Emily Thorson

I never buy luxury brands inconsistent with the characteristics with which I describe myself (seven point scale from strongly disagree to strongly agree)

Guys today is a red-letter day for bad survey questions. Look at this lovely double negative that just rolled in!

Good question! (1) requires too much cognitive work to parse what it's asking for, (2) too much abstract thinking, (3) poorly differentiated response options (e.g. good chance vs strong chance)

The actual survey had probably 20 versions of this question. I almost never satisfice on surveys but I lacked the cognitive focus to interpret each one. Sorry, researchers.

Survey question reading "When deciding whether to pursue a new long-term goal, what is the chance you will overestimate how challenging it will be to achieve?" Answer options: No chance, very slight chance, slight chance, moderate chance, good chance, strong chance, certain

for everyone teaching survey design this fall, great news: a new terrible question example just dropped! (this is from a real survey I just took) 😱

same. just gonna go ahead and put it all on my FSA card.

Or, even worse, accurate.

Survey question asking which of the following best describes your current job title or role, options like "Vice President," "General Manager," and "Individual Contributor"

From a Washington Post reader survey I just took. What a bizarre list of job titles. Also I love the idea that it's critically important to distinguish between the Presidents and Vice Presidents among their readership. TOTALLY different news preferences.

this made me laugh out loud in the middle of a meeting

I'm excited for what I hope will be an increase in LLM-assisted descriptive content analysis...especially looking forward to when someone (who is not me) looks at the topics of news coverage across multiple outlets and time periods.

great call — more experiments and within-subject panels / less emphasis on what the national topline happens to be today while worrying about whether it got the sample right feels like a worthwhile trade for the public polling industry

Agreed. I think they are often BETTER when they hand-write than when they type because they don't try to fancy it up and instead write more like they speak.

I switched to blue books a few years ago. It's fine! Student handwriting is mostly readable, the score distributions are similar, they have not lost the ability to write or concentrate or form coherent sentences. Students who need accommodations take at at our CDR, easy-peasy.

I think figuring out concrete strategies to deal with it is the opposite of letting down one's guard. It's acknowledging that it's a game changer and figuring out how to adapt.

I'm pretty over the doomsaying/complaining about AI and student learning and interested in moving on to the "solutions" part. In-class writing (by hand!) exercises, non-generic paper topics, etc. We can adapt! It might even be fun!

[starts going through all of Drew's coauthors so I can figure out who to high-five at the next APSA]

There is NOTHING on the front page of Fox News about the stock market crash. Nothing at all.

(screenshot taken from the linked paper by @vinarceneaux.bsky.social @m-b-petersen.bsky.social and Mathias Osmundsen)

Pretty horrifying how well the "need for chaos" survey battery describes the Trump/Musk administration's current approach to governing. www.cambridge.org/core/journal...

Also I love that you made a PDF blog.

I totally missed this (this is what happens when you log onto Bluesky only once every six weeks). And yes, I was assuming variation to measure causal effects -- I guess I shouldn't have said "descriptive" when what I meant was "causal inference that is not aimed only at 'proving' a 'theory'"

Work that looks at "who pays attention to what" definitely exists (much more in comm than in PSC, unsurprisingly -- I mean, the ELM is basically about attention) but it's a very scattered literature where the pieces rarely speak to each other.

And the fact that the discipline only rewards causal inference and new theories, not descriptive work. It would not be hard to do a large-N study where participants browse a website and you vary communicator/topic/etc, with the DV being attention. We'd learn a lot. But "where's the novel theory"? 🙄

This is why every political scientist you know has been in a state of panic for the past week. This isn't about policy disagreement, it's about the Constitution being destroyed before our very eyes.

Regardless of your party, is this the future you want for America? Is this the future that the founders wanted? Please contact your elected officials. Tell them that you want to keep this extraordinary democracy we have built.

This is precisely what is happening in the Trump administration. His actions are unconstitutional. Do we want to live in a world where the President, and whoever he appoints, can simply decide what gets funding without Congressional approval?

It is illegal because of our system of checks and balances. Congress has approved that military spending, and it would be unconstitutional for the President to unilaterally decide to end it.

Imagine: in his first week in office, Biden issues an executive order to cut all military R&D. He tells George Soros to carry it out, giving him access to the personal financial information of all military employees. Thousands of soldiers and contractors lose their jobs. This. Would. Be. ILLEGAL. 🧵

This is one of the reasons even “unlikely to succeed” resistance moves (protests, law suits, etc) have value at the moment.

The authors outline some strategies for overcoming indexing and getting the media's attention: (1) outsider counter-spin: protests, personal stories of how this affects Americans (2) on-the-ground footage (aka recordings of Elon's goons/USAID chaos) (3) investigative reporting/leaks.

"When the Press Fails," published in 1990, predicts exactly the situation we're in now. Because the Democrats are largely silent about the Billionaire's Coup, the media don't have "two sides" to cover and so default to repeating whatever line the government feeds them. This is called "indexing."