pith. sign in

arxiv: 2604.04775 · v1 · submitted 2026-04-06 · 💻 cs.CY

Community Driving-Safety Deterioration as a Push Factor for Public Endorsement of AI Driving Capability

Amir Rafe , Subasish Das This is my paper

Pith reviewed 2026-05-10 19:29 UTC · model grok-4.3

classification 💻 cs.CY
keywords AI driving evaluationautonomous vehiclescommunity safety concerninconsistent mediationtechnology acceptancerisk spillovergeneralized AI orientation
0
0 comments X

The pith

Community driving-safety concerns boost specific support for AI driving while suppressing general AI enthusiasm, producing a near-zero net effect.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper models how perceptions of deteriorating driving safety in one's community influence judgments about whether AI or humans are better at driving. It finds that these concerns directly increase approval for AI driving as a domain-specific response, yet they also reduce broader positive views toward AI technology overall. Because general AI enthusiasm itself strongly supports AI driving approval, the indirect path works in the opposite direction. The two routes largely cancel, leaving almost no overall shift in evaluations of AI driving capability. A reader would care because the result shows how everyday safety worries can create targeted technology endorsements without increasing support for the underlying technology in general.

Core claim

Using weighted structural equation modeling on a nationally representative U.S. sample, the study establishes an inconsistent mediation in which perceived community driving-safety concern exerts a small positive direct effect on evaluations of AI versus human driving capability, while simultaneously suppressing Generalized AI Orientation, which itself positively predicts those evaluations; conditional indirect effects remain negative across levels of personal driving frequency, producing a risk-spillover pattern with near-zero net total effect.

What carries the argument

Inconsistent mediation model with perceived community driving-safety concern (PCSC) as the predictor, Generalized AI Orientation as the mediator, and personal driving frequency as the moderator of the indirect path.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Public messaging about local traffic risks could selectively increase acceptance of autonomous vehicles without raising enthusiasm for AI in other domains.
  • Acceptance of AI driving may hinge more on immediate community safety perceptions than on abstract attitudes toward artificial intelligence.
  • Similar risk-spillover patterns could appear in other high-stakes domains such as AI-assisted medical diagnosis or financial advising.

Load-bearing premise

That cross-sectional survey responses and the moderated mediation structure can support causal claims about push and spillover effects without experimental or longitudinal confirmation of directionality.

What would settle it

A longitudinal panel or randomized experiment showing that increases in perceived community driving-safety concern fail to raise AI driving evaluations or to suppress general AI orientation.

Figures

Figures reproduced from arXiv: 2604.04775 by Amir Rafe, Subasish Das.

Figure 1
Figure 1. Figure 1: Analytic pipeline overview. The six-stage design proceeds from data acquisition (Stage 1) through variable operationalization (Stage 2), CFA-based measurement modeling with WLSMV estimation (Stage 3), moderated mediation structural estimation (Stage 4), bias-corrected bootstrap inference (Stage 5), and a pre-registered robustness framework comprising seven complementary sensitivity checks (Stage 6). 𝑥 ∗ 𝑗 … view at source ↗
Figure 2
Figure 2. Figure 2: Structural path diagram for the moderated mediation model. Solid lines denote paths; the dashed line indicates the interaction path. Standardized coefficients (𝛽) are shown alongside each path. 𝑅2 values indicate variance explained in the mediator and outcome equations [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Bootstrap distributions of conditional indirect effects at low (−1 𝑆𝐷), mean, and high (+1 𝑆𝐷) driving frequency (10,000 resamples). Solid vertical lines mark point estimates; red dashed lines indicate 95% bias-corrected confidence interval boundaries. All three intervals exclude zero. In summary, the structural model reveals a pattern of competing effects. The direct path from PCSC to AI Driving Evaluatio… view at source ↗
Figure 4
Figure 4. Figure 4: Multigroup analysis by urbanicity. Path coefficients (𝐵 ± 95% CI) for the three structural paths across urban, suburban, and rural subsamples. The 𝑏1 path (AI Orientation → Outcome) is stable across groups; the 𝑎1 path (PCSC → AI Orientation) reaches significance only in suburban and rural contexts. Multigroup analysis by urbanicity Free multigroup estimation by urbanicity (urban, 𝑛 = 1,302; suburban, 𝑛 = … view at source ↗
Figure 5
Figure 5. Figure 5: Cross-task specificity of the PCSC direct effect (𝑐 ′ ) across six AI-vs.-human evaluation tasks from the HUMANVAI battery. The focal driving-task outcome is highlighted in red; non-driving tasks appear in blue. The positive direct effect of PCSC is statistically significant for the driving task (𝑝 = 0.010) and news writing (𝑝 < 0.001), negative for parole decisions (𝑝 = 0.038), and non-significant for med… view at source ↗
read the original abstract

Road traffic crashes claim approximately 1.19 million lives annually worldwide, and human error accounts for the vast majority, yet the autonomous vehicle acceptance literature models adoption almost exclusively through technology-centered pull factors such as perceived usefulness and trust. This study examines a moderated mediation model in which perceived community driving-safety concern (PCSC) predicts evaluations of AI versus human driving capability, mediated by Generalized AI Orientation and moderated by personal driving frequency. Weighted structural equation modeling is applied to a nationally representative U.S. probability sample from Pew Research Center's American Trends Panel Wave 152, using Weighted Least Squares Mean and Variance Adjusted (WLSMV)-estimated confirmatory factor analysis on ordinal indicators, bias-corrected bootstrap inference, and seven robustness checks including Imai sensitivity analysis, E-value confounding thresholds, and propensity score matching. Results reveal a dual-pathway mechanism constituting an inconsistent mediation: PCSC exerts a small positive direct effect on AI driving evaluation, consistent with a domain-specific push interpretation, while simultaneously suppressing Generalized AI Orientation, which is itself a strong positive predictor of AI driving evaluation. Conditional indirect effects are negative and statistically significant at low, mean, and high levels of driving frequency. These findings establish a risk-spillover mechanism whereby community driving-safety concern promotes domain-specific AI endorsement yet suppresses domain-general AI enthusiasm, yielding a near-zero net total effect.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript examines a moderated mediation model in which perceived community driving-safety concern (PCSC) predicts evaluations of AI versus human driving capability. The model is mediated by Generalized AI Orientation and moderated by personal driving frequency. Using weighted structural equation modeling with WLSMV estimation on ordinal indicators from a nationally representative U.S. sample (Pew ATP Wave 152), along with bias-corrected bootstrap and multiple robustness checks, the study reports a small positive direct effect of PCSC on AI driving evaluation, a negative effect on the mediator, and negative conditional indirect effects across levels of driving frequency, resulting in a near-zero net total effect. These results are interpreted as establishing a domain-specific push and risk-spillover mechanism.

Significance. Should the causal interpretations be supported, the paper makes a valuable contribution by shifting focus in autonomous vehicle acceptance research from purely technology-centered pull factors to include community-level safety concerns as push factors. The dual-pathway finding highlights potential opposing influences on domain-specific versus generalized AI attitudes, which could inform public policy and communication strategies around AI driving technologies. The methodological rigor, evidenced by the use of nationally representative data and extensive robustness checks including Imai sensitivity analysis and propensity score matching, is a notable strength.

major comments (1)
  1. [Abstract and Results section] The paper's central claim relies on interpreting the positive direct path from PCSC to AI driving evaluation as a 'domain-specific push' and the structure as a 'risk-spillover mechanism' that 'promotes' and 'suppresses' attitudes (see abstract and results interpretation). Given that the data are cross-sectional, these interpretations assume causal directionality that cannot be definitively established, despite sensitivity analyses. This is load-bearing because the title and abstract frame PCSC as a 'Push Factor'.
minor comments (2)
  1. [Methods] The full model specification, including all path coefficients, factor loadings, and model fit indices (such as CFI, TLI, RMSEA), should be reported in detail to facilitate evaluation and replication.
  2. [Methods] Exact definitions and survey items used for PCSC, Generalized AI Orientation, and AI driving evaluation should be provided explicitly, as these are central to the analysis.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on the limitations of causal inference. We agree that cross-sectional data preclude definitive causal claims and have revised the abstract, results, and discussion to use more cautious language emphasizing associations and theoretical consistency rather than definitive mechanisms or 'establishment' of effects. We address the comment in detail below.

read point-by-point responses

  1. Referee: [Abstract and Results section] The paper's central claim relies on interpreting the positive direct path from PCSC to AI driving evaluation as a 'domain-specific push' and the structure as a 'risk-spillover mechanism' that 'promotes' and 'suppresses' attitudes (see abstract and results interpretation). Given that the data are cross-sectional, these interpretations assume causal directionality that cannot be definitively established, despite sensitivity analyses. This is load-bearing because the title and abstract frame PCSC as a 'Push Factor'.

    Authors: We fully acknowledge that the cross-sectional design cannot establish causal directionality with certainty, even with sensitivity analyses. The interpretations are theoretically motivated by risk-perception and technology-acceptance frameworks positing community safety concerns as potential push factors, and the observed pattern (positive direct effect alongside negative effect on the mediator) is consistent with inconsistent mediation. To address the concern, we have revised the abstract to state that the results 'suggest' rather than 'establish' a risk-spillover mechanism, replaced 'promotes' and 'suppresses' with 'is associated with' where appropriate, and added explicit caveats in the results and discussion noting the correlational nature and the need for longitudinal or experimental designs to confirm causality. The title retains 'as a Push Factor' to reflect the hypothesized role, but we have inserted a clarifying sentence in the introduction stating that the framing is theoretical and the evidence is associational. The Imai sensitivity analysis and E-value calculations are now described with greater emphasis on their role in assessing robustness rather than proving causality. revision: yes

Circularity Check

0 steps flagged

No circularity in empirical mediation analysis

full rationale

The paper applies standard weighted structural equation modeling (WLSMV CFA, bias-corrected bootstrap) to an external nationally representative Pew ATP Wave 152 dataset. The moderated mediation model estimates direct, indirect, and conditional effects from observed survey responses without any self-referential definitions, fitted parameters renamed as predictions, or load-bearing self-citations. All reported pathways and the near-zero net total effect are statistical outputs from the data and robustness checks rather than tautological constructions, rendering the derivation chain self-contained.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard assumptions of structural equation modeling applied to survey data rather than new postulates; the main free parameters are the estimated path coefficients and factor loadings fitted to the Pew sample.

free parameters (1)
  • SEM path coefficients and factor loadings
    All direct, indirect, and moderated effects are estimated from the observed survey responses.
axioms (2)
  • domain assumption Survey items validly measure the latent constructs PCSC, Generalized AI Orientation, and AI driving evaluation
    Relies on confirmatory factor analysis but assumes the chosen indicators capture the intended concepts without substantial measurement error.
  • domain assumption The moderated mediation model correctly specifies the causal ordering and absence of important omitted variables
    Addressed via sensitivity analyses but remains a core modeling assumption for interpreting the risk-spillover mechanism.

pith-pipeline@v0.9.0 · 5539 in / 1593 out tokens · 64550 ms · 2026-05-10T19:29:17.471843+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages

  1. [1]

    https://newsroom.aaa.com/2025/09/study-finds-almost-all -drivers-experience-road-rage-but-it-can-be-stopped/

    Study finds almost all drivers experience road rage, but it can be stopped. https://newsroom.aaa.com/2025/09/study-finds-almost-all -drivers-experience-road-rage-but-it-can-be-stopped/. Accessed: 2025-03-29. AAAFTS,

    work page 2025

  2. [2]

    https://aaafoundation.org/2024-traffic-safety-culture-index/

    2024 traffic safety culture index. https://aaafoundation.org/2024-traffic-safety-culture-index/. Accessed: 2025-03-29. Acheampong, R.A., Cugurullo, F.,

    work page 2024

  3. [3]

    Psychological Bulletin 103, 411–423

    Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin 103, 411–423. Bandalos,D.L.,2014. Relativeperformanceofcategoricaldiagonallyweightedleastsquaresandrobustmaximumlikelihoodestimation. Structural Equation Modeling: A Multidisciplinary Journal 21, 102–116. Bansal, H.S., Taylor, S.F., St. James, Y.,

    work page 2014

  4. [4]

    https://news.gallup.com/poll/648953/americans-express-real-concern s-artificial-intelligence.aspx

    Americans express real concerns about artificial intelligence. https://news.gallup.com/poll/648953/americans-express-real-concern s-artificial-intelligence.aspx. Accessed: 2025-03-29. Golbabaei, F., Yigitcanlar, T., Paz, A., Bunker, J.,

    work page 2025

  5. [5]

    Journal of the Academy of Marketing Science 43, 115–135

    A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the Academy of Marketing Science 43, 115–135. tzeHu,L.,Bentler,P.M.,1999.Cutoffcriteriaforfitindexesincovariancestructureanalysis:Conventionalcriteriaversusnewalternatives.Structural Equation Modeling: A Multidisciplinary Journal 6, 1–55. Hulse, ...

    work page 1999

  6. [6]

    Whatinfluencesthedecisiontouseautomatedpublictransport?UsingUTAUT to understand public acceptance of automated road transport systems

    Madigan,R.,Louw,T.,Wilbrink,M.,Schieben,A.,Merat,N.,2017. Whatinfluencesthedecisiontouseautomatedpublictransport?UsingUTAUT to understand public acceptance of automated road transport systems. Transportation Research Part F: Traffic Psychology and Behaviour 49, 199–211. Man, S.S., Xiong, W., Chang, F., Chan, A.H.S.,

    work page 2017

  7. [7]

    Lawrence Erlbaum Associates

    Test Theory: A Unified Treatment. Lawrence Erlbaum Associates. Montoro,L.,Useche,S.A.,Alonso,F.,Lijarcio,I.,Bosó-Seguí,P.,Martí-Belda,A.,2019. Perceivedsafetyandattributedvalueaspredictorsofthe intention to use autonomous vehicles: A national study with Spanish drivers. Safety Science 120, 865–876. Naiseh, M., Clark, J., Akarsu, T., Hanoch, Y., Brito, M.,...

    work page 2019

  8. [8]

    Accessed: 2025-03-29

    https://crashstats.nhtsa.dot.gov/A pi/Public/ViewPublication/813561. Accessed: 2025-03-29. National Highway Traffic Safety Administration,

    work page 2025

  9. [9]

    Accessed: 2026-04-03

    https: //crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813800. Accessed: 2026-04-03. Nees, M.A.,

    work page 2026

  10. [10]

    Dangerous driving seen as on the rise, with cellphone distraction a top concern

    Pew Research Center, 2024b. Dangerous driving seen as on the rise, with cellphone distraction a top concern. https://www.pewresearch.org/sh ort-reads/2024/11/12/many-americans-perceive-a-rise-in-dangerous-driving-78-see-cellphone-distraction-as-major-problem/. Accessed: 2025-03-29. Pew Research Center,

    work page 2024

  11. [11]

    public and AI experts view artificial intelligence

    How the U.S. public and AI experts view artificial intelligence. https://www.pewresearch.org/internet/2025/04/03/ho w-the-us-public-and-ai-experts-view-artificial-intelligence/. Accessed: 2025-03-29. Rubin, D.B.,

    work page 2025

  12. [12]

    Computers in Human Behavior Reports 1, 100014

    Initial validation of the general attitudes towards artificial intelligence scale. Computers in Human Behavior Reports 1, 100014. Schepman,A.,Rodway,P.,2023. Thegeneralattitudestowardsartificialintelligencescale(GAAIS):Confirmatoryvalidationandassociationswith personality, government AI strategies, and proximity to AI. International Journal of Human-Compu...

    work page 2023

  13. [13]

    https://safetrec.berkele y.edu/news/2025-california-traffic-safety-survey-released-speeding-aggressive-driving-continue-be-concerns

    2025 california traffic safety survey: Speeding, aggressive driving continue to be concerns. https://safetrec.berkele y.edu/news/2025-california-traffic-safety-survey-released-speeding-aggressive-driving-continue-be-concerns. Accessed: 2025-03-29. VanderWeele,T.J.,Ding,P.,2017.Sensitivityanalysisinobservationalresearch:IntroducingtheE-value.AnnalsofIntern...

    work page 2025

  14. [14]

    Ethics and governance of artificial intelligence for health: Who guidance,

    https://www.who.int/publications/i/item/9789240086517. Accessed: 2025-03-29. Xu, Z., Zhang, K., Min, H., Wang, Z., Zhao, X., Liu, P.,

    work page arXiv 2025