My @NRO piece today has it all:

Back-handed slaps at Macron! Cold-water throwing on leftist Nordiphilia! A call to expand the House of Representatives! Empirical data about apartheid! 11 charts and graphs! Involved arguments about constitutional process! nationalreview.com/2018/12/is-the…
I won't do any spoilers on the article here on twitter. I will say that I really really want you to read it even if you hate it, because it took a ton of work and original research.
Possibly the first (certainly first I've seen) apples-to-apples international comparison of voter participation among adult citizens for a large-ish sample of countries ever published in a political magazine? Yep, that's me!
Historic citizen voting participation by state? Yep, had to make that one myself.

Historic electoral outcomes and participation for a basket of peer countries? Luckily other people collect that data, but the formatting is BLEH.
And that's the thing, see: I am suspicious that the reason we have so much anger over these issues is partly because we have such poor data on them. We see time series charts that begin in 2000, international comparisons with ONE country at a time, etc.
Now, with that done, let me give you some #BonusContent for twitter followers.

NRO's excellent editor for this piece, @RAVerBruggen , made a comment to me that my metric, simple vote share vs. seat share, isn't what academics prefer: they like the efficiency gap.
He is correct. Most academic studies use efficiency gap as their preferred metric. What is the efficiency gap, you ask?

Well, it's complicated, but *basically* it comes from a specific intuition about how districted elections work.
Say you have a country with 2 parties which are both randomly distributed around the country. In some places A is bigger, in some places B is bigger. If A is 51% and B is 49%, A's seat share should be ~50-52%, right?

But what if A is 60%?

A should win *more than* 60% of seats.
Why? Because as share of total vote rises w/ quasi-random distribution of voters, the amount of concentration needed for the minority party to win rises, and concentration is unlikely to arise randomly.
Ergo, a party with 60% of the vote winning e.g. 70% of the seats is what you would expect from fair, districted elections.

So why don't I use this estimate, or the related "wasted votes" calculation?
Several reasons:
1. These calculations come with the tacit assumption that *partisan support will be quasi-randomly distributed*. I do not acknowledge the philosophical view that sectional parties should be viewed as inefficient to be valid.
That is to say, in practice, partisanship is not evenly distributed, not even close; the distribution is not random, nor should we expect or desire it to be. It is fine for regions to have strong, correlated preferences, and I am unwilling to use a calculation that penalizes that
2. I am interested in comparing across widely different systems. The tools that work for describing US politics may not be as informative for Swedish politics. Using a cut-and-dry vote % vs seat % calculation translates very easily to other contexts.
My method "does not care" if you get to vote/seat share harmony by having even distribution of partisans in districted elections, 2-party polarity, or proportional representation. It absorbs all the idiosyncracies to give the net outcome.
Basically, it's a way of punishing countries with high proportional representation thresholds, or crazy gerrymanders, or weird partisan mixes, etc, without having to make specific, complicated decisions for each idiosyncratic piece of an electoral system.
3. In the debate about electoral systems, most people are implicitly carrying in their head a national referendum model. They compare the fairness of an election to what would have happened if it were run as a simple referendum.
In other words, I am interested in creating a metric that *matches typical political intuitions*, and thus is responsive to the arguments of typical people who are upset about e.g. gerrymandering.
I believe that the typical political intuition people carry is that if you win a certain share of the votes, you should get about that share of the power. Some people probably also carry strong winner-take-all biases, and some people strong loss-avoidance biases...
But my suspicion is that these people aren't super common, and that they roughly balance out. On the whole, I suspect most people intuitively think that if 60% of the people support you, you should get something like 60% of the power.
Because I think this is the actual working political intuition people bring to debates about electoral reforms, that's the intuition I am responding to.
And you can see very visibly and annoyingly in the article how concerned I am to speak to this intuition by how I double up every graph: once I show % of votes cast, then I show % of population who voted for the party.
That is, I am agnostic about whether peoples' specific intuition is, "getting 60% of votes should yield 60% of power" or rather, "getting 60% of people to vote for you should yield 60% of power."
But in both cases, I believe the working intuition people bring to politics is that power share should approximately match public support share for a party. I show both ways of calculating this mismatch, but they are variations on the same underlying intuition.
All of that to say....

Enjoy the piece. It took a lot of work. I'm sure somebody will find something in it I did wrong; in a piece this long with this many unfamiliar calculations, there's sure to be something I could have done better.
Ah, and we have our first really good critique! I actually anticipated this one: "partisan bias doesn't matter; ideological bias matters."

That's fair and I actually agree.
But a few things to note here:
1. That's very hard to measure. The *ideal* way we would do this would be to take every Rep and Senator's DW Nominate score and election history, and see if bins of left/middle/right tend to have received more/less votes per seat.
But that would take a truly enormous amount of work. And even then, I'm not sure if it would really tell you anything interesting.
2. While I hear the objection and agree on its validity in principle, in practice the pro-Dem House/Senate bias lasted well into the 1990s, so after the switch among the southern Dems was well under way.
That is, the timing of the change in bias is not super well aligned with the timing of change in party ideologies. So I don't think doing an ideology-adjusted metric would change the aggregate much.
3. Ideologies accomplish their aims through parties. Ideologies matter to people, but parties matter to power. As such, measuring "ideology" is not entirely appropriate.
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Lyman Expand the House of Representatives Stone
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member and get exclusive features!

Premium member ($30.00/year)

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!