Small-Area Income and Poverty Estimates Priorities for 2000 and Beyond (2000) / Chapter Skim
Currently Skimming:
4 Future Model Develpment: The Role of Surveys
4 Future Model Develpment: The Role of Surveys
Pages 82-124
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 82... ...
Research and development of the state and county models, as recommended by the panel, can help. However, marked improvement in the SAIPE estimates, particularly for school districts or other very small areas, will require new data sources.
|
From page 83... ...
This small sample size, together with the clustering of the CPS sample design, results in sizable sampling variability of the CPS state estimates and a lack of any sample in most counties and school districts. Looking to the future, several household surveys could contribute to improved estimates from the SAIPE program, and, in addition, the sample size of the March CPS itself may increase.
|
From page 84... ...
In the remainder of this chapter, we first compare the major features of the 2000 census long-form survey, ACS, March CPS, and SIPP. We then consider alternative uses for these surveys in the SAIPE Program, including: direct estimates for some areas; estimates to use as dependent variables in models; estimates to use as predictor variables in models; estimates for smaller areas of their shares or proportions of the poor population in larger areas; and estimates for controlling or calibrating other estimates on selected characteristics.
|
From page 85... ...
For the SAIPE county models, the panel recommends that the Census Bureau begin research and development now to explore the use of ACS estimates either to provide one of the predictor variables in CPS-based models or to serve as the dependent variable in the models. The Census Bureau should also conduct research on using ACS estimates, in place of or possibly combined with estimates from the previous census, to form within-county shares or proportions for school districts and other subcounty areas to apply to updated county model poverty estimates.
|
From page 86... ...
Offsetting the costs is that many of the activities recommended-such as exact matches of survey and census records-will be helpful for many other uses of household survey data, in addition to SAIPE. SURVEY FEATURES This section describes the main features of the 2000 census long-form sample, ACS, March CPS, and SIPP, including content, sample size and design, data collection schedule and procedures, residence rules, response rates and other quality measures, and data processing and release.
|
From page 87... ...
This design adds one more sampling rate, so that governmental areas with populations only slightly larger than areas with a 1-in-2 sampling rate will have a smaller increase in the proportional sampling error of their estimates compared with the 1990 sample design. For determining sampling rates in 2000, governmental areas were defined to include school districts in addition to counties, towns, and townships.
|
From page 88... ...
per year; months,, mental units and 1-in 8 for design alternatives being months; large census tracts; total considered that would 50,000 oc sample size about 18 oversample rural and plus 2,50 million housing units hard-to-enumerate areas househol In prev~o Data Collection Mail survey, personal Mail survey, telephone 1st and 5 Mode follow-up for follow-up, and then person; c nonresponse personal follow-up for by phone one-third of mail and phone nonrespondents Residence Usual residence; college "Current" or 2-month Usual ret Rules students in dorms counted residence rule students at college location at parent Response 1990 mail response rate Mail response rate 61% in 94-95% hi Rates 74% for occupied house- 4 test sites, plus 8% from respond, holds; net undercount of phone follow-up, plus 9% not respc 1.8% after follow-up; 19% from one-third follow-up of supplem. of aggregate income remaining nonrespondents, househol imputed for weighted response rate coverage of more than 95%; item 92% of cc response may be better than aggregate census, but not coverage
|
From page 89... ...
FUTURE MODEL DEVELOPMENT: THE ROLE OF SURVEYS 89 March Current Population Survey Survey of Income and Program Participation y survey, 1996, 1 sites ull finned ne for 300 Planned non) n acing Bar; being uld nd areas tone for ad ants Beth 61% in from lus 9% ~w-up of Dndents, nse rate item otter than erage Voluntary monthly labor force participation survey, begun in 1940s; income supplement every March Detailed questions on about 28 sources for previous calendar year Clustered sample of household addresses with state-representative design: addresses are in the sample for 4 months, out for 8 months, and in again for 4 months; total sample size of 50,000 occupied households plus 2,500 Hispanic households interviewed in previous November 1st and 5th interviews in person; other six interviews by phone Usual residence; college students in dorms counted at parents' address 94-95% households respond, but some do not respond to income supplement or for all household members; coverage estimated at 92% of census; 20% of aggregate income imputed Voluntary panel survey: each of 1984-1993 panels covered about 2.5 years; 1996 panel covered 4 years; 2000 panel to cover 1 year; 2001 panel to cover 3 years Detailed questions on about 65 sources for each month or for 4-month period preceding interview Clustered sample of household addresses: original sample of occupied households was 12,500-23,500 for 19841993 panels; 37,000 for 1996 panel, with oversampling of low-income households; 11,000 for 2000 panel; 37,000 planned for 2001 panel 1st, 2nd, and one interview in each subsequent year of a panel in person; other interviews by phone Similar to CPS; members of originally sampled households followed for life of panel 91-95% households respond to 1st wave, but sample attrition occurs; cumulative response only 69% by wave 8 of 1996 panel; coverage similar to CPS; 11% of aggregate income imputed Table continued on next page
|
From page 90... ...
Item nonresponse rates in 1990 were generally higher for income than for most other items. When household income information is missing, the Census Bureau uses statistical techniques to impute it on the basis of
|
From page 91... ...
For 2000, the long-form data are planned to be controlled to short-form data that have been corrected for measured population undercount. Long-form data, including income and poverty estimates, are provided for areas as small as census tracts, school districts, and block groups.
|
From page 92... ...
The current design calls for the ACS to use a sample design similar to that of the 2000 census long form, with higher sampling rates for small governmental units (including school districts) and lower sampling rates for large census tracts.
|
From page 93... ...
Preliminary results from the 1996 ACS test sites showed lower item nonresponse rates than in the 1990 census, at least for some items (Salvo and Lobo, 1998; Tersine, 1998~. But the ACS, like other household surveys, may cover the population less well than the census, based on one
|
From page 94... ...
compared ACS estimates for the counties in the 1996 and 1997 test sites before adjustment to population controls with the population estimates for those counties and found some degree of undercoverage for most of the counties relative to the population estimates. These comparisons include both within-household and whole-household misses.
|
From page 95... ...
population, has 1.3 percent of the CPS sample. This sample design means that income and poverty estimates in large states are generally more precise than those in smaller states.
|
From page 96... ...
In addition, some people who respond to the basic CPS labor force questionnaire do not respond to the March Income Supplement. To adjust for whole household nonresponse to the basic CPS, the Census Bureau increases the weights of similar responding households.
|
From page 97... ...
Publication Publication of detailed official income and poverty estimates from the CPS for the nation as a whole, geographic regions, and population groups occurs each year about 6 months after data collection in March. Limited statistics are also published for states on the basis of 3-year averages.
|
From page 98... ...
In addition, the design of the sample will be modified to represent all states and provide estimates for the largest states that have about the same level of error due to sampling variability as the current March CPS. Such a sample redesign, however, cannot be made until after the 2000 census results have been analyzed and used to redesign the samples for all major household surveys, which could take several years.
|
From page 99... ...
SIPP has lower item nonresponse rates than the March CPS: overall, only 11 percent of total regular money income obtained for calendar year 1984 from the first four waves of the 1984 SIPP panel was imputed, compared with 20 percent in the March 1985 CPS. The SIPP and March CPS imputation rates for 1984 for earnings were 10 percent and 19 percent,
|
From page 100... ...
If another survey were used for this purpose in place of the March CPS, comparability of the survey income and poverty measurements with the CPS measurements would be desirable, to reduce the likelihood of anomalies in the time series of estimates. For this use, survey estimates must be available on a frequent, timely basis.
|
From page 101... ...
EVALUATING ALTERNATIVE USES This section discusses several critical considerations for determining feasible and desirable roles for the 2000 census long form, ACS, March CPS, and SIPP in the SAIPE Program. These considerations are: sampling variability, timeliness, and comparability and quality of income and poverty measurements.
|
From page 102... ...
Nonetheless, the ACS estimates, when averaged over a year to provide a sample size of 3 million or about 1 in 36 housing units, will exhibit considerably higher sampling variability than estimates from the 2000 census long-form sample. Even when cumulated for 5 years, the ACS estimates will be more variable than the long-form estimates-not only is the 5-year ACS sample size somewhat smaller than the long-form sample size (about 1 in 7 compared to 1 in 6 households)
|
From page 103... ...
, which further reduces the effective sample size to about 62 percent of the originally designated sample (78% reduced by a factor to take account of the loss of precision from variable weights; see Kish, 1992~. Below we compare the sampling variability of direct estimates of poor school-age children for states from the 2000 census long-form sample, ACS, and CPS and for counties and school districts of different population sizes from the census and ACS.
|
From page 104... ...
104 En ._, 5a be ¢1 o ~0 u cry 5o be MU o V)
|
From page 105... ...
For areas with 50,000 or more total population (27% of counties in 1990, but only 6% of school districts) , the estimates of poor school-age poverty rates from the census long form have fairly low levels of error due to sampling variability, with coefficients of variation of less than 10 percent.
|
From page 106... ...
The calculations assume average sampling rates and do not allow for differences in sampling rates across geographic areas. County population size percentages are from Census Bureau data for 3,141 counties in 1990; school district population size percentages are from Census Bureau data for 9,243 school districts defined for 1990 in the Bureau's 1980-1990 evaluation file.
|
From page 107... ...
for a 3 percent annual sample and a design factor of 1.6. ACS 3-year average and 5year average CVs are calculated by applying a factor of 0.578 and 0.447, respectively, to the 1-year average CVs.
|
From page 108... ...
In the ACS, oversampling of a small school district will reduce the coefficient of variation for 5-year averages from about 47 percent to about 21 percent.9 T· ~. menses The second key consideration in using survey data to produce smallarea income and poverty estimates for such purposes as annual fund allocation is how regularly data can be provided and on what time schedule.
|
From page 109... ...
If ACS direct estimates can be used for SAIPE, averaged over 1 or more years, they will be more timely than the current model-based CPS estimates. However, if ACS estimates are used indirectly in models, reducing the time lag in the estimates will require efforts to improve the timely availability of other data used by the models or perhaps changes in how the data are used (e.g., perhaps using an earlier year of food stamp data as a predictor variable; see Chapter 3~.
|
From page 110... ...
Income and poverty estimates from the ACS will likely differ not only from CPS and SIPP estimates, but also from census estimates, even though the ACS is designed to be very similar to the census long form. A study that compared median household income in the 1996 ACS test sites and the 1990 census, adjusted to 1996 dollars, found that the ACS produced significantly lower medians than the census in all four sites (Posey and Welniak, 1998~.
|
From page 111... ...
In the 1990-1993 panels, about 20 percent of respondents used at least one type of record. Reference Period Both the 2000 census and the March CPS ask about receipt of income over the most recent calendar year at a time when many people have just completed or are preparing their income tax returns.
|
From page 112... ...
However, such estimates will still represent an average of different reporting periods, and, consequently, may differ from the estimates that would be obtained from the March CPS or the census. Residence Rules The ACS current or 2-month residence rule (see "American Community Survey," above)
|
From page 113... ...
ANALYSIS AND CONCLUSIONS Having considered the reliability, timeliness, and likely quality of data from the 2000 census long form, ACS, March CPS, and SIPP and the alternative uses that could be made of them for the SAIPE program (e.g., providing direct estimates, serving as predictor or independent variables in models) , we have reached several conclusions and recommendations.
|
From page 114... ...
Consequently, using the ACS to provide direct estimates for the SAIPE Program, except for states, does not seem warranted absent considerable evaluation work. The March CPS provides high-quality annual estimates, but it does not currently provide reliable direct estimates for any subnational areas, except for the very largest states.
|
From page 115... ...
Another useful set of comparisons would be exact matches of the 2000 census, 2000 ACS, 2000 March CPS, and 1996 SIPP with Internal Revenue Service (IRS) tax return records for 1999.15 Census-IRS, CPSIRS, and SIPP-IRS matches have been performed in the past (see, e.g., Childers and Hogan, 1984; Coder, 1991, 1992; David et al., 1986~.
|
From page 116... ...
Both uses, for which the Census Bureau should begin research and development now, involve indirect rather than direct estimation. One approach is for the Census Bureau to continue to base county estimates on statistical models for which the March CPS estimates form the dependent variable and ACS estimates are used as one of the predictor variables, along with the other variables that are currently in the models.
|
From page 117... ...
The use of 2-year or 3year average ACS estimates would place more weight on the direct estimates when they are combined with the model estimates than if average annual estimates, which have greater sampling variability, were used.l7 Given the likely measurement biases for ACS income and poverty estimates, estimates from the ACS-based county models could perhaps be improved by calibrating them in some way to selected estimates from the March CPS. For example, counties could be grouped into broad categories on such dimensions as race, ethnicity, and geographic region, and raking factors could be developed that would achieve consistency between the ACS model-based estimates for each county group and the corresponding March CPS estimates.
|
From page 118... ...
that could improve the SAIPE estimates for school districts and other subcounty areas is to use ACS data to form within-county shares to apply to updated county poverty estimates. The advantage of this approach, in comparison with the current estimation procedure in which the most recent census data are used to develop within-county shares to apply updated county model estimates, is that the ACS estimates will be more current.
|
From page 119... ...
For the greatest improvement in subcounty estimates of income and poverty, it will likely be necessary to develop a statistical regression model for these areas that makes use of administrative data for predictor variables (see Chapter 5~. However, development of appropriate administrative data is a long-range effort, so the Census Bureau should pursue the alternative of using ACS estimates, perhaps together with 2000 census estimates, in a within-county shares model.
|
From page 120... ...
The sampling variability in the census estimates would weaken the predictive power of the census variable, but the model would produce unbiased predictions. Direct Estimates for 1999 While it clearly makes sense to plan to use 2000 long-form estimates as predictor variables in SAIPE state and county regression models and, for the time being, in a county shares model for school districts and other subcounty areas, it is far from clear what use, if any, to make of the direct long-form estimates for income year 1999.
|
From page 121... ...
; using the long-form estimates with a ratio adjustment to short-form data to reduce the sampling variability of the estimates (see Chapter 3~; using the long-form estimates with a calibration to CPS aggregate estimates; not using the direct long-form estimates but, instead, using the current SAIPE models to produce indirect estimates for income year 1999. The Bureau should convene a meeting of key users to discuss these options so that the basis for the Bureau's decision is well understood.
|
From page 122... ...
Poverty estimates reflecting the current measure from the 2000 census or the ACS could also be used to form within-county shares for subcounty areas to apply to updated county poverty estimates developed from CPS-based models for which the dependent variable reflected a revised measure. Finally, if ACS estimates are calibrated in some manner to March CPS estimates of poverty developed with a revised measure, then the calibrated ACS estimates could be used as a dependent variable in regression This use of census long-form or ACS estimates of shares would require the assumption that the distribution of poverty within counties is similar under the current and revised poverty measures.
|
From page 123... ...
For this purpose, the Census Bureau should conduct a series of exact matches and analyses: · the planned exact match of the March 2000 CPS and the 2000 census long form; · an exact match of interviews from the 1996 SIPP panel covering 1999 income and the 2000 census long form; · the planned set of aggregate comparisons of income and poverty estimates from the 2000 ACS and the 2000 census long form; · an exact match of the 2000 ACS and the 2000 census short form to examine differences in measurement of household composition and demographic characteristics that relate to income and poverty; and · exact matches of Internal Revenue Service tax returns for income year 1999 with the 2000 census long form, 2000 ACS, March 2000 CPS, and 1996 SIPP. 4-3 Research and development by the Census Bureau should begin now to explore two possible uses of ACS estimates in SAIPE models for counties: to form one of the predictor variables in statistical models for which the March CPS continues to provide the dependent variable and to serve as the dependent variable in county models.
|
From page 124... ...
Reductions in funding could jeopardize its usefulness for SAIPE and, more generally, make it difficult to properly assess the potential uses of ACS data in small-area estimation. 4-6 The Census Bureau should plan to use 2000 census long-form estimates to form one of the predictor variables in the SAIPE state and county models.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.