (as opposed to the (period, age)-diagram that we nowadays ascribe to Lexis), also proposed the equilateral diagram reinvented by Weinkam and Sterling more than 100 years later. Lexis even discussed a three-dimensional plot for studying marriage}death models, in our context illness}death models.
A particular recommendation of Robertson and Boyle is that since it is hard to read details o! the pretty &three-dimensional' surface plots, it is necessary to add level curves or image plots (in grey scales or colour).
Examples of this, earlier than the suggestion of Jolley and Giles, were given by Keiding and Keiding et al., who also plotted sections of the surface along the three directions corresponding to constant age, period or cohort.