2005 National Census Test

ACTION:

Proposed collection; comment request.

SUMMARY:

The Department of Commerce, as part of its continuing effort to reduce paperwork and respondent burden, invites the general public and other Federal agencies to take this opportunity to comment on proposed and/or continuing information collections, as required by the Paperwork Reduction Act of 1995, Pub. L. 104-13 (44 U.S.C. 3506(c)(2)(A)).

DATES:

Written comments must be submitted on or before January 3, 2005.

ADDRESSES:

Direct all written comments to Diana Hynek, Departmental Paperwork Clearance Officer, Department of Commerce, Room 6625, 14th and Constitution Avenue, NW., Washington, DC 20230 (or via the Internet at DHynek@doc.gov).

FOR FURTHER INFORMATION CONTACT:

Requests for additional information or copies of the information collection instrument(s) and instructions should be directed to Edison Gore, U.S. Census Bureau, Building 2, Room 2012, Washington, DC 20233-9200, (301) 763-3998.

SUPPLEMENTARY INFORMATION:

I. Abstract

The 2005 National Census Test (NCT) is part of the research and development cycle leading up to the re-engineered 2010 Census. The NCT will help the U.S. Census Bureau achieve one of its Strategic Goals—developing a census that is cost-effective, improves coverage, and reduces operational risk.

The Objectives of the 2005 NCT include studying methods for:

• Improving completeness and accuracy of reporting for short form items.
• Reducing respondent and data capture errors.
• Making questionnaires more respondent friendly.
• Improving coverage accuracy.
• Improving the operational feasibility of a targeted mailing for replacement questionnaires.
• Improving self-response and maintaining data quality by mailing bilingual questionnaires.

In conjunction with the 2005 NCT, the Census Bureau will conduct the 2005 Coverage Followup (CFU) operation. This operation is a continuation of the research and testing program begun in 2002 that is intended to develop and evaluate new procedures to improve coverage and reduce duplication. The CFU operation will collect data to evaluate the different versions of the coverage questions and different presentations of the residence rules instructions (See Definition of Terms). A separate Federal Register notice will be submitted for this operation.

Components of the Test

A. Control

The Control questionnaire will include short-form topics from the Census 2000 questionnaire. The standard mailing strategy will be used for both the initial and replacement Control questionnaires. All wording changes in the control questionnaire that are different from the Census 2000 questionnaire reflect refinements based on the 2004 Census Test questionnaire. The short form questions included in the Control questionnaire are currently considered to be the “best” version of each question. Our objective is to determine whether the experimental panels' question wording can improve the item response and data completeness over the control panel questions.

The control questionnaire will use the Residence Rules Instructions (See Definition of Terms) tested in the Alternative Questionnaire Experiment (AQE) 2000 (See Definition of Terms). This design will serve as the control for some of the experimental treatments because the results of tests conducted during Census 2000 indicated that the residence rules instructions used in the AQE questionnaire yielded better quality data than did the Census 2000 questionnaire residence rules instructions. The changes in format, presentation, and wording of the residence rules instructions used in the AQE resulted in a significantly higher response to the household count question (an important indicator of missing data and a flag for large household followup). In addition, the AQE questionnaire also produced better data for Hispanics who were likely to be left off census forms.

The Control Component includes four Self Response Option (SRO) treatments, each using the same form, content and initial questionnaire mailing strategy. Previous tests have shown that sending non-respondents a replacement questionnaire significantly increases response rates. We will employ four variations of the traditional replacement mailing strategy.

• Two treatments are planned to test the operational feasibility of two different replacement questionnaire-packaging strategies. Since the questionnaires designed to be included in these treatments may also “look” different, we also need to evaluate the response to them.
• A third treatment is planned to test the effect of providing a letter encouraging respondents to send in their original questionnaire or respond via the Internet. Households in this panel will not receive a replacement questionnaire.
• The fourth treatment is planned to test whether using messaging on the replacement questionnaire (that distinguishes it from the initial questionnaire) will increase response rates as well as reduce response duplication. This treatment is intended to create a clear differentiation between the replacement questionnaire and the original in order to make it easier for respondents to understand the intent of the replacement questionnaire.

The questionnaires for the Control component and the four (SRO) treatments share the same design and mailing strategy for the initial questionnaire. Consequently, we will be able to compare the results from the Control questionnaire with the results from the initial questionnaires in the SOR treatments. Doing so will give us a much larger sample for comparisons.

B. Hispanic Origin/Race

For the 2005 NCT, we plan to test modifications of the questions on race and Hispanic origin that are consistent with the 1997 Office of Management and Budget (OMB) Directive 15 (i.e., test a question on race that includes only the five minimum OMB race categories—See Definition of Terms). The version of the question chosen for future testing must produce data that is comparable in quality or better than the data produced by the Census 2000 questions.

The Hispanic origin and race component of the 2005 NCT is intended to evaluate the following elements: question design, the use of examples, revised wording of the questions and instructions, and a tribal enrollment question.

Question Design

We plan to test two fundamentally different designs—the traditional Hispanic origin and race design and a new design that includes shortened questions on Hispanic origin and race combined with a third question on ancestry. The shortened design includes only the five minimum OMB race categories and eliminates all write-ins. The Hispanic origin question component will consist of a yes/no option with no write-in option. The ancestry question component will include write-in lines that are intended to permit respondents to provide detailed information on their ancestry or country of origin.

Examples

We need to determine how useful it is to include examples of the OMB race categories to help respondents understand the intent of the design that includes the shortened questions. Consequently we plan to test this design with and without examples. We are currently conducting cognitive tests in order to select the most promising sets of examples for the 2005 NCT. We will choose the examples that yield the highest quality data for use in the 2008 Dress Rehearsal.

Secondly, based on the results of past censuses, we know that the specific examples used in the ancestry question can affect reporting. Since the detailed Hispanic origin (e.g., Mexican) and race information (e.g., Japanese) only will be collected in the ancestry question, we intend to evaluate the effect of using two different sets of examples on the reporting of detailed ancestry groups in that question.

Wording and Instructions

We plan to test the effect of changing the word order of the Hispanic origin item so that it reads, “Is this person of Hispanic, Latino, or of Spanish origin?” (Census 2000 order: “Spanish, Hispanic, Latino”). The instruction for the Hispanic origin item will reflect the OMB definition of Hispanic origin (See Definition of Terms) rather than relying on examples to communicate the intent of the question.

In addition, we plan to test revisions to the “MARK ONE OR MORE” instruction in order to make it more user-friendly, and we plan to test the effectiveness of the revised note that is intended to encourage respondents to answer both of the traditionally formulated race and Hispanic origin questions. The note is intended to reduce the number of Hispanics who report “Some other race”. The final wording of the question will be determined by cognitive testing that is currently underway.

Tribal Enrollment

The 2005 NCT plans to test a tribal enrollment question to attempt to determine what proportion of those who report a tribe are enrolled. We are currently conducting cognitive testing to determine the final wording of the question. We plan to evaluate the quality of tribal enrollment data.

C. Tenure and Other Population Questions

Tenure

We plan to test the following elements in the Tenure (own or rent) and other population questions: dropping the reference to “cash” rent, adding an instruction to improve the reporting of home equity loans, and a version that combines both treatments.

We intend to test the effect of eliminating the term “cash” from the tenure question, since the traditional formulation of the question has been criticized as not accurately depicting how rent is actually paid (e.g., by check). We plan to compare missing item data rates for the test questions to those for the Census 2000 question in order to determine the effect of eliminating the term.

Since we are not sure whether respondents understand that home equity loans are liens against the home, we also plan to test an instruction asking respondents who own their own homes whether they have a mortgage or loan, including home equity loans. We plan to evaluate the resulting owner distribution (owned free and clear vs. owned with a mortgage or loan). We plan to evaluate both variables in order to choose the version of the question that yields the highest quality data for use in future tests.

Age

In Census 2000, many respondents incorrectly reported the age of babies under one year of age. The 2005 NCT will test an instruction in the Age question to help respondents correctly determine the age of babies who are less than one year old. We also plan to reverse the order of the Age and Date of Birth questions to make them consistent with electronic modes such as the Internet.

Modified Categories in the Relationship Question

We plan to test the effect on response distributions of replacing “Foster Child” with “Foster child or foster adult”. Cognitive tests indicate that respondents understand the phrase “foster adult” and do not consider it to be offensive.

We plan to test the effect on data quality of replacing “Natural-born son/daughter” with “Biological son/daughter” [used in the Survey of Income and Program Participation]. Adoptive parents have received the term “Natural-born” unfavorably.

D. Residence Rules Instructions and Coverage Questions

Improving the accuracy of census coverage is one of the major goals of reengineering the 2010 Decennial Census Program. As a result of the Census 2000 Testing, Evaluation, and Experimentation Program and the Coverage Measurement Program, we implemented a research and development program to investigate ways of improving our coverage of persons and housing units in preparation for the 2010 Census. The 2005 NCT is part of this effort. Specific areas that we intend to evaluate in the 2005 NCT include improving within-household coverage and revising residence rules instructions so that they are clear and unambiguous to the respondent.

The panels in this section of the 2005 NCT are planned to evaluate the main effects and anticipated interactions of the residence rules instructions (See Definition of Terms) and two versions of the 2004 Census Test coverage questions.

The Control for this section of the 2005 NCT is the experimental roster tested in the AQE in the 2000 Census. This design was chosen as the control because it out-performed the Census 2000 residence rules instructions in two ways: It had significantly lower item nonresponse, and, in low coverage areas, the rate at which Hispanics were not included on the AQE questionnaire was significantly lower than for the Census 2000 questionnaire.

The following are the features of the residence rules instructions that will be evaluated:

• Content, order, and wording of cues and bullets in the include/exclude lists (See Definition of Terms).
• List order.
• Presentation of the basic “usual residence” concept.
• Approach to structuring the residence rules instructions (for example, using an alternative approach that eliminates the include/exclude lists and relies instead on explaining the basic concept behind the lists).

The Coverage Followup (CFU) operation will evaluate the effects of the alternative residence rules instructions on gross coverage errors.

The Census Bureau is currently conducting cognitive tests using four versions of the 2004 Census Test undercount question (Question 2) and overcount question (Question 10). The versions of the questions that produce the best results will be chosen for use in the 2005 NCT. The experimental treatments are intended to isolate some of the individual effects of each version of the coverage questions by crossing them with the different residence rules instructions. We plan to evaluate the coverage questions' efficacy in flagging potential omissions or erroneous enumerations by implementing the CFU operation.

E. Respondent-Friendly Design

The questionnaire for this panel will have design changes intended to make it easier to use. The changes that this questionnaire will test include:

• Color as a navigational tool.
• Lightly embedded text that describes what should be entered in response boxes.
• Consistent formats between check boxes and write-in answer fields (for example, all answer fields will be outlined with a strong black line). Using consistent formats for all answer fields is intended to help respondents identify all fields where a response is required.

F. Language

The 2005 NCT will include a bilingual English/Spanish questionnaire panel in an effort to improve self-response in the growing number of households in which Spanish is a primary language. We plan to implement an English/Spanish questionnaire mailout treatment that is intended to evaluate the effect of a bilingual questionnaire on response rates, public reaction, and data quality.

G. Internet Option

All respondents (including those in the Language panel) will have the opportunity to respond via the Internet, but while the general content of the Internet questionnaire will be the same as other test questionnaires, it is not planned to mirror the exact wording of any one specific paper questionnaire. Instead, the Internet Questionnaire will be designed using questions from several of the 2005 NCT questionnaires and rewording the questions to reflect the wording that works best for this response mode.

Respondents who ask to respond via the Internet (including those in the language panel) will be randomly assigned to answer either a person-based or a topic-based format. In the person-based approach, responses for all items (e.g., name, date of birth, gender, race) are collected for one household member (person), after which the same questions are repeated for each successive household member. In the topic-based approach, responses for a given topic/item (e.g., age) are collected for all persons in the household, after which responses for the next topic/item (e.g., date of birth) are collected. This process continues for each successive topic/item. We will evaluate the quality of Internet data collected using these two design treatments.

II. Method of Collection

In late August, we will mail an advance letter to a national sample of about 420,000 households. This letter will explain why we are conducting the mandatory 2005 NCT. The letter also will assure respondents that their answers are confidential. We will inform them of the measures we take to keep their personal information secure. The 2005 NCT questionnaires will be mailed approximately a week later. Respondents will be asked to mail back their completed questionnaires or respond via the Internet by Census Day (September 15, 2005). Early in September, we will send reminder/thank you postcards thanking those who have already responded and asking non-respondents to send in their questionnaires or reply via the Internet. As part of the Census Bureau's efforts to improve response rates and contain costs, most nonrespondents will receive replacement questionnaires a few days after Census Day. Households assigned to the panel for which the experimental treatment consists of a letter in lieu of a replacement questionnaire will not receive the second questionnaire.

The Coverage Followup (CFU) operation is scheduled to begin in December. The CFU operation will obtain additional information by telephone from a sample of respondents in order to evaluate the residence rules instructions and coverage questions. Approximately six months after Census Day, we will begin formal evaluations of population and housing content, coverage, language, race and ethnicity, and self-response options.

Definition of Terms

Alternative Questionnaire Experiment—The 2000 AQE incorporated three separate experiments, one involving census long forms and the other two involving short forms, with different objectives. This experiment was conducted during Census 2000 under census conditions. Consequently, we were able to compare the effectiveness of the AQE questionnaire designs with the Census 2000 questionnaires.

All three experiments tested combinations or “packages” of design features, rather than testing each design change separately in a controlled fashion that would permit inferences about their individual effects. Thus, firm conclusions only were drawn about the combined effect of multiple design features. This is an important limitation of all three experiments. The three experiments were:

• 1.1 Experiment A: Effects of Altering the Design of Branching Instructions on Navigational Performance.
• 1.2 Experiment B: An Experiment to Improve Coverage Through Revised Roster Instructions.
• 1.3 Experiment C: Questionnaire Effects on Reporting of Race and Hispanic Origin: Results of a Replication of the 1990 Mail Short Form in Census 2000.

The report describing the experiment is located at the following address: http://www.census.gov/pred/www/rpts/TR17.pdf.

Include/exclude List—The list of the people the respondent should include in the household count and those who should be left out because they should not be counted or will be counted elsewhere.

Office of Management and Budget (OMB) definition of Hispanic origin—A person of Mexican, Puerto Rican, Cuban, Central or South American or other Spanish culture or origin, regardless of race.

Office of Management and Budget (OMB) race categories—American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, and White.

Residence Rules Instructions—Instructions that respondents use to determine who should be counted in that household. They are meant to insure that everyone is counted once and in the right place for the primary purposes of apportionment and redistricting.

III. Data

OMB Number: None.

Form Number: DC-1A through DC-1X (2005 Census Test questionnaires).

Type of Review: Regular.

Affected Public: Individuals or households.

Estimated Number of Respondents: Approximately 420,000 households.

Estimated Time Per Response: 10 minutes.

Estimated Total Annual Burden Hours: 70,000.

Estimated Total Annual Cost: There is no cost to respondents except for their time to respond.

Respondent's Obligation: Mandatory.

Legal Authority: Title 13 of the United States Code, sections 141 and 193.

IV. Request for Comments

Comments are invited on: (a) Whether the proposed collection of information is necessary for the proper performance of the functions of the agency, including whether the information shall have practical utility; (b) the accuracy of the agency's estimate of the burden (including hours and cost) of the proposed collection of information; (c) ways to enhance the quality, utility, and clarity of the information to be collected; and (d) ways to minimize the burden of the collection of information on respondents, including through the use of automated collection techniques or other forms of information technology.

Comments submitted in response to this notice will be summarized and/or included in the request for OMB approval of this information collection; they also will become a matter of public record.