Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Engstrom and Kernell. 2005. Manufactured responsiveness. AJPS.
America's founders sought to divide legislative and executive power by using electoral institutions to give legislators different electoral incentives and constituencies than those faced by the president. If they were successful, we should expect to frequently observe divided government. And indeed we do, at least in the twentieth century. But, strangely, the nineteenth century was a period of frequent unified government. In particular, the House was almost always controlled by the same party that controlled the presidency. Why?
The pre-reform ballot made it easy for local party leaders to know exactly how much support to expect in each area; they knew that anybody supporting their presidential candidate would support their House candidate, and they could observe this in non-secret voting. Thus, it was possible to draw very precise districts. A state party could draw legislative districts that would be sure to ensure that the party would just barely win in almost every district. This is "efficient" gerrymandering.
But the post-reform ballot made it so that state parties had less precise information about where the votes came from. Thus, drawing "efficient" gerrymanders was more risky--it might result in the party barely winning every district, but small mistakes might allow the minority to win several districts. Thus, twentieth-century district drawers switched to a different strategy. Rather than draw a marginal win in every district, they would "pack" all the opposition supporters in as few districts as possible, and spread their own supporters into several comfortable districts. Although this would guarantee the opposition a certain number of districts, it also guaranteed the majority party a large number of non-risky seats. This is the "packing" strategy.
Lo and behold, the authors are correct. Their time-series data bear this out.
By using ballot-type as a proxy for the "efficiency" of gerrymandering, the authors predict:
Research by the same authors
Research on similar subjects