How does response modality affect the quality of insights collected through AI-moderated interviews? Results from 252 AI-moderated interviews across three groups (voice, text, and hybrid, in which respondents had free choice between voice and text) show that voice responses are longer (236% more overall words), more diverse (138% more “unique” words), more descriptive (213% more “content” words) and thematically richer (28% more thematic codes assigned) than text responses. Hybrid responses fall in between these extremes. Participants rated all modalities similarly well in terms of: ease of use, empathy of the AI moderator perceived, and willingness to disclose (open up) though most (55%) would rather use text modality, for comfort and privacy. These findings highlight trade-offs and offer researchers and practitioners guidance in balancing depth with respondent preference.
The pandemic reshaped consumers’ relationship with cash, resulting in more frequent use of digital transactions (e.g., credit cards, phone payments). This change in consumer behavior may have prompted a similar shift in incentive preference, with respondents finding cash less enticing post-Covid. This analysis explores incentive preference using vignette questions about how respondents would like to receive $5 and $40 promised incentives for completing a web survey. The results reveal that cash is not the preferred monetary incentive for all subgroups, and the majority of respondents prefer digital payments (i.e., digital gift cards, bank transfers). In addition, people’s forms of payment for common expenses correlate with incentive preference indicating that these post-pandemic shifts in consumer behavior are relevant to survey practitioners looking to offer incentives that are persuasive to potential respondents.
As survey practitioners, we try to offer incentives that are enticing to our potential respondents, both in amount and type of payment. Common wisdom is that cash is most persuasive for gaining response. However, the growth of contactless payment during the pandemic changed the way people pay for goods and services, and therefore changed their relationship with cash. This analysis investigates payment behaviors among survey respondents as a precursor to further research on incentive preference. While most people carry cash, the findings point to a strong preference for digital payments. Those more likely to use cash were older, male, low-income, non-White, and less educated. With such a small subgroup using cash, this research prompts further questions as to whether cash is still king in the incentive world.
Latino adults, particularly those residing in rural areas, remain underrepresented in survey research due to language barriers, housing instability, and limited internet access. Designing an effective data collection methodology for this underrepresented group is difficult given the limited literature on best practices for this population. Latino people are the largest foreign-born population group in the United States, underscoring the need for greater representation. To address this question, the 2024 Community Health Survey (CHS) drew on the principles of responsive and adaptive survey design (RASD) to target rural-dwelling Latino adults in the western United States to measure their experiences regarding health care access and immigration policies. The survey employed a dual-frame design using address-based sampling (ABS) and prepaid cell phone samples. Over three waves, the study made modifications across four key areas: a prepaid incentive experiment, participant recruitment materials, sample design, and a post-completion incentive. The prepaid incentive experiment conducted in Wave 1 compared a single $2 bill to two $1 bills and found that a single $2 bill yielded a marginally higher response rate than two $1 bills. Increasing the prevalence of Spanish-language outreach, reducing the number of sampled cases flagged as unlikely to match the population of interest, and incorporating a $10 post-incentive in subsequent survey waves collectively improved response rates and cost efficiency. By Wave 3, the cost per interview declined by 40 percent compared to Wave 1 across both sample frames. Demographic shifts across waves, including increased participation among younger adults and larger households, highlight the effectiveness of these responsive changes. This study demonstrates the utility of RASD principles in improving survey response and representation among rural Latino adults and offers practical insights for tailoring future survey designs to this population.
West Virginia is experiencing a shortage of Direct Care Professionals (DCPs), a critical workforce providing hands-on support to help people remain in their homes. Based on recommendations from a statewide taskforce led by AARP WV, the West Virginia Department of Human Services, Bureau for Medical Services, initiated a Direct Care Workforce Training Surge. This training surge aimed to train up to 800 DCPs over two years (2024 and 2025), while optimizing the training through evaluation, leading to a more prepared and confident workforce. This paper highlights the details and importance of a flexible mixed methods approach utilized to determine an ideal training length during the first eight months of the training surge (the pilot phase). The evaluation applied an explanatory sequential mixed methods design; the flexibility inherent in the approach allowed evaluators to pivot when necessary to improve data collection. West Virginians can be considered a “hard-to-reach” population as the state is fully within the Appalachian region and residents live in predominantly rural areas with potentially limited transportation and connectivity services; potential DCPs, the target audience for the training, are situated within this environment and therefore, a flexible design was crucial to ensure all voices were heard when making training length recommendations.
