Back to previous section

MANUFACTURED HOUSING IN NONMETROPOLITAN AREAS:
A DATA REVIEW

© Housing Assistance Council, 1996

Permission is granted ONLY to nonprofit community-based organizations to reproduce and/or adapt this document, and only for their own use.

APPENDIX A:
ABOUT THE AMERICAN HOUSING SURVEY⁵¹

The American Housing Survey is a national sample survey conducted every two years by interviewing occupants of homes. The 1993 national survey (used for this manufactured housing study) included about 55,000 interviews from July through December 1993. Additional surveys are conducted of samples of housing units in 44 metropolitan statistical areas. In both surveys, the samples are made up of housing units enumerated in the 1990 Census and updated to include changes since then. Each home interviewed represents a large number of other homes. 

Interviewers visit the household living in the selected housing unit. Data collected include general housing characteristics; housing costs; and some additions, alterations, and repairs made to the property. Data are collected on income in conjunction with housing costs to estimate percentage of incomes spent on housing. The Bureau of the Census notes that the AHS historically underreports income and overreports poverty in comparison to the Current Population Survey, income tax returns, and national income accounts. Use of telephone interviewing in part of the sample produced lower estimates for housing quality problems and higher estimates for some income and cost items. 

The American Housing Survey records location of housing units so that they may be identified as inside or outside of Metropolitan Statistical Areas (MSAs), urban or rural, or by region. Four regions of the country (Northeast, Midwest, South and West) are the smallest geographic levels at which data is reported. Area definitions used for the 1993 AHS for urban, rural, farm and nonfarm are from the 1980 (not 1990) Census, and for MSAs are from the Office of Management and Budget 1983 determinations.

The numbers reported by the AHS include both sampling and nonsampling errors (such as incomplete data and wrong answers). The probability of sampling errors can be calculated and expressed as ranges, also known as confidence intervals. For example, the survey reports that there are 2,666,000 mobile homes outside of MSAs. There is a 90 percent probability that the actual number is between 2,457,000 and 2,875,000, and a 99 percent probability that the actual number is between 2,338,000 and 2,994,000.

This sampling error renders uncertain the comparison of some statistics reported by the AHS. Generally, the greater the difference between any two numbers or any two percentages calculated using AHS figures, the greater the likelihood that there is a real difference between them. The extent of the sampling error varies, however, as does its effect on the reliability of the figures and the apparent differences between figures, so there is no general rule that a difference of a certain size is, for example, 90 percent reliable. 

Intervals for all AHS figures used in this study are not provided here, either in the text or in this appendix, because they would make this report extremely difficult to read. It is clear, however, that caution is needed in interpreting relatively small differences between small proportions of the housing stock. 

On to next section

Back to table of contents