September 20, 2017

MEMORANDUM FOR Associate Directorate for Research and Methodology (ADRM)
From:

Center for Survey Measurement (CSM)

Subject:

Respondent Confidentiality Concerns

CSM researchers have noticed a recent increase in respondents spontaneously expressing
concerns about confidentiality in some of our pretesting studies conducted in 2017. We
recommend systematically collecting data on this phenomenon, and development and
pretesting of new messages to avoid increases in nonresponse among hard-to-count
populations for the 2020 Census as well as other surveys like the American Community Survey
(ACS).
Below is a preview of findings relating to respondent confidentiality concerns from recent CSM
projects, followed by a more detailed recommendation from CSM. These findings are drawn
from usability interviews with English- and Spanish-speaking respondents (N=15), cognitive
interviews with Spanish-speaking respondents (N=10), four focus groups with Spanish-speaking
Field Representatives (FRs) (N=16), five focus groups with Field Supervisors (FSs) and Field
Representatives (N = 24), and 42 focus groups with respondents (N=366). These interviews and
focus groups were conducted in different regions of the country in English, Spanish, Chinese,
Korean, Vietnamese, Russian, and Arabic since January of 2017. All projects were small,
qualitative studies and as such, unrepresentative of the population as a whole, and none of
them were specifically designed to examine confidentiality concerns. However, respondents
and field representatives spontaneously brought up these concerns at a much higher rate than
CSM researchers have seen in previous pretesting projects, and as such, this information may
have implications for nonresponse on U.S. Census Bureau studies and surveys.
In particular, CSM researchers heard respondents express new concerns about topics like the
“Muslim ban,” discomfort “registering” other household members by reporting their
demographic characteristics, the dissolution of the “DACA” (Deferred Action for Childhood
Arrival) program, repeated references to Immigration and Customs Enforcement (ICE), etc. FRs
and FSs emphasized facing a “new phenomenon” in the field and reported that respondents’
fears, particularly among immigrant respondents, have increased markedly this year.
Respondents reported being told by community leaders not to open the door without a warrant
signed by a judge, and CSM researchers observed respondents falsifying names, dates of birth,

 and other information on household rosters. FRs requested additional training to help them
overcome respondents’ fears regarding confidentiality and data sharing with other agencies like
ICE, as well as materials they could share with respondents to reassure them about these
concerns.
Usability Findings (2017 PEGA Internet Self-Response Instrument; N = 15)
Overall, four of fifteen respondents who participated in usability interviews in the DC-metro
area to pretest the 2017 PEGA internet self-response (ISR) instrument in English and Spanish
intentionally provided incomplete or incorrect information about household members due to
concerns regarding confidentiality, particularly relating to perceived negative attitudes toward
immigrants.
One Spanish-speaking respondent said she was uncomfortable “registering” other household
members and tried to exit the survey at the dashboard when she realized she would have to
provide information on others who live with her. She mentioned being afraid because of the
current political climate and news reports about changing immigration policy. The researcher
had to help the respondent delete the other household members from the roster to avoid a
break-off; she only provided her own information.
A second Spanish-speaking respondent filled out information about herself and three family
members but intentionally left three or four roomers off the roster because, “This frightens me,
given how the situation is now” and mentioned being worried because of their “[immigration]
status.” Both Spanish-speaking respondents stated that they would not complete the survey at
home.
A third Spanish-speaking respondent, who the researcher had reason to believe was not
concerned about whether his data would be shared with other federal agencies because of his
status as legal resident in the country, commented: “Particularly with our current political
climate, the Latino community will not sign up because they will think that Census will pass
their information on and people can come looking for them.” This theme came up repeatedly
even for those without concerns about the immigration status of members of their household.
One English-speaking respondent entered false names and some incorrect dates of birth for his
roommates because he was not comfortable providing their information without their consent
due to data sharing concerns.
A second English-speaking respondent did not report five unrelated household members (some
of whom were immigrants) because she does not report their rental income to the IRS and
because of what she referred to as the “Muslim ban.”

 It should be noted that this level of deliberate falsification of the household roster, and
spontaneous mention of concerns regarding negative attitudes toward immigrants, is largely
unprecedented in the usability interviews that CSM has been conducting since 2014 in
preparation for the 2020 Census. In general, we assume that pretesting respondents are in fact
more willing to fill out the survey than most respondents would be during the 2020 Census,
given that they are being paid a cash incentive for their participation and being interviewed by
a researcher with whom they have established rapport. As such, these concerns might be even
more pronounced during a production survey than researchers observed during pretesting.
Cognitive Findings (CBAMS Paper Testing; N = 10)
Spanish-speaking respondents who participated in paper testing of the CBAMS (Census Barriers,
Attitudes, and Motivators Survey) expressed concern about whether their answers might be
shared with other government agencies. One respondent said, "The possibility that the Census
could give my information to internal security and immigration could come and arrest me for
not having documents terrifies me.” Later she commented that she was worried that her
information could be used against her if she answered that she is not satisfied with the
government here. She thought someone could say, ‘If you're not satisfied, why are you here?’
and this could be used against her to expel her from the country.
Respondent concerns on this survey were eye-opening for CSM researchers because some of
the respondents who participated in cognitive interviews had previously taken part in CSM
pretesting projects. Despite having participated in the past, they seemed visibly nervous and
reticent and required extensive explanations regarding how their data would be used and their
personal identifying information would be redacted. This behavior was in contrast to their
demeanor during prior CSM pretesting projects.
Multilingual Focus Groups on Doorstep Messages for the 2020 Census (N = 366)
Respondents also raised concerns in 42 focus groups conducted this spring in order to test
doorstep messages that enumerators can use to overcome reluctance in the 2020 Census.
These focus groups were conducted in English, Spanish, Chinese, Korean, Vietnamese, Russian,
and Arabic, and the topic of confidentiality concerns came up in several groups.
For example, Spanish-speakers brought up immigration raids, fear of government, and fear of
deportation. Respondents talked about having received advice not to open the door if they fear
a visit from Immigration and Customs Enforcement (ICE) and that they could instead ask that
warrants be slipped under the door. They suggested that the Census Bureau have something in
writing that enumerators could slip under the door to indicate why an enumerator is at a
respondent’s home. They felt that the most important message to encourage participation was
confidentiality and the greatest barriers to Latino participation are fear and mistrust.

 Several Chinese-speaking focus group respondents stated that the Chinese community’s main
fear or concern was immigration status and how the data are used. They also expressed
concern about opening the door to a government official and not wanting to be “investigated.”
Arabic-speakers reported that they had concerns about their perception of the current
environment as unwelcoming to Arabic-speaking immigrants and said that they feared
deportation. One respondent said, “The immigrant is not going to trust the Census employee
when they are continuously hearing a contradicting message from the media everyday
threatening to deport immigrants.” Respondents wanted to have more assurance about how
the data would be used before providing personal information.
English-speakers expressed similar reservations when discussing the current “environment.” In
one English focus group, respondents spontaneously expressed concerns that their personal
information would be shared with other agencies, and mentioned in particular that data could
be shared with Immigration and Customs Enforcement and the Department of Homeland
Security. One participant recommended that Census materials should explicitly explain that
personal information is not shared with these agencies.
Overall, concerns about the confidentiality of data, including between agencies, negative
perceptions of immigrants, and deportation emerged across languages in this project.
Focus Groups with Spanish-speaking Field Representatives (N = 16)
CSM conducted four focus groups from July to September with Spanish-speaking Census Bureau
Field Representatives who work in different states regarding the Spanish translation of a health
survey. Many of the FRs spontaneously brought up the topic of an upsurge in respondent
confidentiality concerns.
Many FRs stated that before they can begin an interview, they have to spend several minutes
calming respondents and gaining their trust due to the current “political state.” One FR said,
“The politics have changed everything. Recently.” Another mentioned that this is especially
relevant given that the DACA (Deferred Action for Childhood Arrival) program is “on the
chopping block.” FRs reported that some respondents worry about giving out legitimate names
or completing the roster; they often do not feel comfortable giving out information about other
people in the household. One FR said, “This may just be a sign of the times, but in the recent
several months before anything begins, I’m being asked times over, does it make a difference if
I’m not a citizen?” FRs reported that many Spanish-speaking respondents distrust the
statement on confidentiality in the survey mailing materials, even when they understand it.

 Many respondents believe that “the less information they give out, the better. The safer they
are.”
One FR said that in June she was doing a Census Bureau survey interview with questions about
citizenship status. A Spanish-speaking respondent answered that he was not a citizen, and then
appeared to lie about his country of origin. When the FR started asking about his year of entry
into the U.S., he “shut down” and stopped responding to her questions. He then walked out
and left her alone in the apartment, which had never happened to her during an interview
before.
Another FR commented that she had seen this scenario many times while administering the
ACS, although this was the first time she had heard of a respondent actually leaving the FR
alone in his or her home. She suggested that respondents might have concerns about
confidentiality given “the current political climate."
A third Spanish-speaking FR added that she had observed Hispanic members of a household
move out of a mobile home after she tried to interview them. She said, “There was a cluster of
mobile homes, all Hispanic. I went to one and I left the information on the door. I could hear
them inside. I did two more interviews, and when I came back, they were moving.... It's because
they were afraid of being deported."
FRs reported using various strategies to overcome respondents’ fears. They are often asked if
they work for other federal agencies, and reassure respondents that this information is not
reported to other federal agencies; their information is not shared with “immigration or taxes.”
They explain that the respondent’s immigration status does not matter. The FRs reported that
sometimes they encourage respondents to do the interview anonymously with fake names,
when it seems like the respondent is about to refuse.
The FRs recommended that ad campaigns be used to reduce the mistrust the public has toward
completing our surveys. They also requested “an immigration letter” like one used on the
NHANES (National Health and Nutrition Examination Survey) that mentioned “la migra” [a slang
term for ICE] that was very effective. The FRs could use it selectively when it was needed. It
clearly said that the Census Bureau was not in any way related with “la migra”.
FRs were asked to share the most important change that they wanted to see made to the
Spanish translation of the survey materials. In one focus group, the three FRs agreed
unanimously that they would like an “immigration statement” to appear on mailing materials
because of current “political issues.” They reported that immigration concerns are the “topic of

 the day” and that they always have to allay fears about immigration by saying, “We do not
share information with other agencies.” They suggested that the statement should convey that
while the Census Bureau is part of the federal government, it is a statistical agency, and that the
respondent’s legal status in the country does not matter at all.
Focus Groups with Field Supervisors and Field Representatives (N = 24)
CSM conducted five focus groups in September with Field Supervisors and Field
Representatives to collect feedback on FR training, the availability of printed materials in
various languages, and the usage of printed materials during a recent housing survey operation.
The topic of respondent concerns regarding confidentiality came up repeatedly in these focus
groups.
In one focus group of Field Supervisors, an FS reported having a respondent produce papers
proving US citizenship of household members during an interview. Another FS reported that
each time she spoke to a Spanish-speaking respondent, her focus was on convincing the
respondent of the confidentiality of their answers “given the political temperature these days.”
One FS said, “we have to let [respondents] know where this information is going. That’s their
biggest fear.” When asked if the training the FRs had received was adequate, an FS commented
that more training was needed on respondent confidentiality concerns, but that “this climate
didn’t exist before [when training was designed last time], when you did the study th ree years
ago, so of course it wasn’t planned in there.” FSs reiterated that the main issue they saw was
privacy concerns of Latino respondents, and that FRs should do more practice interviews where
someone models those concerns and concerns about immigration so that the FRs are more
prepared to respond adequately in the field.
FRs who spoke a language other than Spanish or English (e.g., Cantonese) reported that
completing interviews for the survey in question this year was much harder than the last ti me
the survey was fielded: “Three years ago was so much easier to get respondents compared to
now because of the government changes… and trust factors [and] also because of what
happened here [in the United States].…Three years ago I didn’t have problems with the
immigration questions.” Another FR commented, “There will always be political situations that
are out of our control …. Sometimes I just come right out and say, this isn’t for immigration.”
Even FRs who only speak English reported needing additional training for encountering
households where respondents are especially fearful. One FR reported that respondents have
been confusing him with someone from Immigration and Customs Enforcement (ICE, formerly
known as INS). He reported that respondents that identified him as working for the government
were hesitant to answer any questions, and it was difficult to gain their trust. Another FR
agreed that most incompletes were due to a distrust of the government. When asked whether

 their training adequately prepared them, several FRs mentioned that training regarding
concerns about ICE could not have been included in the training they received because it was a
new phenomenon. The FRs in this focus group emphasized that they were having to reorder
the questions in this housing survey to collect demographics last in order to avoid breakoffs.
Spanish bilingual FRs shared many of the same concerns as the Field Supervisors, speakers of
languages other than English or Spanish, and the monolingual English-speaking FRs. They
emphasized that when completing interviews with Spanish-speaking households, immigration
concerns were challenging and that respondents seemed fearful. They requested more training
focusing on respondent fears, particularly immigrant respondents’ fears. They mentioned
respondents giving out false names and reordering survey questions to collect demographics
last.
Recommendation
Overall, these findings, in various languages from respondents, Field Representatives, and Field
Supervisors across the country who have participated in recent projects are raising concerns
within CSM regarding potential barriers to respondent participation in the 2020 Census, as well
as other Census Bureau surveys. The findings listed above are a sampling of what CSM
researchers have observed on recent projects, and these concerns were all expressed
spontaneously to researchers during the course of pretesting various survey materials. These
findings are particularly troubling given that they impact hard-to-count populations
disproportionately, and have implications for data quality and nonresponse.
A systematic pretesting study evaluating respondent confidentiality concerns, both from the
perspective of respondents as well as Field Representatives, would shed light on the nature and
prevalence of these concerns, particularly for Limited English Proficient (LEP) or immigrant
populations in the U.S. Quantitative analysis could also be done to examine any changes in
response rates, mode of administration, item non-response, or number of contact attempts for
surveys such as the ACS among non-English speakers and hard-to-count, immigrant
respondents. Similarly, we could review whether the number of residents reported or the
number of unrelated household members within households has declined in recent months.
In addition to gathering data on any uptick in confidentiality concerns that may exist, we
recommend designing and pretesting wording that could address these concerns in mailing
materials, the Decennial Internet Self Response instrument, FAQs provided to enumerators, etc.
This text could inform respondents that the Census Bureau does not collect information on
immigration status or religion (similar to the language stating that we do not collect social
security numbers), or that we do not share data with agencies like ICE. Pretesting with
respondents from a variety of backgrounds would be vital given that such a message could be

 reassuring to some respondents but may have other effects for different populations. Care
should be taken in crafting new messages. CSM also recommends that additional training be
provided to FRs across surveys regarding allaying respondents’ confidentiality concerns.