Attachment Y - DOB Research

Research Section Proposal for 2020: Date of Birth

May 29, 2019

Background

Prior to 2009, CE asked respondents for the day, month, and year of birth for each household member. In 2009, CE switched to asking only month and year of birth. Finally in 2011 the question on date of birth was removed entirely to “eliminate unnecessary PII”. It was replaced with the question, “As of today, how old are you?” These decisions were made in order to reduce any potential privacy concerns from respondents and because there was no business need for exact date of birth, just age.

One potential business need for asking respondents for exact date of birth is to facilitate assigning a Protected Identification Key (PIK) that can be used to link members in CE households with other person-level administrative and alternative data sources at the Census Bureau. In 2016, as part of contracted work with the Census Bureau to link CE data to IRS data, we requested that a simulation be done to identify PIK rates using age at time of interview in lieu of birth date. Census compared PIK assignment rates for the American Community Survey using full date of birth with rates using age. For 2013, the PIK assignment rate using date of birth was 94 percent and using age was 84 percent. Assuming CE is similar, a 10 percent sample increase could be expected for any of CE’s linking projects if date of birth were available for the PIK matching procedure. To date, the CE program has considered re-adding date of birth in the future, if any person-level linking to administrative records (e.g., IRS, HUD public housing data, social security data) were to move from a research phase into production.

Recently, there has been increased interest from internal stakeholders, and external researchers, in matching administrative records to the CE for the development of consumption based policy measures. Interest was expressed in asking the date of birth question to improve match rates for the PIK-ing process. After providing reasoning described above for not including date of birth as a survey questions, we proposed adding a question to the research section. This would not only allow for a CE-based PIK exercise with exact data, but would also allow CE to (1) explore the impact on the respondent of asking date of birth and (2) explores measurement error properties of the currently collected age by comparing reported age to date of birth calculated ages.

Proposal

In the 2020 Policy Analysis and Research section (asked to 4^th wave respondents), add a question for the respondent and CU-members as outlined in Attachment 1, with the following embedded experiment:

1. Ask half the sample the date of birth questions prior to the four burden questions in the research section and half of the sample the date of birth questions after the four burden questions. The burden questions include one on question sensitivity, so this may provide some indication of respondents perception of being asked date of birth. Given that the rates of burden/sensitivity don’t change much from quarter to quarter and no other changes are anticipated for 2020, the previous year can be used as the control.

Time Period of Collection: 2020q1 to TBD (minimum of 4 quarters)

Attachment 1. Date of Birth Questions

Now I’d like verify demographic information provided earlier for you and members of your household.

What is your/(name)’s date of birth?

* Enter Birth Month

1. January

2. February

3. March

4. April

5. May

6. June

7. July

8. August

9. September

10. October

11. November

12. December
    
    

* Enter Birth Day [enter text] _____________

* Enter Birth Year [enter value] _____________

As of today, that would make you/(name) "age" year/years old.

Is that correct?

1. Yes

2. No

Even though you don’t know your/(name)’s exact birthdate, what is your best guess as to how old you/he/she were/was on your/his/her last birthday?

1. [enter value] _____________

99. 99 years or older

* Ask if necessary

3