Jan C. Zoellick, Adelheid Kuhlmey, Liane Schenk, Daniel Schindel, Stefan Blüher
9 min
Historically, research into public attitudes toward autonomous vehicles (AVs) has relied heavily on hypothetical scenarios, which often yielded inconsistent results regarding the influence of age and gender. To improve ecological validity, the researchers conducted a field study involving 125 participants who experienced a real-world ride in an electrically powered, level-4 automated shuttle on clinic campuses in Berlin. By measuring participants' reactions immediately after physical exposure, the authors aimed to model the relationships between attitudes (acceptance, trust, perceived safety), emotions (amusement, fear, surprise), and the ultimate intention to use the technology in the future.
The study highlights that core psychological constructs—acceptance, perceived safety, and trust—are strongly correlated with one another and function as key drivers of usage intention. Specifically, hierarchical regression analysis showed that acceptance was the most potent predictor, followed by perceived safety. Interestingly, while age was a significant predictor of the intention to use AVs—with younger participants generally being more inclined to use them—it remained predictive even after accounting for attitudes like trust and acceptance. In contrast, gender-based differences in attitudes, while present, lost their predictive power when combined with core attitude metrics in the model. The researchers also attempted to integrate emotional measures but found these correlations to be weak or inconsistent, suggesting that future research needs more robust ways to capture emotional states in transit.
This paper moves the field beyond hypothetical surveys, emphasizing that "in-vivo" experiences provide more reliable insights into how humans interact with emerging mobility technologies. By clarifying that acceptance and safety perceptions are central to usage intentions, the findings provide a roadmap for policymakers and vehicle developers: focusing on public feelings of safety and broad acceptance is more critical for adoption than targeting specific demographic groups. The researchers also advocate for greater transparency and consistency in how constructs like 'acceptance' are defined, arguing that standardized methodology is the only way to resolve the current ambivalence in the literature.
Alex: The analysed sample contained one hundred and twenty-five participants. About half were male, and the authors describe the sample as disproportionately young, tech-interested, and urban. They included passengers without driving licences and participants below adulthood.
Sam: Let’s separate the measurements. If acceptance already means willingness to use something, wouldn’t predicting use from acceptance be nearly circular?
Alex: That’s precisely the conceptual problem the authors raise. Here, acceptance means favourable or unfavourable attitudes, not behaviour. Intention to use means someone’s reported likelihood of taking a future ride. They measured these with separate questionnaire scales.
Sam: And safety and trust also sound close. How did the authors avoid treating them as different names for the same response?
Alex: Perceived safety means believing that vehicle use will affect your well-being. Trust means a belief that lets you willingly become vulnerable after considering the vehicle’s characteristics. The definitions differ, though the responses were strongly related.
Sam: Separate definitions are a start, but questionnaires don’t always behave as planned. Did these scales hold together in the data?
Alex: The authors checked whether items grouped into the intended scales. Most did, but perceived safety needed an item removed. Boredom didn’t form the expected grouping, so they dropped it from further analyses. That matters when interpreting their emotion findings.
Sam: Before we reach the emotions, give us the main evidence. What did the attitude measures tell them about intended use?
Alex: Acceptance, perceived safety, trust, and intended use were all strongly correlated. But correlation alone doesn’t show which measure adds information beyond the others. For that, the authors used hierarchical regression: adding predictors in stages and comparing their contributions.
Sam: So they began with demographics, then asked whether knowing someone’s attitudes improved the prediction. How much difference did that make?
Alex: Age and gender alone accounted for thirteen percent of the variation in intended use. Adding acceptance made the largest improvement. With acceptance, trust, and perceived safety all included, the model accounted for forty-nine percent.
Sam: That’s substantial within this sample. But it doesn’t mean changing attitudes would cause that much change in usage, or that they predicted actual travel.
Alex: Neither claim is established here. These are associations among post-ride questionnaire responses, with intended use as the outcome. There’s also no before-and-after comparison showing that the ride itself changed attitudes.
Sam: Then why does trust lose its place in the final model? Is the study saying trust doesn’t matter?
Alex: Trust was related to intended use and contributed when first added. Once perceived safety entered, trust no longer made a statistically clear independent contribution. Safety and trust were especially closely correlated, so their information overlapped.
Sam: The practical lesson sounds like choosing measures carefully, not abandoning trust. What happened to the demographic predictions once those attitudes were included?
Alex: Gender lost predictive power after acceptance was added. Age remained a significant predictor even with all the attitude measures included: older participants reported lower intended use. That contradicted the expectation that both demographic predictors would fade away.
Sam: The authors also expected young men to be especially positive. Did this real-ride sample support that familiar early-adopter story?
Alex: No. They couldn’t establish that young men scored higher than the other age-and-gender groups. Unequal variances required a stricter significance threshold. There were indications of gender differences, with perceived safety clearing that threshold, but the broader pattern needs replication.
Sam: Now the title brings in amusement. Were people’s emotional reactions useful additions, or mostly a different description of their attitudes?
Alex: The emotional evidence was weaker. Participants generally reported amusement and little fear. Amusement’s associations with attitudes were small and uncertain. Surprise wasn’t correlated with the attitude measures, while fear was negatively associated with intended use.
Sam: Low fear seems unsurprising in an accompanied, slow-moving shuttle. Does that make the emotion measurements a poor test of future passenger reactions?
Alex: The authors raise that concern themselves. Fear may not fit this setting well; subtler anxiety might be more appropriate. They also suggest different measurement methods and larger samples, rather than concluding that emotions are irrelevant.
Sam: Where does that leave your confidence in the headline finding? Physical experience improves on imagining a ride, but the conditions still seem narrow.
Alex: I’d treat this as useful evidence about relationships in this particular passenger sample. The larger attitude associations are better supported than the small emotion links. But willing science-event visitors on prepared campus routes don’t represent all potential users.
Sam: And the operator limits what the experience can tell us about surrendering control completely. Who should read the full document, and where should they start?
Alex: Researchers studying automated transport or technology acceptance should start with the methods’ construct definitions. Then read the hierarchical regression results to see what acceptance, trust, and safety each add. Read the limitations alongside them, especially the supervised setting and selective sample.
Sam: For everyone who only needs the main lesson, what should they carry back to their colleagues?
Alex: After these rides, favourable attitudes and feeling safe predicted intended use better than demographics alone. But the line to remember is this: liking a vehicle, trusting it, and planning to use it are related—not interchangeable.
Sam: Keep those distinctions in your next survey.