Interpreting data: Boxplots and tables
Interpreting data: Boxplots and tables

Start this free course now. Just create an account and sign in. Enrol and complete the course for a free statement of participation or digital badge if available.

2.5 Early retirement from the National Health Service

Example 2.2: Early retirement from the National Health Service

A study was carried out to investigate various aspects of early retirement from the British National Health Service (NHS). In 1998–99, 5469 NHS employees from England and Wales were granted early retirement because of ill health. The researchers examined the records of a sample of 1994 of these people. Table 2.8 gives data on these people, classifying each of them by occupational group and by a broad classification of the health reason for which they retired.

Table 2.8 Retirements from the NHS because of ill health, 1998–99

Occupational groupReason for retiring because of ill health
Ambulance workers651261295
Healthcare assistants or support3396177117594
Nurses or midwives36414470153731
Technical or professional staff422542394
Administration or estates staff118943166309
Doctors or surgeons33402028121
This table is adapted from Pattani, S., Constantinovici, N., and Williams, S. (2001) Who retires early from the NHS because of ill health and what does it cost? A national cross sectional study. British Medical Journal, 322, 208–209.

Activity 5 Early retirement from the National Health Service

Suppose that the main interest of the researchers was to see whether (and, if so, how) the pattern of causes of retirement differed between occupational groups. How does the table, as it stands, match up to the guidelines given at the start of this section?



The labelling of rows and columns is reasonably clear as it stands (guideline 1). Assuming that the researchers were interested separately in all these occupational groups and all these reasons for retirement, there seem to be no good reasons for breaking up the table or combining cells (guideline 2). The numbers in the table are counts and not particularly large ones (three digits at most, apart from the overall total) and there seems no reason to simplify them (guideline 3).

However, it might help to include some calculation results (guideline 4). As the table stands, it is reasonably easy to see that (for instance) in each occupational group, the greatest number of retirements was due to musculoskeletal reasons, but it is not easy to compare just how much bigger that count is relative to the others in each occupational group, because the total number of retirements differs considerably from one occupational group to another. This sort of comparison would be more straightforward if we knew, for instance, the proportion or percentage of people in each occupational group who retired for musculoskeletal reasons.

Activity 6 Early retirement from the National Health Service: percentages

  • (a) For each occupational group, calculate the percentage of people who retired because of each cause of ill health. Use these percentages to comment on the different patterns of causes of retirement in the different occupational groups.

  • (b) Assuming you found the percentages useful for making these comparisons, say whether you think that a table presenting this information should include only the counts (as in Table 2.8), only the percentages that you calculated, or both.



(a) The percentage of people in each occupational group who retired because of each cause of ill health is given in Table 2.9.

Table 2.9 Retirements from the NHS because of ill health, 1998–99

Occupational groupReason for retiring because of ill health (% of row total)
Ambulance workers6813613
Healthcare assistants or support57101320
Nurses or midwives50201021
Technical or professional staff452742
Administration or estates staff38301021
Doctors or surgeons27331723

For instance, the proportion of the ambulance workers who retired because of ill health for musculoskeletal causes is 65/95=0.68421, or 68.421%. However, there is no need to include three decimal places to portray the patterns in the data clearly. Whole percentages are sufficiently accurate; so this percentage has been entered in the table as 68%. The other percentages were calculated in a similar way.

(Note that, because of the rounding of the percentages, the sums of some of the rows are 99% or 101% rather than 100%. In the context of communicating the general pattern of the data, this does not matter.)

Perhaps the most obvious difference between the occupational groups is that the percentage of retirements for musculoskeletal causes was considerably greater in the first two groups than in some of the others, particularly administrators and doctors. The authors of the paper from which these data are taken attribute this difference to the greater amount of manual work done by workers in the first two categories. The occupational groups with relatively low levels of retirement for musculoskeletal causes also had relatively high percentages of retirements for psychiatric causes. Without further investigation, and in particular without having looked at what proportion of workers in each of these groups actually retired on grounds of ill health (rather than continuing to work), it is difficult to say more about the reasons for these patterns.

(b) The question of whether to include the percentages in a table as well as, or instead of, the counts does not have a clear-cut answer. The table given in the paper from which these data were obtained includes both. This makes the table rather complicated, and the patterns of different causes of retirement is not entirely clear at a glance. However, in interpreting the data it is important to know that the number of ill-health retirements in some of these groups was not particularly large. A useful compromise would have been to include the total number of retirements from which the percentages in each row were calculated, as in Table 2.10. In general, when calculating row percentages (or column percentages) in a table in this way, it is good practice to include the totals that were used to calculate the percentages as well.

Table 2.10 Retirements from the NHS because of ill health, 1998–99

Occupational groupReason for retiring because of ill health (%of row total)
MusculoskeletalPsychiatricCardiovascularOtherTotal (=100%)
Ambulance workers681361395
Healthcare assistants or support57101320594
Nurses or midwives50201021731
Technical or professional staff45274249
Administration or estates staff38301021309
Doctors or surgeons27331723121

Take your learning further

Making the decision to study can be a big step, which is why you'll want a trusted University. The Open University has 50 years’ experience delivering flexible learning and 170,000 students are studying with us right now. Take a look at all Open University courses.

If you are new to University-level study, we offer two introductory routes to our qualifications. You could either choose to start with an Access module, or a module which allows you to count your previous learning towards an Open University qualification. Read our guide on Where to take your learning next for more information.

Not ready for formal University study? Then browse over 1000 free courses on OpenLearn and sign up to our newsletter to hear about new free courses as they are released.

Every year, thousands of students decide to study with The Open University. With over 120 qualifications, we’ve got the right course for you.

Request an Open University prospectus371