Charles Stewart and Joseph Loffredo have posted this draft on SSRN. Here is the abstract:

In recent years, confidence and trust in the American electoral system have drawn the attention of academics, the media, the public, and policymakers. Much of the academic literature about confidence in the mass public has focused on longitudinal analysis, ignoring cross-state variability arising from differences in political context, administrative practices, and state legislation. This paper fills this gap in the literature by focusing on interstate variability in election trust. Using the Survey of the Performance of American Elections to measure confidence at four levels (personal, local, state, and nationwide), we explore the influence of the winner/loser effect, mail-ballot usage, vote-counting speed, and the passage of restrictive and expansive state legislation on confidence, aggregated to the state level, from 2012 to 2024. The paper culminates in a multivariate statistical analysis in which we find strong evidence for the influence of the winner/loser effect, mail-ballot usage, and vote-counting speed on interstate variability in trust, especially among Republicans. The paper concludes with a discussion of nuances revealed by the analysis that should support further investigation. Among these are the partisan asymmetry of effects on confidence and how different independent variables have differing degrees of influence on the four levels of trust.