This blog auto-posts. But I’m not on twitter. You can tweet at me all you want; I won’t hear it (unless someone happens to tell me about it).

So if there’s anything buggin ya, put it in a blog comment.

Posted by Andrew on 5 October 2017, 9:59 am

This blog auto-posts. But I’m not on twitter. You can tweet at me all you want; I won’t hear it (unless someone happens to tell me about it).

So if there’s anything buggin ya, put it in a blog comment.

## Recent Comments

- Carlos Ungil on The statistical significance filter leads to overoptimistic expectations of replicability
- Shravan Vasishth on The statistical significance filter leads to overoptimistic expectations of replicability
- Paul Alper on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Shravan Vasishth on The statistical significance filter leads to overoptimistic expectations of replicability
- Strangetruther on Write your congressmember to require researchers to publicly post their code?
- Sander Greenland on The statistical significance filter leads to overoptimistic expectations of replicability
- Anonymous on The statistical significance filter leads to overoptimistic expectations of replicability
- Dale Lehman on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Andrew on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Andrew on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Sander Greenland on The statistical significance filter leads to overoptimistic expectations of replicability
- Dale Lehman on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Jeff on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Sameera Daniels on The statistical significance filter leads to overoptimistic expectations of replicability
- Jonathan Baron on Click here to find out how these 2 top researchers hyped their work in a NYT op-ed!
- Keith O'Rourke on The statistical significance filter leads to overoptimistic expectations of replicability
- Bill Jefferys on The anthropic principle in statistics
- Carlos Ungil on The statistical significance filter leads to overoptimistic expectations of replicability
- Shravan Vasishth on The statistical significance filter leads to overoptimistic expectations of replicability
- Carlos Ungil on The statistical significance filter leads to overoptimistic expectations of replicability

## Categories

Nothing bugging me. Just wanted to say that I’m looking forward to hearing your keynote at SSI!

Not you? https://mobile.twitter.com/statmodeling

Sure, but obsiously these are automatic tweets posted by the blog software. Andrew seems to ignore his account.

So all the retweets and replies will go unnoticed. Sad!

Many of my friends have asked me why I’m not on Twitter, and I tell them it’s because it limits the number of characters you can tweet to 14

Not to mention the irony of a tweet stating “I’m not on Twitter.”

Sometimes typos suggest good words: e.g., “obsiously” = both obsequious and obvious”? Or perhaps “Not sure if obvious or just obsequious”?

Should have been “obviously”, obsequiously… ;-)

After you posted about abandoning statistical significance, I was wondering what your impression of work in the “Behavioural Economics” field by organisations like the Behavioural Insights Team is? Most of their work is based on running RCT’s in policy and they rely heavily on p-values when highlighting “positive” results. Is this less of a problem in your view due to strong experimental setup and use of RCT’s or do you still see it as a problem regardless?

Kp:

There are two things going on here:

1. I am suspicious of “behavioral economics” in general in that it seems like a step backward from a theoretical or conceptual perspective; see discussion here.

2. I haven’t looked carefully at the particular studies you’re referring to. But, yes, often it does seem that “evidence-based policy” is up there with “drug-free school zone,” “best linear unbiased prediction,” and “Holy Roman Empire” as something rarely to be believed. As I’ve written various times recently, randomized controlled trials are fine, but if effects are small and measurements are highly variable, all the randomization in the world’s not gonna save ya. Bias and variance, baby, bias and variance.

I was also under the reasonable assumption that you would see your tweet responses. Thanks for clarifying. I enjoy your blog very much & respect your views. We have some similar viewpoints from a cursory study. I too am skeptical of ‘behavioral economics. Be back.

This post is filed under decision theory…

Hi Andrew. Your message popped up on my twitter too. I have responded here: http://andrewgelman.com/2017/10/04/worry-rigged-priors/#comment-578758 . Huw.