Testing of Coverage Questions Plan

The Census Bureau plans to conduct additional research under the generic clearance for questionnaire pretesting research (OMB number 0607-0725). The overall goal of the proposed research is to improve coverage of the population and minimize mode differences in a cost-effective manner for the 2020 Census. A major goal of the 2020 Census is to reduce data collection costs by reducing the need for follow up due to incomplete or incorrect data from the initial census questionnaire. We can begin to achieve this by cognitive interviews designed to reduce measurement error in all phases of data collection, but particularly at the initial census enumeration. The specific objectives of this research are twofold:

• To cognitively test residence questions on all primary census forms to assess respondent comprehension and understanding of the forms in paper and electronic modes; and

• To gather qualitative information on the usability of electronic versions of all forms, including respondent perceptions of security.

This study will consist of three iterative rounds. This letter seeks clearance for the second round only. The second round of this study will include 80 cognitive interviews with census questionnaires across a range of data collection modes (paper, Smartphones, and tablets) and with a range of populations who would be possible respondents to a self-response household census questionnaire, a Be Counted questionnaire, a Group Quarters (GQ) questionnaire or a Nonresponse Followup (NRFU) interview. A total of 6 questionnaires will be tested in Round 2

We will test 4 electronic versions (2 different instruments which could be used as either a self-response household form or an enumerator form (NRFU), a Be Counted instrument, and a GQ instrument) and 2 paper forms (Be counted and Household).  

Included in Round 2 is an experimental comparison of two approaches to the household and NRFU electronic form (we are calling them versions 3 and 4 to distinguish them from versions 1 and 2 tested in Round 1). Both versions 3 and 4 use a topic level approach to the census overcount questions. However, version 3 asks about the topic at the household level. For example, the question is worded “did you or any of these people…?” This is similar to the way the 2012 National Census Test online form is designed. Version 4 asks about the same topic for each person individually and then goes on to the next topic. The goal is to test whether the more detailed version 4 uncovers additional places where someone could be counted.

As was done in Round 1, we are testing using a single electronic form as both a self-enumeration and an interviewer enumeration (i.e., testing whether the same questionnaire could be used for either purpose). The Smartphone V3 and Tablet V3 for Household and NRFU are identical questionnaires, as are the Smartphone V4 and Tablet V4 for Household and NRFU. Also included in Round 2 are electronic versions of the self-administered Be Counted and Group Quarters forms as well as paper versions of the household form and the Be Counted form for comparison purposes. All electronic forms were developed using similar strategies to maintain consistency across forms.

All six unique questionnaires are attached.

Table 1 shows the distribution of respondents by mode and questionnaire type.

Table 1: Round 2 Targets for Questionnaire Type by Mode

Questionnaire type	Paper	Smart phone	Tablet	Total
Be Counted	5	7	8	20
GQ	0	8	7	15
Household	8	9	9	26
NRFU*	0	9	10	19
Grand Total	13	33	34	80

*The version of the NRFU instrument will have no impact on the recruitment of participants for these interviews. RTI/RSS plans to equally distribute the interviews between V3 and V4 (approximately 9-10 cases for each version). They will also distribute cases between the smartphone and tablet versions of the NRFU within V3 and V4 as equally as possible (2-5 each for smartphone and tablet); distribution of cases within NRFU will depend on the availability of cases in each interview location and by device. They will monitor recruitment/interviewing and will adjust plans based on the priorities of the Census project team.

In October of 2012 through January of 2013, under the direction of the Census Bureau, staff from RTI, International and RSS will conduct a maximum of 80 cognitive interviews with respondents in the Raleigh-Durham, NC; Chicago, IL; Southern California and Washington, DC areas. The contractor will recruit using the following criteria, in addition to demographic diversity:

• Household and NRFU questionnaires –

  • Complex households including households with the following types of members:

    • temporary or tenuously attached people,

    • college students living away,

    • people on military deployment,

    • people who travel for work for months at a time,

    • people who maintain other residences to be closer to work,

    • non-family such as housemates or roommates, and

    • non-nuclear family members (other than parents and children).

  • Respondents who have a seasonal or second home, and

  • Those who have moved within 1-2 months of the interview.

• Be Counted –

  • Persons who are experiencing homelessness,

  • Persons who are highly mobile or tenuously attached who might be omitted from a household census questionnaire,

  • Respondents living in a household that could be missed in the census (such as a basement apartment attached to a single-family home unnoticeable from the outside).

  • Respondents who may have difficulty providing address information, including,

    • Doubled up families who share an address, but not living space,

    • Persons using a PO Box for mail, who do not have a street address that accepts mail, and

    • Persons with a rural route address.

• Group Quarters –

  • Persons living in group homes including at least one of each of the following categories:

    • Group home

    • Nursing home

    • College dorm or housing / homeless shelter

Cognitive interviews will be tape-recorded, with the participants' permission, to facilitate summary of the results. For interviews conducted with mobile devices, the devices will be hooked up to laptop computers for screen-capture, to facilitate the interview and for analysis. All participants will be informed that their response is voluntary and that the information they provide is confidential. Respondents will receive $40 for their participation.

The estimated time for completion of each of the cognitive interviews is an hour. Thus, the estimated burden for this research is 80 hours.

The contact person for questions regarding data collection and study design is:

Patricia Goerman

Center for Survey Measurement

U.S. Census Bureau

Room 5K024B

Washington, D.C. 20233

(301) 763-1819

Patricia.l.goerman@census.gov