Living Together Shaped House Votes — But The Effect Was Smaller Than Believed

📌 What Was Studied

This study examines whether nineteenth-century U.S. House members who lived together influenced one another’s behavior or simply selected similar housemates. The analysis covers members' residences from 1801 to 1861 — several decades more than prior work — and revisits the so-called "boardinghouse effect."

📊 How the Question Was Answered

• Residential data on U.S. House members, 1801–1861, were compiled to map coresidence ties across decades.
• Improved identification strategies were used to separate social influence from selection, including:
• Weighting to adjust for differential selection into shared residences.
• A death-in-office design treating unexpected deaths as shocks to coresidential ties.

📈 Key Findings

• Legislators tended to live with politically similar colleagues, consistent with selection into shared residences.
• Coresidents who were politically divergent were more likely to end their shared living arrangements, supporting selective separation.
• After accounting for selection with weighting, coresidence still increased voting agreement — but the effect size is only about half as large as previously reported.
• The influence of coresidence was larger for weaker social ties and for relationships involving newly elected members.
• Exploiting deaths in office shows causal dynamics: unexpected deaths increased ideological distance between surviving legislators and their deceased coresidents, consistent with an influence mechanism.

💡 Why It Matters

Findings reconcile two views: social influence among nineteenth-century House members existed, but prior estimates overstated its magnitude because of selection into residences. The results highlight that informal, residential ties matter most for weaker ties and newcomers, and that natural experiments like deaths in office can help identify causal social effects in legislative behavior.