Email or username:

Password:

Forgot your password?
Terri O 🍁's posts Post Back to profile
Terri O 🍁

I really love this description of a retracted study: not only does it explain what was retracted (turns out men don't generally divorce their sick wives), but also it covers what the error was (a coding problem treated people who left the study as divorced) how it all went down (someone tried to replicate, asked for data and didn't get the same analysis. Contacted the authors and they were horrified and immediately worked to retract).

It's a really nice story of why replication matters and how to be good at science. This is how I was taught science should work, but I rarely come across such good retrospectives.

retractionwatch.com/2015/07/21

#science #PeerReview

21 August at 18:37 | Open on social.afront.org
23 comments
marqle

@terri

This is why I distrust social science data. It's even harder than physical sciences to discover mistakes because it's never acid tested until some assays it themselves.

In physics, you usually find the error because it stops working :)

21 August at 18:39 | Open on mastodonapp.uk
MylesRyden

@terri

I will say this, even in the face of the comment that already exists here...

People like to say things like "science doesn't work" and then go on to name fraud, commercial interests and so on. Which is all true enough.

But when we find that "science doesn't work" it is because "science" itself found the errors. It was other researchers who tried to replicate, reanalyzed, looked at the books, etc to determine that the results were in question.

Figuring out that "science doesn't work" is part of the process of science. And Retraction Watch is an important part of that!

@terri

I will say this, even in the face of the comment that already exists here...

People like to say things like "science doesn't work" and then go on to name fraud, commercial interests and so on. Which is all true enough.

But when we find that "science doesn't work" it is because "science" itself found the errors. It was other researchers who tried to replicate, reanalyzed, looked at the books, etc to determine that the results were in question.

21 August at 22:22 | Open on social.vivaldi.net
tyx

@MylesRyden @terri
Totally, unless you try to find the data to make a decision and realize that the field from which the data come is deep in the reproducibility crisis and funding/publication bias.
Trust is a continuous variable and many fields are waay below the standards in STEM (which is not perfect on itself).
It's great that one wrong paper was retracted, it adds a bit trust, but how many bad papers were published/amplified by media the same day?

21 August at 22:52 | Open on lor.sh
David Mitchell :CApride:

@MylesRyden @terri

I will add that fraud, commercial interests, etc aren’t actually science, they are part of the all too human structures in which we embed science. People fail, science ultimately uncovers the failures.

22 August at 16:43 | Open on mstdn.ca
PJ Coffey

@terri

That's nice. I often see it repeated and it makes me very sad 😔

21 August at 23:18 | Open on mastodon.ie
Moz

@terri it's also the ongoing preconception bias. That's problematic inside science, but as we see from which studies "happen" to go viral, it's a much wider societal problem.

If only news that contradicts prejudices and biases was received the same way.

22 August at 0:01 | Open on fosstodon.org
Iris Young (he/they/she) (PhD)

@terri is the code excerpt rendering weird for just me or for everyone else too? I can't figure out what raw text would have rendered as these i`' bits. (I know very little html.)

Here’s the original coding line:     replace event`i’ = 1 if delta_mct`i’ != 0 | spouse_delta_mct`i’ != 0 And here’s the corrected coding:     replace event`i’ = 1 if (delta_mct`i’ != 0 | spouse_delta_mct`i’ != 0) & delta_mct`i’ != . & spouse_delta_mct`i’ != .
22 August at 1:47 | Open on neuromatch.social
JohnBJohnBJohnB

@iris @terri
that isn’t HTML.
Maybe Perl or R?
I guess the dot means “left the study” - so don’t replace the value if either spouse left the study

22 August at 4:00 | Open on mastodon.social
Morten Grøftehauge

@JohnBJohnBJohnB @iris @terri I don't think those languages have a `replace` keyword? I can't find one at least.

22 August at 7:02 | Open on sigmoid.social
Dan
Iris Young (he/they/she) (PhD)

@Dan @drgroftehauge @JohnBJohnBJohnB @terri re:html, I meant what could have picked up special characters and done something terrible and wrong with them in the article rendering. Agree Stata looks likely for the programming language. Looks vaguely SQL-y to me and that's as far as I interrogated it.

22 August at 8:12 | Open on neuromatch.social
Iris Young (he/they/she) (PhD)

@Dan @drgroftehauge @JohnBJohnBJohnB @terri I think the i came from html <i> tags and the quotes and backticks from escaping the wrong characters. Which still doesn't answer what should have been inside those tags.

22 August at 8:20 | Open on neuromatch.social
Irenes (many)
Slideruth :cascadia:

@terri That's great! I had heard of this study and it made me sad, but hearing that (a) it was wrong, and (b) that the researchers did a good job of finding and fixing their error, is wonderful news! Thanks for sharing it.

22 August at 5:53 | Open on social.seattle.wa.us
M0KHR

@terri unless you are a US Republican leader?

22 August at 6:54 | Open on mastodon.radio
Tim Ward ⭐🇪🇺🔶 #FBPE

@terri Yeahbut ... and yes, I know this is 20-20 hindsight ... surely that obviously suspicious number should have flagged itself up to someone doing a sanity check?

22 August at 10:57 | Open on c.im
selmins

@terri please note that still

"What we find in the corrected analysis is we still see evidence that when wives become sick marriages are at an elevated risk of divorce, whereas we don't see any relationship between divorce and husband's illness. We see this in a very specific case, which is at the onset of heart problems. So basically its a more nuanced finding. The finding is not quite as strong"

22 August at 12:53 | Open on mastodon.social
wb x64

@selmins @terri that's interesting, I wonder why specifically heart problems and not something more traumatizing like cancer. Maybe given how invisible and subtle heart problems are, it can breed miscommunication and resentment vs the more socially protected and visible chemotherapy etc?

22 August at 17:24 | Open on glitch.social
Gabriel Pettier

@terri thanks for sharing, and to the scientists who tried to reproduce it, i believed that to be a fact, good to know the science, at least doesn't support it (anymore).

22 August at 15:05 | Open on mas.to
Lady of Mystery and Science

@terri
So Raphael ‘Ted’ Cruz is just an asshole, and it’s not genetically evolved behavior.

22 August at 15:54 | Open on xeno.glyphpress.com
Only Ohm

@terri So was the retracted study the only primary source of the widely-repeated claim?

22 August at 16:26 | Open on mas.to
Joan Albright

@terri The most ironic thing for me reading this is that my sister's husband divorced her while she was in chemo.

22 August at 16:55 | Open on mstdn.games
L'égrégore André ꕭꕬ

@terri I'm really glad you posted this, cause I hadn't seen the retraction before! Thank you!

[* Edit because of autocorrect]

22 August at 16:57 | Open on mastodon.social
Go Up