It has nothing to do with believing and everything to do with evidence. The available evidence suggested that she would underperform relative to last year. I'm not a huge believer in prediction markets but for what it's worth Koizumi’s chances in Polymarket were something like 80% at the start of the day. Clearly there was a combination of (a) either a double-digit polling miss or late shift among voters, (b) shy Takaichi supporters in the parliamentary party and/or a polling miss, and (c) her over performance in rank and file voting led to a cascade of parliamentary support. That was going to be her best chance of winning; my mistake was trusting that the available polls -- which were all consistent -- were correct.
I know, Tobias. Still and all, if you had predicted a Takaichi win, I would now be singing your praises and hawking subscriptions to your substack all over town. But the pundits didn’t see it coming any more than us regular folk.
I think the polling was inaccurate because Ishiba (and then subsequently Koizumi) got a lot of support from centrist or centre-left opposition voters who claimed to support the LDP but were not dues-paying members qualified to vote in the election. I imagine many of the survey respondents did not realize that they were not able to vote in the election, which is always a major difficulty with polling elections where the electorate is so limited.
This result is also a defeat for Tobias Harris — and all the other pundits who couldn’t quite believe this would happen.
It has nothing to do with believing and everything to do with evidence. The available evidence suggested that she would underperform relative to last year. I'm not a huge believer in prediction markets but for what it's worth Koizumi’s chances in Polymarket were something like 80% at the start of the day. Clearly there was a combination of (a) either a double-digit polling miss or late shift among voters, (b) shy Takaichi supporters in the parliamentary party and/or a polling miss, and (c) her over performance in rank and file voting led to a cascade of parliamentary support. That was going to be her best chance of winning; my mistake was trusting that the available polls -- which were all consistent -- were correct.
I know, Tobias. Still and all, if you had predicted a Takaichi win, I would now be singing your praises and hawking subscriptions to your substack all over town. But the pundits didn’t see it coming any more than us regular folk.
Alas, much as I wish otherwise, it's impossible to bat 1.000. The only thing to do is learn where I went wrong and try better next time.
Another in-depth and timely article, Tobias. Thank you for the background and your analysis. Exciting times.
I think the polling was inaccurate because Ishiba (and then subsequently Koizumi) got a lot of support from centrist or centre-left opposition voters who claimed to support the LDP but were not dues-paying members qualified to vote in the election. I imagine many of the survey respondents did not realize that they were not able to vote in the election, which is always a major difficulty with polling elections where the electorate is so limited.
That's not what the polling said though. Koizumi did better when looking at LDP supporters only.