Table A-12. Measures of Central Tendency and Variability for Year-to-Year Percent Changes in National Health Expenditures by Type of Expenditure, Current Dollars, 1966-19811.
| Type of National Health Expenditure | Measures of Variability2 | ||||
|---|---|---|---|---|---|
|
|
|||||
| Measures of Central Tendency | Standard Deviation | Standard Error | Coefficient of Variation of The Mean3 | ||
|
| |||||
| Mean | Median | ||||
|
|
|
|
|
|
|
| Total | 12.8% | 12.8% | 1.5% | 0.377% | 0.029 |
| Personal Health Care (Total Systems Cost) |
13.1 | 13.0 | 1.8 | 0.467 | 0.036 |
| Hospital Care | 14.3 | 14.7 | 2.0 | 0.517 | 0.036 |
| Community Hospital Care | 15.8 | 16.6 | 2.8 | 0.716 | 0.045 |
| Inpatient | 15.5 | 16.2 | 2.8 | 0.732 | 0.047 |
| Outpatient | 19.0 | 19.3 | 3.6 | 0.919 | 0.048 |
| Federal Hospital Care | 11.1 | 11.8 | 3.2 | 0.819 | 0.074 |
| Physicians' Services | 12.4 | 11.9 | 2.9 | 0.742 | 0.060 |
| Dentists' Services | 12.1 | 12.5 | 2.8 | 0.725 | 0.060 |
| Other Professional Services | 12.2 | 12.6 | 5.1 | 1.309 | 0.108 |
| Drugs and Medical Sundries | 9.3 | 9.2 | 2.1 | 0.537 | 0.058 |
| Eyeglasses and Appliances | 10.5 | 11.9 | 5.1 | 1.317 | 0.125 |
| Nursing-Home Care | 16.7 | 16.5 | 3.3 | 0.862 | 0.052 |
| Nursing-Home Care Excluding ICF-MR | 15.9 | 15.1 | 3.5 | 0.905 | 0.057 |
| Other Health Services | 12.3 | 12.1 | 6.2 | 1.606 | 0.130 |
| Program Administration and Net Cost of Insurance | 13.8 | 14.6 | 15.3 | 3.949 | 0.286 |
| Government Public Health Activities | 14.8 | 15.5 | 6.4 | 1.648 | 0.111 |
| Research | 8.6 | 7.6 | 4.1 | 1.053 | 0.122 |
| Construction of Medical Facilities | 8.9 | 7.6 | 7.8 | 2.014 | 0.227 |
There are 16 annual percent changes for the period 1966-1981.
ln textbook examples, measures of variability typically measure sampling variability, that is, variations that might occur by chance because a sample of the population is surveyed. As calculated in this paper (and for typical applied time-series analyses) measures of variability also reflect variability associated with evolving causal structures and variability associated with various types of nonsampling errors such as data processing mistakes, nonresponse, misreporting by respondents, etc. The calculated measures are approximate and are meant as a general guide. It is important to keep in mind the potential dangers of extrapolating historical measures of variability into the future. That is, there can be no guarantee that future variability will replicate historical variability.
For cautions in using the coefficient of variation when the mean of the variable measures change, see Kish (1965, pp. 47-49).