Att 5b _2023 Q2 NHIS Proposed New Content Concepts Measured, Duplication, and Proposed Uses of Data

Attachment 5b

2023 Q2 NHIS Proposed New Content

Concepts Measured, Duplication, and Proposed Uses of Data

Questions are being added to quarter 2 (Q2) of the 2023 NHIS to inform NCHS’s ongoing efforts to evaluate the quality of data collected from probability-based online survey panels. New content includes questions on speaking a language other than English at home and using a language other than English while engaging in different activities.

Non-English Speakers/Limited English Proficiency (LEP) – Sample Adult

 Concepts to be Measured

• Speak a language other than English at home (LANGHM_A; taken from the ACS). If YES:

• Other language spoken most often at home (LANGSPEC_A; adapted from the ACS).

We discussed asking a third ACS item (“How well do you speak English?”) to measure LEP but identified it as a poor fit for the NHIS given interviewer administration. Some adults who would receive this item would have already answered several questions in English and may find this item offensive. Instead, we propose to ask about the language (English vs. some other language) used most often during different activities:

• Language used most often when watching television, reading news online or in print, or listening to the radio (LANGMED_A)

• Language used most often when seeing a doctor or other health professional (LANGDOC_A)

• Language used most often when participating in social activities, such as visiting friends, attending clubs and meetings, or going to parties (LANGSOC_A)

Duplication and Previous NHIS

• The first two questions are currently on or adapted from the American Community Survey.

• The remaining three questions are new.

Expected Prevalence and Reliability

• LANGHM_A: 2021 ACS prevalence = 21.7%.

• LANGSPEC_A: 2021 ACS prevalence: Spanish = 58.2% (15.1% of all adults); other Indo-European languages = 18.9% (4.9% of all adults); Asian and Pacific Island languages = 17.2% (4.5% of all adults); other languages = 5.7% (1.5% of all adults). We anticipate creating a variable that separates Spanish from all other languages.

• For LANGMED_A, LANGDOC_A, and LANGSOC_A there are no benchmarks for determining likely prevalence. However, we anticipate that the percentage of adults who answer “yes” to these items will be the same as or exceed the ~42% of adults in the 2021 ACS who reported speaking English well, not well, or not at all (after indicating that they speak a language other than English at home). This translates to at least 9% of the adult population.

Proposed Use of the Data

• Like many surveys, probability-based, online panels tend to achieve lower response rates among non-English speaking/LEP adults (Lavrakas et al., 2016; Ramirez et al., 2017). However, coverage and nonresponse error are further compounded for probability-based online panels in that non-English speaking/LEP adults also tend to have lower recruitment rates and higher attrition rates once empaneled. Hence, non-English speaking/LEP adults are often under-represented in panel surveys (Kocar and Biddle, 2023; Ventura et al., 2018).

• From a health perspective, research has shown non-English speaking/LEP individuals to be less likely to have a usual source of care and to receive routine preventative care, and more likely to have an unmet medical need and to be uninsured, relative to their English proficient counterparts (Foiles Sifuentes et al., 2020; Gulati and Hur, 2022; Jang and Kim, 2019). Research has also shown non-English speaking/LEP individuals to be at higher risk of activity limitations, fair or poor health, and depression (Jang et al., 2016).  

• We hypothesize that questions on non-English speaking/LEP adults will produce variables useful for calibrating Rapid Survey System (RSS) weights because 1) we expect the RSS to underestimate, relative to the NHIS, the percentage of adults who use a language other than English at home and during different activities, and 2) we expect the language items to be associated with health outcomes of interest on the RSS. While LANGHM_A could be calibrated to the ACS, the other variables (LANGMED_A, LANGDOC_A, and LANGSOC_A) do not have corresponding benchmarks on the ACS.

• These questions will also enable data users to explore links between LEP and a variety of health outcomes, but that is not the primary reason for their inclusion on NHIS.

References

Foiles Sifuentes AM, Cornejo MR, Li NC, Castaneda-Avila MA, Tjia J, and Lapane KL. 2020. “The role of limited English proficiency and access to health insurance and health care in the Affordable Care Act era.” Health Equity, 4.1: 509-517.

Gulati RK and Hur K. 2022. “Association between limited English proficiency and healthcare access and utilization in California.” Journal of Immigrant and Minority Health, 24: 95–101. https://doi.org/10.1007/s10903-021-01224-5

Jang Y, Yoon H, Park NS, and Chiriboga DA. 2016. “Health vulnerability of immigrants with limited English proficiency: A study of older Korean Americans.” Journal of the American Geriatrics Society, 64:1498–1502.

Jang Y and Kim MT. 2019. “Limited English proficiency and health service use in Asian Americans.” Journal of Immigrant and Minority Health, 21: 264–270. https://doi.org/10.1007/s10903-018-0763-0

Kocar S and Biddle N. 2023. “Do we have to mix modes in probability-based online panel research to obtain more accurate results? Methods, Data, Analyses, 17(1): 93-120. DOI: 10.12758/mda.2022.11

Lavrakas P, Dirkz G, Lusskin L, and Ponce B. 2016. “Experimenting with the addressee line in a mail survey of Hispanic households.” Survey Practice, 9(5). https://doi.org/10.29115/SP-2016-0030

Ramirez AS, Willis G, and Rutten LF. 2017. “Understanding Spanish-language response in a national health communication survey: Implications for health communication research.” Journal of Health Communication, 22(5): 442-450. doi:10.1080/10810730.2017.1304470.

Ventura I, Bautista R, and Hendarwan E. 2018. “An experiment in panel recruitment for Spanish speaking populations: The AmeriSpeak case study.” Proceedings of the Joint Statistical Meetings, 3115-3126.

3