re

Family.--The term "family," as used here, fers to a group of two persons or more related by blood, marriage, or adoption and residing together; all such persons are considered as members of one family. A family may comprise persons in either a household or a quasi household. A lodger and his wife who are not related to the head of the household, or a resident employee and his wife living in, are considered as a separate family and not as part of the head's family. Thus, a household may contain more than one family. However, if the son of the head of the household and the son's wife are members of the household, they are treated as part of the head's family. A household head living alone, or with unrelated persons only, is regarded as a household but not as a family. Thus, some households do not contain a family.

Married couple.--A married couple, as defined for census purposes, is a husband and his wife enumerated as members of the same household or quasi household. The married couple may or may not have children living with them. The expression "husbandwife" before the term "household," "family," or "subfamily" indicates that the head of the household, family, or subfamily is a married man whose wife lives with him. For example, a husband-wife family is a family with a head who is "married, wife present."

Parent-child group.--A parent-child group is a parent and one or more own sons or daughters under 18 years old living together. The most common example is a widow and her child sharing the same living quarters. The expression "parent-child" before the term "household," "family," or "subfamily" indicates that the head of the household, family, or subfamily is a parent who is not living with his (her) spouse.

Unrelated individuals.--Unrelated individuals are persons (other than inmates of institutions) who are not living with any relatives. An unrelated individual may be (a) a household head living alone or with nonrelatives only, (b) a lodger or resident employee with no relatives in the household, or (c) a quasi-household member who has no relatives living with him. Thus, a widow who occupies her house alone or with one or more other persons not related to her, a roomer not related to anyone else in the dwelling unit, a maid living as a member of her employer's household but with no relatives in the household, and a resident staff member in a hospital living apart from any relatives are all examples of unrelated individuals.

Primary individual.--A primary individual is a household head living alone or with nonrelatives only.

living quarters who are related

to the head of the family by blood, marriage, or adoption. "Size of subfamily" includes the head of the subfamily, his wife (if any), and their sons and daughters under 18 years of age. If a primary family has a subfamily among its members, the size of the primary family includes, the members of the subfamily.

Related persons and family members.--In the classification of households by number of related persons, the head of the household and all persons in the household related to the head are included. In the classification of families by number of fam1ly members, the head of the family and all other persons in the family are included. The number of family members is the same as the size of the family.

Own children and related children.--"Own" children in a family are sons and daughters, including stepchildren and adopted children, of the family head. Similarly, "own" children in a subfamily are sons and daughters of the subfamily head. "Related" children in a family include own children and all other children in the household who are related to the family head by blood, marriage, or adoption. All children shown as members of subfamilies are own children of the subfamily head. For each type of family unit the count of children is limited to single (unmarried) children.

SOURCE AND RELIABILITY OF THE ESTIMATES

Source of data.--The estimates presented in this report are based on data obtained in connection with the monthly population sample survey of the Bureau of the Census. The statistics for 1956 are based on a sample design instituted in January 1954. This sample is spread over 230 sample areas comprising 453 counties and independent cities. A total of 24,000 to 26,000 dwelling units and other living quarters are designated for the sample at any time, and completed interviews are obtained each month from about 20,000 to 22,000 households. meinder, about 500 to 1,000 are households for which information should be obtained but is not (1.e., households temporarily absent, households living in areas with impassable roads, etc.), and the rest are vacant dwellings or those otherwise not to be enumerated for the survey. The data for 1947, for which similar information was obtained, are based on a different sample, which consisted of about the same number of units but which covered only 68 sample areas.

Of the re

The estimating procedure used in this survey involved the inflation of weighted sample results to independent estimates of the civilian noninstitutional population of the United States by age, color, and sex. These independent estimates were based on statistics from the 1950 Census of Population; statistics of births, deaths, immigration, and emigration; and statistics on the strength of the Armed Forces. To these totals were added the population in the Armed Forces living off post or with their families on post and the institutional population in 1950. For the years prior to 1953, the independent estimates of the population were based on the 1940 Census data brought forward to the survey month to take account of births, deaths, net immigration, and aging of the population. The introduction of the new population levels should improve the accuracy of the statistics, but it creates some discontinuities in the series. For example, the estimated number of widowed males for 1950 on the new basis is 225,000 greater than on the former basis; for widowed females the difference is 175,000.

Reliability of the estimates.--Since the estimates, except the independent estimates and complete census data, are based on sample data, they are subject to sampling variability. The following illustration indicates the order of magnitude of the sampling errors for some typical statistics in March 1956. At that time there were an estimated 8,365,000 households in the United States with female heads. This number was 17.1 percent of all households in the United States. The standard error of the estimate of 8,365,000 is about 140,000; the standard error of the estimate of 17.1 percent is roughly 0.3 percentage points. about 68 out of 100 that the sample differ from the results tained from a complete census by less than the standard errors indicated for the above items. The chances are about 95 out of 100 that the differences would be less than twice the specified sampling errors and about 99 out of 100 that they would be less than 2 times the errors indicated.

The chances are estimates from the which would be ob

Caution should be exercised in considering trends or comparisons of these data from one year to the next; comparisons over relatively longer periods will provide more reliable evidence of change. Estimates of annual change are sensitive to small errors in the estimates of the total number of units in a category. The joint effect of sampling variability and any biases in response or coverage on the total number of persons in a given category for each year may, therefore, be large in relation to the amount of annual change.