The two major parties present American voters with increasingly stark electoral alternatives. How do independents make political decisions in this era of heightened partisanship? Prevailing accounts portray independents either as inattentive, low-information voters or as “undercover” partisans whose behavior mirrors that of Democrats and Republicans. We challenge both characterizations. Drawing on novel survey experiments, we compare the information-seeking behavior of independents and partisans by examining the importance they attach to different candidate attributes and the quantity of information they seek before choosing between candidates. We find that independents and partisans rank candidate attributes in similar ways, with both groups prioritizing issue positions over socio-demographic characteristics. Yet, independents rely less on candidate partisanship and acquire more information than partisans across a variety of hypothetical electoral conditions. These differences persist across levels of political knowledge, suggesting that independents and partisans use distinct decision rules when evaluating political candidates. 

Electoral systems shape not only who governs, but also how citizens evaluate the rules and goals of democratic competition. I argue that voters update their institutional preferences in response to the way electoral rules translate votes into seats, and test this argument using British Election Study panel data spanning five general elections from 2010 to 2024. Within individuals, support for proportional representation declines as their party's seat-vote gap increases: a ten-point increase in a party’s seat bonus is associated with a 13-point decline in the probability that its supporters favor proportional representation. This relationship persists after accounting for national and constituency-level electoral success and is roughly twice as strong among supporters of parties outside government. Electoral systems, therefore, generate feedback through multiple representational outcomes. Because institutional preferences track electoral outcomes rather than accumulating over time, the foundation for electoral reform is, in part, rebuilt after each election.

Pollsters underestimated support for Donald Trump in 2016, 2020, and 2024 and the UK Conservative Party in 2015 and 2024, leading many to argue that polls systematically understate right-wing vote shares. We provide the first comparative test of this argument, fitting Bayesian hierarchical models to more than 38,500 election-poll-party observations from 372 elections in 32 countries between 1945 and 2025. The dominant bias, however, favors the left: pooling across elections, polls overstate support for left-wing parties by 1.4 percentage points, inflating the left's lead over the right by 1.6 points. Left-wing parties are overestimated in most countries and in polling dating back to the 1940s. Right-wing parties are more modestly underestimated, chiefly among conservative and agrarian parties rather than the far right, and this bias has declined in recent decades. Persistent partisan bias challenges explanations built on shorter-term, election-specific mechanisms and matters because polls shape campaign strategies and voter behavior. 

Does information about one branch of government affect trust in another? We show that institutional trust is neither institution-specific nor uniformly system-wide. In a survey experiment embedded in the British Election Study Internet Panel, respondents received information about parliament, national and supranational courts that varied the policy content of decisions, party framing of these decisions, and the conduct of individual officeholders. Policy decisions conflicting with respondents' preferences reduce trust well beyond the institution responsible, while congruent decisions leave trust unchanged. Spillovers are strongly asymmetric: parliamentary decisions depress trust in the courts more than in parliament itself, whereas court decisions produce little spillover beyond the judicial domain. Criticism from parties offsets rather than deepens the penalty, regardless of how much respondents like the cueing party. Information about officeholders' conduct moves trust neither within nor across institutions. We conclude that institutional trust is selectively contagious, shaped above all by institutions' policy decisions.