Memo

Generic Information Collection Request

Request: The Census Bureau plans to conduct additional research under the Generic Clearance for 2020 Census Field Test to Automate Field Data Collection Activities (OMB Control Number 0607-0971). The Center for Survey Measurement in conjunction with the Decennial Statistical Studies Division plans to conduct iterative small-scale mailout tests to refine invitations to an online survey. Results from these tests will inform the 2020 Decennial Census testing program for further testing. This submission documents our fourth test in this series. As a part of this current submission we are seeking approval to conduct this test.

Purpose: There are three purposes to this test. First, we want to evaluate a survey invitation mailing strategy which compares response rates based on the day of the week the mailings are received at the household. In our larger production tests, we have observed an increase in online response on Mondays and Tuesdays of the week and then a tapering off of response until the next Monday or Tuesday. This phenomenon appears to be independent of the day of the week the mailings are mailed. This test will determine whether mailings which are received earlier in the week generate proportionally more response than mailings which are received at the end of the week. The second purpose of this test is to determine how a staggered mailing affects calls to our telephone help line. Again, we have observed increases in calls on Mondays, independent of the day the mailings are received at the household. The current 2020 Census plan is to use staggered mailings reduce the peaks to the telephone lines on particular days. This test will allow us to begin to evaluate the call volume using a staggered mailout schedule. The third purpose of this test is to determine whether the instructions on the web survey instrument screens affect responses and usability. We have observed in usability testing that often respondents do not appear to read all the instructional text on the screen. This test will allow us to determine for different sized devices whether a clutter-free design aids usability without affecting response distributions.

Population of Interest: Results will inform 2020 Census planning and thus we are interested in data representing housing units in the U.S.

Sample: We will select a sample of 8,000 addresses from the Census Bureau’s Master Address File, stratifying for region of the U.S (for mailing purposes), urban and rural areas, and high and low Internet access areas (this could affect calls the telephone line). Each mailout day of the week treatment will have a sample of 4,000 housing units and each survey design treatment will have a sample of 4,000 housing units. The design will be fully-crossed as shown in Table 1.

Table 1: Sample design for January/February 2017 Small-Scale Mailout Test

	Mailout day treatment = 1 (Mail arrives late week)	Mailout day treatment = 2 (Mail arrives early week)
Survey Design Treatment = 1 (Instructions on screen)	2,000	2,000
Survey Design Treatment = 2 (Instructions in help)	2,000	2,000

Given a response rate of 35 percent, we should be able to detect an estimated 3 percent difference in response using alpha=0.05 and beta=0.80. The sample will exclude housing units selected for other recent Census tests and the American Community Survey.

Timeline: The surveys will be open between January 23 and March 15, 2017.

Language: The survey and mailing materials are in English only.

Method: We plan to conduct this test with a Census Day of February 1, 2017. There will be a maximum of four mailings to each housing unit. The mailings in one treatment (Mail arrives early week) will be mailed to generally arrive on Monday or Tuesday of the week and the mailings in the other treatment (Mail arrives late week) will be mailed to generally arrive on Thursday or Friday of the week. The survey will close out approximately six weeks after the first mailing. The third and fourth mailings will not be sent to any housing unit for which a response has been received.

This is an online-only data collection test with the URL https://respond.census.gov/ncbs. There is no paper questionnaire or telephone data collection for this test. Because some sampled households might not be able to respond online, we offer a telephone number with an automated message indicating that we will contact them if we need their information if they cannot respond on the Internet. To track calls to the telephone line, two pre-recorded lines will be used. In one mailout treatment, one telephone number will be used, and in the other mailout treatment, the other telephone number will be used. Telephone calls will be tallied daily for the two lines.

For the online survey, the login id will trigger the appropriate path of the instrument (either the control (Instructions on screen), or the experimental condition (Instructions in help). The survey will be optimized for the device accessing it. The control version of the online survey is similar to the 2015 National Content Test online survey used approved under the (OMB number 0607-0985) until 6/30/2018 and is identical to the online survey approved for the National Census Bureau Survey conducted in August and September 2016 approved under the OMB number 0607-0971 until 04/20/2019, with the exception of the questions on the last page of the survey. The survey collects general demographic data.

Incentives: None

Enclosures:

The first enclosure includes:

• Initial Letters for the two treatments – (only the mailout dates and telephone numbers are different)

• Internet card (this card is included with the initial letter and the final letter)

• Postcard Reminders for the two treatments – (only the mailout dates and telephone numbers are different)

• Second Postcard Reminders for the two treatments – (only the mailout dates and telephone numbers are different)

• Final Letters for the two treatments – (only the mailout dates and telephone numbers and dates in the box are different)

The second enclosure includes the survey questions screen shots for the control with a description of the version with the instructions in help.

Length of interview: We estimate that users will spend 8 minutes on average completing the survey and approximately 2 minutes reading mailing material. Thus, the total estimated respondent burden for this study is approximately 1,333 hours, which assumes everyone reads the mailing material and answers the survey.

The contact person for questions regarding data collection and statistical aspects of the design of this research is listed below:

Elizabeth Nichols

Center for Survey Measurement

U.S. Census Bureau

Washington, D.C. 20233

(301) 763-1724

[email protected]

2