Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Justin Wolfers Profile picture
@JustinWolfers
Sep 2, 2020 • 3 tweets • 2 min read • Read on X
My latest @nytimes piece: In normal times the official inflation statistics overstate the rate of inflation. But in this upside-down pandemic economy, the usual biases are reversed, and the official numbers almost certainly understate actual inflation.
nytimes.com/2020/09/02/bus…
@nytimes It follows that many of our key economic statistics -- like real GDP, real wages, and poverty -- rely on incomplete inflation adjustments, and almost certainly overstate our economic health.
@nytimes Also, for those teaching intro to macro courses, a deep dive into how Covid shapes the cost of living is a great teaching opportunity for digging into biases in inflation measurement. #teachecon

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Justin Wolfers

Justin Wolfers Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @JustinWolfers

Justin Wolfers Profile picture
Feb 17
One thing I've learned to do when I have questions about social security number holders who are age 100 or older is to look up the SSA Inspector General audit report, "Numberholders Age 100 or Older Who Did Not Have Death Information on the Numident."

oig.ssa.gov/assets/uploads… x.com/elonmusk/statu…
It's a gripping read. It tells me, for instance, that 98 percent of these folks have received no payments. Image
Why are there dead people on (this one table of) the social security database? They died before the use of electronic death records. Image
Read 4 tweets
Justin Wolfers Profile picture
Nov 4, 2024
This counter-response essentially says that any form of weighting in survey research is herding. If so, I love herding!

He's right about the motivation: All weighting is done to ensure that you don't get crazy results. But that's a feature of a good poll, not a bug!
After all: Is there a principled difference between weighting on age (to ensure that your sample includes youngs and olds) and weighting on past vote (to ensure you get folks from across the political spectrum)?
Both age and past vote are:
- Predetermined (before this poll)
- Non-manipulable
- Though self-reported
- And we have good population estimates to weight them to.

What principle would make one of these a legitimate survey design weight and the other "herding"?
Read 9 tweets
Justin Wolfers Profile picture
Nov 4, 2024
Lately @NateSilver has been arguing that pollsters are "herding" —tweaking or hiding their results to avoid publishing outliers.

I don't know any (reputable) pollsters who do this.

And I think the problem here is Nate making a simple math/stats error 🧵 Image
Let's explore binomial distributions and the standard errors of weighted samples.

TL, DR: Weighting across groups with different voting patterns should change how you calculate confidence intervals.

Fail to do so, and you'll falsely accuse pollsters of herding.
First, what @NateSilver538 does:

In a “vanilla” poll—a simple random sample—the standard error of an estimate of a candidate’s vote share, p, in a two horse race is √[p(1-p)/n].

This is the perspective that animates Nate's analysis of "herding". Image
Read 34 tweets
Justin Wolfers Profile picture
Oct 24, 2024
Focusing for a moment on the POLICIES at play.

Voters were asked whether they supported 100 different policies, but weren't told whether the policy was from Trump or Harris.

Nearly all of Harris' policies got majority support from the public, and... 🧵 Image
Far fewer of Trump's policy proposals garnered majority support, and almost none were broadly popular (say, >75% support). Image
Most undecided voters like most of Harris' policies (left graph).

Most undecided voters dislike most of Trumps policies (right graph). Image
Read 5 tweets
Justin Wolfers Profile picture
Oct 13, 2024
Senator, I think it's worth acknowledging that violent crime actually has falllen, and indeed it is lower today than it has been for decades. Image
This decline in violent crime is evident in not just the FBI reports, but also an independent survey by the BJS. Image
If you don't trust data from G-men, the decline in homicide rates the FBI reports is also evident in a count of death certificates in which coroners cute homicide as the cause of death. Image
Read 6 tweets
Justin Wolfers Profile picture
Oct 4, 2024
Wowza: HUGE jobs report.

September payrolls grew +254k, well above expectations.

August payrolls revised up +17k to +159k, and July revised up +55k to +144k.

Unemployment fell to 4.1%

This economic expansion that is motoring along.
Honestly, there's not much to say here other than that fears the job market had slowed turned out to be a statistical illusion due to incomplete data.

Over the past three months, payrolls is motoring along at +186k per month, on average, which is pretty much where you want it.
(The conspiracy theories on revisions to economic numbers confuse me, and I can't remember whether numbers getting revised up to look good is evidence of a conspiracy, or numbers later get revised down is evidence of an initial conspiracy that falls apart.)
Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

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

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

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

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(