While the quantitative argument presented here is fairly convincing on its own, I think there's a qualitative argument to make, which you may have intentionally not included so as to not muddy your argument.
It goes like this: progressive candidates often face more brutal primaries, more attacks from within, undermining by the Democratic party, itself, which then makes subpar performance of these candidates more likely in the general election. Let's take the Mamdani situation as an example. He's currently heavily favored to win, with some models suggesting a 75% probability of a win. While that feels large, especially in our era of frequent toss-up elections, note that it's not that different from what 538 predicted for Clinton in 2016, so it's hardly certain.
Now imagine if Mamdani had the backing of the party, all of the notable endorsements, and didn't have massive media machines organized against him. Do all of these things hurt his election chances? Maybe we're at the point where too many mainstream endorsements are actually harmful. But my gut still tells me that the Democratic nominee would be a shoo-in for an election in this anti-Trump political environment, and it's the "Democratic" moderates, themselves, who are preventing this.
"But the value of understanding electability is not in figuring out how to win safe seats by greater margins. It lies in the clues it provides to winning swing voters who could tip the balance in battleground districts. Nominating candidates who are relatively more extreme can cost a party seats, while nominating contenders who are more moderate can gain them seats." - LJ
FWIW, I think he's relatively correct in his prescription and solution of a key problem: battleground states are difficult to win, and nominating moderates relative to progressives can help tip those scales favorable to Dems at large. Given that GEM also found an effect of 1-1.5 points, it seems that at least that is directionally correct. If the effect is overstated, that seems like a technical argument that I'm glad you three are having.
Note on the Aaron Judge metaphor - I think ST's WAR does not tell people that changing their name to Aaron Judge means that they'll do better. I think it communicates "players that bat worse than Aaron Judge could consider trying some of things that Aaron Judge does." And sure, maybe Aaron Judge is so unique and context specific that no batters should consider trying what Aaron Judge uses to be such a good player.
But at the very least, I think it's valuable to study how the Aaron Judge's habits with batting rather than trying to say "Because we don't know what can drive batting performance, there's no value in Max Kepler trying to learn from Aaron Judge."
"Nominating candidates who are relatively more extreme can cost a party seats, while nominating contenders who are more moderate can gain them seats."
I think the problem with this generality is that it assumes that all swing voters are moderates. Having registered a lot of voters, I find that a lot of people don't vote because they don't see anyone who actually represents their views... Considering that most politicians we have today are moderate or Far Right, I think that might tell you something about which views have been cut out of the Overton Window (and why cynicism pervades our political discourse).
"Directionally correct" while claiming a 5x difference in effect size is also known as "incorrect", if you take GEM's model as the baseline. If you just like moderates, you probably don't care about effect size, but it is important to measure accurately for most purposes.
Changing from the most liberal to most consevative candidate potentially increasing voteshare by either 0 or 1-1.5 pct points is just not a large factor no matter how you cut it.
1) Making no comment around who is "correct" given that my preference is for ST's model and I haven't fully read through GEM's article from yday, being "directionally correct" means to me that moderates have tended to outperform partisans. I do care about the accuracy of the effect size, but I remain unconvinced that the adjustments in this model are superior to what is going on at ST.
2) The 1-1.5% moderation outperformance would be from moderate dem to progressive dem, not from progressive dem - far right lunatic.
3) Even a 1-1.5% moderation bonus is a significant given the narrow margins that battleground states operate under.
FWIW, I'm pretty keen on the Dem party becoming more left wing in general.
The ST 7 point advantage is over the entire range of the candidate ideology spectrum, and I assume the cited 1-1.5 is over the same given the authors compare the two. This could be wrong but GEM is paywalled, so unclear.
Further GEM highlights the vast majority of WAR difference comes from safe seats so is not directly applicable to competitive races.
So that means ST's moderation effect is smaller than 7%. That makes the differences between Blue Dogs and Progressives like ~4% for House 2024, which is more in line with I understand GEM was claiming.
Just looking on his twitter, looks like the difference in WAR was optimistically 2.5% (more conservatively 1.4%?) Moderate Dems compared to Average Dems... so not across the entire ideological spectrum.
No, the within-party range-wide effect size is 7 percentage points for ST and 1-1.5 for GEM. That is a difference of about 5x and in both cases not directly applicable to competitive races.
I am unsure what you are saying though, so not clear if that is the crux.
I don't think that's correct for GEM. I am a GEM subscriber so got the full article. He listed the difference between Jared Golden, one of the more moderate (but not most moderate by his metric) and the *average* dem at 1.4 points, the difference between Golden and Omar (who is also pretty liberal but not the most left wing by his metric) as 2.5 points. So total range from most moderate to most liberal is somewhat more than 2.5 points, maybe 3.5? So maybe half that of the ST range, not a 5x difference.
I mean yeah if every single party member moves right the distance between Kamala Harris and Joe Manchin, you would average 1-1.5pp mostly concentrated in safe seats, idk if that sounds like a good trade to you but certainly doesn't to me.
president joe manchin and speaker joe manchin for failed presidential candidate kamala harris and house minority leader hakeem jeffries sounds like an excellent trade, actually
You'd also need to replace Joe Manchin with Thom Tillis and Chris Murphy with Joe Manchin across the board in every even slightly competitive seat (where, again, the result does not necessarily hold as GEM's WAR is concentrated in safe seats)
Seems to me that your finding is the outlier, not SplitTicket. Bailey/Reese 2024 find that ideological moderation was associated with higher vote shares in 2022/2024, with effects between 2 and 4 pct points in competitive districts. That's a lot! Additionally, I can't remember SplitTicket claiming that WAR was *only* due to moderation. Of course they acknowledge that many factors play a role. You seem to claim that moderation doesn't matter - which is inconsistent with the empirical literature. Or you might claim that it's no "silver bullet," which nobody argued making it a straw man. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5169898
This comment doesn't respond substantively to the content.
First, ST-WAR does claim a 7 point moderation advantage, counter to your claim that they did not argue moderation strongly influences success and their writing further emphasizes the point they think it is a major causal factor. Your straw man claim is false.
Second, this article shows multiple analyses demonstrating moderation has little to no advantage. That one more paper shows the opposite does not make these analyses outliers, counter to your claim
Finally, Bailey and Reese (2025) is susceptible to nearly all the critiques in this post, rather than being a different confirmation of ST's methodology. You are saying "look, we tried the same wrong thing twice, you lot doing the right thing are outliers" which is facially ridiculous.
This is not a serious response to any of the substance of the post.
First, this blog post makes claims about the SplitTicket metric that are flat out false. The ST database covers *every* race contested by both parties, so the survivor bias critique is completely meaningless.
Second, ST's original post about the WAR doesn't even mention moderation. They clearly say they "quantify the impact of candidate quality," which of course is about much more than moderation. Not sure where your seven-point claim is coming from, the WaPo piece by Jain doesn't feature it. Original WAR ST Post here: https://split-ticket.org/2025/01/15/our-2024-wins-above-replacement-war-models/
Third, Bailey/Reese is not a WAR measure. How is it supposed to be "susceptible to all the critiques in this post?" Clearly you haven't read it. They use original ideology estimates based on the language candidates use on social media and their campaign websites.
Fourth, contrary to the claims in this post, difference-in-difference identification strategies also show that ideologically moderate candidates are rewarded at the ballot, and extremists are punished. See http://www.chriswarshaw.com/papers/accountability.pdf
Five, effects of 2 to 4 pct points (consistent finding) are absolutely decisive in competitive elections/districts. Nobody (not Jain, nor Yglesias) claimed that moderation is somehow needed in safe districts. Nor did anyone claim that context doesn't matter. Straw man.
1. False: “Although our database covers every race contested by both parties, we’ll limit our analysis to incumbents here for ease of classification1 — while quantifying challenger ideology is quite difficult, it’s much easier to classify Congressional incumbents, as clearer ideological groupings begin to emerge in caucuses.” - https://split-ticket.org/2025/03/17/are-moderates-more-electable/
2. Read the post you are replying to there are several links to their moderation analysis and I linked again in 1
3. That it is not WAR and that they use a novel (poor) ideology measure does mean that Bailey and Reese (2025) is immune to the critiques in this post, for instance the peeking problem
4. Again that a different DiD analysis shows something else does not mean you are responding substantively to the post, you are gish galloping with a literature this is a response to without reading the response.
5. GEM is clear that the vast majority of the small amount of moderation WAR they discover is not relevant to competitive races, you are once again citing past literature without responding to the substance at hand here
Almost farcical. Calling their ideology measure “poor” without any evidence. Also, what the heck is "gish galloping"? The paper linked looks at more elections than Grumbach and Bonica, with a similar DiD design. Clearly you're not interested in discussing anything substantive.
Gish galloping is, e.g., making a comment with five points and when all are responded to, reacting to only the weakest of the responses, as you do here
I would respond but split ticket did a much better job of going through each point in this article. And while I was on the fence, split ticket made a compelling case is why this article is wrong, but way off the mark.
You’ll find it surprisingly easy to claim the literature in your favor if you just throw away any finding that doesn’t already fit your belief. I am not totally convinced by the post either, but they have some pretty serious and legitimate qualms and it is a good piece to include with the others
Question: How does this model define "moderate"? Do they measure candidates on a linear scale, from “Left” to “Center”? Based on what? You need at least two dimensions, one on IDENTITARIAN versus UNIVERSALIST beliefs and messaging, the other on ECONOMIC POPULISM versus NEO-LIBERAL beliefs and messaging.
Just off the top of my head, I would bet that the top performer (accounting for all the other confounding factors) would be Universalist/Populist (Bernie Sanders) the worst would be Identitarian/Neo-Liberal (Corey Booker). The interesting contest would be between a Universalist/Neo-Liberal (what you all seem to identify as “centrist”) and an Identitarian/Populist (AOC?).
Anybody who wants to pursue this line of reasoning, I'd be happy to work with you to develop the parameters.
Thank you for writing this! To build upon the Aaron Judge metaphor, I think it might be even more akin to arguing that joining the Dodgers leads to becoming a better player. It's true players who join the Dodgers tend to do better, but that's because joining a very well-resourced team generally leads to becoming a better player -- not because there's anything magical about the Dodgers themselves. Similarly, joining the well-resourced moderate team may lead to winning more, but if the progressives were as well-resourced then we would observe a similar effect of becoming a progressive.
Focusing on margin of victory as opposed to predicted probabilities of victory also seems problematic to me, and I'm curious if you would agree. AOC, for instance, tends to take positions which might be a little out of step with her electorate because she's in a safe district and is at no risk of losing her seat for taking a slightly unpopular stance. In the long run, this may slightly decrease her margins of victory. Does that really mean she's a worse candidate if her odds of victory remain constant at 99.9?
Yes. If you’re taking positions that do not reflect your constituents and underperform the baseline, it means you are a subpar candidate. A job of a elected official is to represent their voters
Ai ya. Underperforming *at what?* We are talking about getting elected. 67% support and 66% support are 100% equal likelihood to get elected. WAR distinguishing between those, when it is a measure of ability to get elected and note maximize vote share, is very silly.
I don’t think it’s silly at all. If you’re under performing in your district that you are accurately representing the views and desires of the people in your district. Personally, I think that’s a problem.
Matt Yglesias makes the point that in a highly nationalized era you should expect relatively modest effects from candidate moderation. Even if a hypothetical swing voter believes, say, that Jared Golden is truly moderate, the voter will rightly be concerned that the effect of a Democratic majority in the House won't be so moderate. But doesn't this also, Yglesias reasons, reduce the usefulness of this whole line of research? It can't tell you anything at all about the effect that would be produced if the party itself moderated its image, for example by nominating a more moderate candidate for president. It can't even tell you what would happen if there were more moderate candidates nationwide, rather than the very small number we have now. Is there anything wrong with Yglesias's reasoning? Doesn't it suggest we should be rather agnostic about the effects of moderation, given that the smartest version of the hypothesis is extremely hard to test, or even unfalsfiable?
While I greatly appreciate all the back-and-forth number crunching and have plenty of respect for data science, in some ways, the discussion of electoral WAR works to dehumanize what is ultimately an exercise in civic engagement involving real people. It's hard to imagine that Robert Kennedy or Lee Atwater ever approached elections like this - attempts at hacking or gamifying seem to miss the point of the whole thing.
Caro talks about this in Master of the Senate: in the 1950s, people in both parties were using data (specifically, Eisenhower's share of the Black vote, and Black voters' share of the electorate in northern battleground states) to argue that their party needed to embrace civil rights to become the national "majority party". The argument was that as Black voters moved from South to North, they were becoming an increasingly pivotal bloc in key battleground states in the Electoral College, so whichever party moved first on civil rights could secure an Electoral College advantage by winning over Black voters.
In The Victory Lab, Sasha Issenberg discusses how JFK used polling to decide how to respond to attacks on his religion in 1960. Which is to say that using number crunching to try and win elections has been going on for a long time.
Oh, I don't discount that one bit, but as sports-adjacent analytics has moved into politics, it feels less about winning elections and more about trying to win them with the least amount of effort or motivation needed. Embracing civil rights is an outcome of winning elections, which isn't really what Jain I think is suggesting; instead, it's about winning for the sake of winning.
Winning elections is not the end, but none of the ends are possible without it. People in politics generally significantly discount how important it is. If Joe Manchin had decided to retire in January 2018 like he was reportedly considering everything Joe Biden accomplished wouldn't have happened. If Maggie Hassan had won 1000 fewer votes in 2016 there are 15 or 20 million people who wouldn't have healthcare right now. It really doesn't matter how progressive your 47 senators are.
While the quantitative argument presented here is fairly convincing on its own, I think there's a qualitative argument to make, which you may have intentionally not included so as to not muddy your argument.
It goes like this: progressive candidates often face more brutal primaries, more attacks from within, undermining by the Democratic party, itself, which then makes subpar performance of these candidates more likely in the general election. Let's take the Mamdani situation as an example. He's currently heavily favored to win, with some models suggesting a 75% probability of a win. While that feels large, especially in our era of frequent toss-up elections, note that it's not that different from what 538 predicted for Clinton in 2016, so it's hardly certain.
Now imagine if Mamdani had the backing of the party, all of the notable endorsements, and didn't have massive media machines organized against him. Do all of these things hurt his election chances? Maybe we're at the point where too many mainstream endorsements are actually harmful. But my gut still tells me that the Democratic nominee would be a shoo-in for an election in this anti-Trump political environment, and it's the "Democratic" moderates, themselves, who are preventing this.
"But the value of understanding electability is not in figuring out how to win safe seats by greater margins. It lies in the clues it provides to winning swing voters who could tip the balance in battleground districts. Nominating candidates who are relatively more extreme can cost a party seats, while nominating contenders who are more moderate can gain them seats." - LJ
FWIW, I think he's relatively correct in his prescription and solution of a key problem: battleground states are difficult to win, and nominating moderates relative to progressives can help tip those scales favorable to Dems at large. Given that GEM also found an effect of 1-1.5 points, it seems that at least that is directionally correct. If the effect is overstated, that seems like a technical argument that I'm glad you three are having.
Note on the Aaron Judge metaphor - I think ST's WAR does not tell people that changing their name to Aaron Judge means that they'll do better. I think it communicates "players that bat worse than Aaron Judge could consider trying some of things that Aaron Judge does." And sure, maybe Aaron Judge is so unique and context specific that no batters should consider trying what Aaron Judge uses to be such a good player.
But at the very least, I think it's valuable to study how the Aaron Judge's habits with batting rather than trying to say "Because we don't know what can drive batting performance, there's no value in Max Kepler trying to learn from Aaron Judge."
"Nominating candidates who are relatively more extreme can cost a party seats, while nominating contenders who are more moderate can gain them seats."
I think the problem with this generality is that it assumes that all swing voters are moderates. Having registered a lot of voters, I find that a lot of people don't vote because they don't see anyone who actually represents their views... Considering that most politicians we have today are moderate or Far Right, I think that might tell you something about which views have been cut out of the Overton Window (and why cynicism pervades our political discourse).
"Directionally correct" while claiming a 5x difference in effect size is also known as "incorrect", if you take GEM's model as the baseline. If you just like moderates, you probably don't care about effect size, but it is important to measure accurately for most purposes.
Changing from the most liberal to most consevative candidate potentially increasing voteshare by either 0 or 1-1.5 pct points is just not a large factor no matter how you cut it.
1) Making no comment around who is "correct" given that my preference is for ST's model and I haven't fully read through GEM's article from yday, being "directionally correct" means to me that moderates have tended to outperform partisans. I do care about the accuracy of the effect size, but I remain unconvinced that the adjustments in this model are superior to what is going on at ST.
2) The 1-1.5% moderation outperformance would be from moderate dem to progressive dem, not from progressive dem - far right lunatic.
3) Even a 1-1.5% moderation bonus is a significant given the narrow margins that battleground states operate under.
FWIW, I'm pretty keen on the Dem party becoming more left wing in general.
The ST 7 point advantage is over the entire range of the candidate ideology spectrum, and I assume the cited 1-1.5 is over the same given the authors compare the two. This could be wrong but GEM is paywalled, so unclear.
Further GEM highlights the vast majority of WAR difference comes from safe seats so is not directly applicable to competitive races.
So that means ST's moderation effect is smaller than 7%. That makes the differences between Blue Dogs and Progressives like ~4% for House 2024, which is more in line with I understand GEM was claiming.
Just looking on his twitter, looks like the difference in WAR was optimistically 2.5% (more conservatively 1.4%?) Moderate Dems compared to Average Dems... so not across the entire ideological spectrum.
No, the within-party range-wide effect size is 7 percentage points for ST and 1-1.5 for GEM. That is a difference of about 5x and in both cases not directly applicable to competitive races.
I am unsure what you are saying though, so not clear if that is the crux.
I don't think that's correct for GEM. I am a GEM subscriber so got the full article. He listed the difference between Jared Golden, one of the more moderate (but not most moderate by his metric) and the *average* dem at 1.4 points, the difference between Golden and Omar (who is also pretty liberal but not the most left wing by his metric) as 2.5 points. So total range from most moderate to most liberal is somewhat more than 2.5 points, maybe 3.5? So maybe half that of the ST range, not a 5x difference.
in 2024 a 1.5 pp change in vote share would have won democrats the presidency, the house and a senate seat
I mean yeah if every single party member moves right the distance between Kamala Harris and Joe Manchin, you would average 1-1.5pp mostly concentrated in safe seats, idk if that sounds like a good trade to you but certainly doesn't to me.
president joe manchin and speaker joe manchin for failed presidential candidate kamala harris and house minority leader hakeem jeffries sounds like an excellent trade, actually
You'd also need to replace Joe Manchin with Thom Tillis and Chris Murphy with Joe Manchin across the board in every even slightly competitive seat (where, again, the result does not necessarily hold as GEM's WAR is concentrated in safe seats)
that sounds fine to me
I don't see the paper listed on your working papers page. What is the full title?
Seems to me that your finding is the outlier, not SplitTicket. Bailey/Reese 2024 find that ideological moderation was associated with higher vote shares in 2022/2024, with effects between 2 and 4 pct points in competitive districts. That's a lot! Additionally, I can't remember SplitTicket claiming that WAR was *only* due to moderation. Of course they acknowledge that many factors play a role. You seem to claim that moderation doesn't matter - which is inconsistent with the empirical literature. Or you might claim that it's no "silver bullet," which nobody argued making it a straw man. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5169898
This comment doesn't respond substantively to the content.
First, ST-WAR does claim a 7 point moderation advantage, counter to your claim that they did not argue moderation strongly influences success and their writing further emphasizes the point they think it is a major causal factor. Your straw man claim is false.
Second, this article shows multiple analyses demonstrating moderation has little to no advantage. That one more paper shows the opposite does not make these analyses outliers, counter to your claim
Finally, Bailey and Reese (2025) is susceptible to nearly all the critiques in this post, rather than being a different confirmation of ST's methodology. You are saying "look, we tried the same wrong thing twice, you lot doing the right thing are outliers" which is facially ridiculous.
This is not a serious response to any of the substance of the post.
Let's talk about the substance then.
First, this blog post makes claims about the SplitTicket metric that are flat out false. The ST database covers *every* race contested by both parties, so the survivor bias critique is completely meaningless.
Second, ST's original post about the WAR doesn't even mention moderation. They clearly say they "quantify the impact of candidate quality," which of course is about much more than moderation. Not sure where your seven-point claim is coming from, the WaPo piece by Jain doesn't feature it. Original WAR ST Post here: https://split-ticket.org/2025/01/15/our-2024-wins-above-replacement-war-models/
Third, Bailey/Reese is not a WAR measure. How is it supposed to be "susceptible to all the critiques in this post?" Clearly you haven't read it. They use original ideology estimates based on the language candidates use on social media and their campaign websites.
Fourth, contrary to the claims in this post, difference-in-difference identification strategies also show that ideologically moderate candidates are rewarded at the ballot, and extremists are punished. See http://www.chriswarshaw.com/papers/accountability.pdf
Five, effects of 2 to 4 pct points (consistent finding) are absolutely decisive in competitive elections/districts. Nobody (not Jain, nor Yglesias) claimed that moderation is somehow needed in safe districts. Nor did anyone claim that context doesn't matter. Straw man.
1. False: “Although our database covers every race contested by both parties, we’ll limit our analysis to incumbents here for ease of classification1 — while quantifying challenger ideology is quite difficult, it’s much easier to classify Congressional incumbents, as clearer ideological groupings begin to emerge in caucuses.” - https://split-ticket.org/2025/03/17/are-moderates-more-electable/
2. Read the post you are replying to there are several links to their moderation analysis and I linked again in 1
3. That it is not WAR and that they use a novel (poor) ideology measure does mean that Bailey and Reese (2025) is immune to the critiques in this post, for instance the peeking problem
4. Again that a different DiD analysis shows something else does not mean you are responding substantively to the post, you are gish galloping with a literature this is a response to without reading the response.
5. GEM is clear that the vast majority of the small amount of moderation WAR they discover is not relevant to competitive races, you are once again citing past literature without responding to the substance at hand here
Almost farcical. Calling their ideology measure “poor” without any evidence. Also, what the heck is "gish galloping"? The paper linked looks at more elections than Grumbach and Bonica, with a similar DiD design. Clearly you're not interested in discussing anything substantive.
Gish galloping is, e.g., making a comment with five points and when all are responded to, reacting to only the weakest of the responses, as you do here
I would respond but split ticket did a much better job of going through each point in this article. And while I was on the fence, split ticket made a compelling case is why this article is wrong, but way off the mark.
https://split-ticket.org/2025/08/15/deconstructing-war/
You’ll find it surprisingly easy to claim the literature in your favor if you just throw away any finding that doesn’t already fit your belief. I am not totally convinced by the post either, but they have some pretty serious and legitimate qualms and it is a good piece to include with the others
Question: How does this model define "moderate"? Do they measure candidates on a linear scale, from “Left” to “Center”? Based on what? You need at least two dimensions, one on IDENTITARIAN versus UNIVERSALIST beliefs and messaging, the other on ECONOMIC POPULISM versus NEO-LIBERAL beliefs and messaging.
Just off the top of my head, I would bet that the top performer (accounting for all the other confounding factors) would be Universalist/Populist (Bernie Sanders) the worst would be Identitarian/Neo-Liberal (Corey Booker). The interesting contest would be between a Universalist/Neo-Liberal (what you all seem to identify as “centrist”) and an Identitarian/Populist (AOC?).
Anybody who wants to pursue this line of reasoning, I'd be happy to work with you to develop the parameters.
Thank you for writing this! To build upon the Aaron Judge metaphor, I think it might be even more akin to arguing that joining the Dodgers leads to becoming a better player. It's true players who join the Dodgers tend to do better, but that's because joining a very well-resourced team generally leads to becoming a better player -- not because there's anything magical about the Dodgers themselves. Similarly, joining the well-resourced moderate team may lead to winning more, but if the progressives were as well-resourced then we would observe a similar effect of becoming a progressive.
Focusing on margin of victory as opposed to predicted probabilities of victory also seems problematic to me, and I'm curious if you would agree. AOC, for instance, tends to take positions which might be a little out of step with her electorate because she's in a safe district and is at no risk of losing her seat for taking a slightly unpopular stance. In the long run, this may slightly decrease her margins of victory. Does that really mean she's a worse candidate if her odds of victory remain constant at 99.9?
Yes. If you’re taking positions that do not reflect your constituents and underperform the baseline, it means you are a subpar candidate. A job of a elected official is to represent their voters
I think you're missing the point. WAR is about getting elected -- not about your personal beliefs about the job of an elected official.
Yeah, that’s exactly the point. And if you are underperforming, that means you are a worse candidate
Ai ya. Underperforming *at what?* We are talking about getting elected. 67% support and 66% support are 100% equal likelihood to get elected. WAR distinguishing between those, when it is a measure of ability to get elected and note maximize vote share, is very silly.
I don’t think it’s silly at all. If you’re under performing in your district that you are accurately representing the views and desires of the people in your district. Personally, I think that’s a problem.
Great post over at the Silver Bulletin about these guys being full of a bunch of academic BS. Great read and very convincing.
Split ticket ended up responding and it seems pretty convincing: https://split-ticket.org/2025/08/15/deconstructing-war/
Matt Yglesias makes the point that in a highly nationalized era you should expect relatively modest effects from candidate moderation. Even if a hypothetical swing voter believes, say, that Jared Golden is truly moderate, the voter will rightly be concerned that the effect of a Democratic majority in the House won't be so moderate. But doesn't this also, Yglesias reasons, reduce the usefulness of this whole line of research? It can't tell you anything at all about the effect that would be produced if the party itself moderated its image, for example by nominating a more moderate candidate for president. It can't even tell you what would happen if there were more moderate candidates nationwide, rather than the very small number we have now. Is there anything wrong with Yglesias's reasoning? Doesn't it suggest we should be rather agnostic about the effects of moderation, given that the smartest version of the hypothesis is extremely hard to test, or even unfalsfiable?
While I greatly appreciate all the back-and-forth number crunching and have plenty of respect for data science, in some ways, the discussion of electoral WAR works to dehumanize what is ultimately an exercise in civic engagement involving real people. It's hard to imagine that Robert Kennedy or Lee Atwater ever approached elections like this - attempts at hacking or gamifying seem to miss the point of the whole thing.
Caro talks about this in Master of the Senate: in the 1950s, people in both parties were using data (specifically, Eisenhower's share of the Black vote, and Black voters' share of the electorate in northern battleground states) to argue that their party needed to embrace civil rights to become the national "majority party". The argument was that as Black voters moved from South to North, they were becoming an increasingly pivotal bloc in key battleground states in the Electoral College, so whichever party moved first on civil rights could secure an Electoral College advantage by winning over Black voters.
In The Victory Lab, Sasha Issenberg discusses how JFK used polling to decide how to respond to attacks on his religion in 1960. Which is to say that using number crunching to try and win elections has been going on for a long time.
Oh, I don't discount that one bit, but as sports-adjacent analytics has moved into politics, it feels less about winning elections and more about trying to win them with the least amount of effort or motivation needed. Embracing civil rights is an outcome of winning elections, which isn't really what Jain I think is suggesting; instead, it's about winning for the sake of winning.
Winning elections is not the end, but none of the ends are possible without it. People in politics generally significantly discount how important it is. If Joe Manchin had decided to retire in January 2018 like he was reportedly considering everything Joe Biden accomplished wouldn't have happened. If Maggie Hassan had won 1000 fewer votes in 2016 there are 15 or 20 million people who wouldn't have healthcare right now. It really doesn't matter how progressive your 47 senators are.