Why is determining the specific purpose of this speech such a vital step in the process of preparing it?

Page 2

figures are first calculated for each calendar quarter by averaging estimates for all weeks of interviewing in the quarter. Prevalence data for a year are then obtained by averaging the four quarterly figures.

For other types of statistics-namely those measuring the number of occurrences during a specified time period-such as incidence of acute conditions, number of disability days, or number of visits to a doctor or dentist, a similar computational procedure is used, but the statistics are interpreted differently. For these items, the questionnaire asks for the respondent's experience over the 2 calendar weeks prior to the week of interview. In such instances the estimated quarterly total for the statistic is 6.5 times the average 2-week estimate produced by the 13 successive samples taken during the period. The annual total is the sum of the four quarters. Thus the experience of persons interviewed during a year-experience which actually occurred for each person in a 2-calendar-week interval prior to week of interview-is treated as though it measured the total of such experience during the year. Such interpretation leads to no significant bias.

been published as well as a detailed description of the sample design and a report on the estimation procedure and the method used to calculate sampling

5 errors of estimates derived from the survey.

Collection of Data.-Field operations for the survey are performed by the U.S. Bureau of the Census under specifications established by the National Center for Health Statistics. In accordance with these specifications the Bureau of the Census participates in survey planning, selects the sample, and conducts the field interviewing as an agent of NCHS. The data are coded, edited, and tabulated by NCHS.

Estimating procedures.-Since the design of the HIS is a complex multistage probability sample, it is necessary to use complex procedures in the derivation of estimates. Four basic operations are involved:

Inflation by the reciprocal of the probability of
selection. The probability of selection is the prod-
uct of the probabilities of selection from each step
of selection in the design: PSU, segment, and house- hold.

Nonresponse adjustment. The estimates are in-

Poststratification by age-sex-color. –The esti-

The effect of the ratio-estimating process is to make the sample more closely representative of the civilian, noninstitutional population by age, sex, color, and residence, which thereby reduces sampling variance.

As noted, each week's sample represents the population living during that week and characteristics of :he population. Consolidation of samples over a time period, e.g., a calendar quarter, produces estimates of average characteristics of the U.S. population for :he calendar quarter. Similarly, population data for a rear are averages of the four quarterly figures.

For prevalence statistics, such as number of persons with speech impairments or number of persons :lassified by time interval since last physician visit,

Nonresponse. — Data were adjusted for nonresponse by a procedure which imputes to persons in a household which was not interviewed the characteristics of persons in households in the same segment which were interviewed. The total noninterview rate was about 5 percent-1 percent was refusal, and the remainder was primarily due to the failure to find an eligible respondent at home after repeated calls.

The interview process. The statistics presented in this report are based on replies obtained in interviews of persons in the sampled households. Each person 19 years of age and over present at the time of interview was interviewed individually. For children and for adults not present in the home at the time of the interview, the information was obtained from a related household member such as a spouse or the mother of a child.

There are limitations to the accuracy of diagnostic and other information collected in household interviews. For diagnostic information, the household respondent can usually pass on to the interviewer only the information the physician has given to the family. For conditions not medically attended, diagnostic information is often no more than a description of symptoms. However, other facts, such as the number of disability days caused by the condition, can be obtained more accurately from household members than from any other source since only the persons concerned are in a position to report this information.

Rounding of numbers. The original tabulations on which the data in this report are based show all estimates to the nearest whole unit. All consolidations were made from the original tabulations using the estimates to the nearest unit. In the final published tables, the figures are rounded to the nearest thousand, although these are not necessarily accurate to that detail. Devised statistics, such as rates and percent distributions, are computed after the estimates on which these are based have been rounded to the nearest thousand.