Current Events > Vaccine only gives 3% protection after 6 months?

(+) Yeah! (+)

WeeWeiWiiWie posted...

Peer review has nothing to do with that.

I said tend, never said does.

---

badjay posted...

An actual journal article that has been peer reviewed, rather than some layman's misinterpretation of such an article like most news sites tend to put out.

So point me in the right direction where I can get news I can trust

---

Johnson & Johnson is pretty weak and should only be taken if you can't get Moderna or Pfizer.

---

LightHawKnight posted...

I said tend, never said does.

I mean, the vast majority of articles are never directly replicated.

And again, peer review has nothing to do with replication.

---

Honestly, it's bad enough that this came from a pre-published paper, but the fact that they arbitrarily chose a single data point, one that has a non-statistically significant p-value, shows the extent to which science illiteracy makes it into journalism and is propagated quickly.

---

DarkRoast posted...

Honestly, it's bad enough that this came from a pre-published paper, but the fact that they arbitrarily chose a single data point, one that has a non-statistically significant p-value, shows the extent to which science illiteracy makes it into journalism and is propagated quickly.

I think you just don't understand what that p-value represents.

---

+ side, JnJ is a megacorporation with a vested interest in this kind of bad press being punished with lawsuits, so the trashbag article probably won't last long

WeeWeiWiiWie posted...

I think you just don't understand what that p-value represents.

seems like she does tbh

---

WeeWeiWiiWie posted...

p-values test against a null, not the evidentiary strength of an estimate. The test here seems to be against 0, or a hazard of 1.

---

yes, and she's correct in concluding that given there's a 50% chance of finding results that extreme under the assumption that the null hypothesis is correct--which is to say there's no correlation between time and vaccine efficacy--the findings for august are meaningless

if your gripe is with her use of the phrase 'statistically significant', ok

---

Sackgurl posted...

yes, and she's correct in concluding that given there's a 50% chance of finding results that extreme under the assumption that the null hypothesis is correct--which is to say there's no correlation between time and vaccine efficacy--the findings for august are meaningless

if your gripe is with her use of the phrase 'statistically significant', ok

...no, that is incorrect. The test that is resulting in the p-value is whether the protectiveness of the vaccine is distinguishable from no protection.

They found no statistical difference from no protection with the J&J.

---

I am incorrect for my purely statistical definition standpoint. Which I acknowledge. Although that wasn't really what I was trying to say to begin with, but that's my fault. A p-value of 0.5 in this particular case is completely meaningless.

---

WeeWeiWiiWie posted...

...no, that is incorrect. The test that is resulting in the p-value is whether the protectiveness of the vaccine is distinguishable from no protection.

They found no statistical difference from no protection with the J&J.

In one single data point for one month

---

WeeWeiWiiWie posted...

I mean, the vast majority of articles are never directly replicated.

And again, peer review has nothing to do with replication.

Eh then whats the point of science if no one else is confirming the results?

---

Yo just get the pfizer one.

Who cares if I need to get another shot?! I for one enjoy the free bag of mircrochips for you to munch on.

---

The p-value is in reference to the null! It doesn't give you room to judge about how good the estimate of 3% is! In fact, an estimate with the same evidentiary value but with a hazard ratio shifted down by .12 would be statistically significant with this test! So the same amount of evidence, but a 15% for the estimate, and that p-value is very different.

---

Jx1010 posted...

So point me in the right direction where I can get news I can trust

What news are you looking for? Are you looking for news on vaccines? Which vaccine would you like more info on? What exactly are you looking for on the vaccine? Efficacy rates? Side effects? There's journal articles covering anything a general CNN article puts out. It's just stupidly interpreted when you read it from a news website rather than the actual scientific journal article.

---

Jx1010 posted...

So point me in the right direction where I can get news I can trust

Reuters is probably the best mainstream news source, I really like Dr. John Campbell on youtube, he's not providing opinion, he will just read off the latest papers and data points.
https://www.youtube.com/watch?v=wCRj3wygulc

He's also good at sharing the data/articles that debunk misinformation about vaccine safety and other things.

Many other sources are garbage.

I also recommend doing what I/others did with this article. They cited a study, I basically said "ok I'm ignoring the rest of the article and reading the study." And went to the study, then other people doing diligence caught the odd data point and we were able to learn what this study was and wasn't saying.

---

yeah, COVxy, you're right. sorry for arguing with you on this one.

i had to pull out my old stats textbook for this one because i haven't had the pleasure of running tests that experimentally proved the null hypothesis directly

but as the results are presented as a hazard ratio, the null hypothesis is a result of 1

so it makes sense that an experimental result very close to 1 would yield an unusually high p-value

---

LightHawKnight posted...

Eh then whats the point of science if no one else is confirming the results?

There's kind of a lot of angles to come at this from. First, science is typically done with mostly an assumption of good faith. So the goal of replication is explicitly not to see if some other group was bullshitting. So, then the effect of replication simply compounds your confidence in a result. But there are many ways in which you can be confident in a result without direct replication, like the initial evidence is very strong, it builds off of prior strong work, it has multiple results pointing at the same conclusion, etc, etc...

There's something to be said for the long game incentive for publishing work that you think will hold up, particularly after a very short period of time in the field. Short term gain of flimsy results are really only beneficial very early on in a scientist's career. While publishing flimsy results after having been in the field gains you a reputation which leads to decreased attention, citation, and expansion of your work.

---

Sackgurl posted...

yeah, COVxy, you're right. sorry for arguing with you on this one.

i had to pull out my old stats textbook for this one because i haven't had the pleasure of running tests that experimentally proved the null hypothesis directly

but as the results are presented as a hazard ratio, the null hypothesis is a result of 1

so it makes sense that an experimental result very close to 1 would yield an unusually high p-value

No worries, these things get complicated and nuanced.

---

TotallyNotAGirl posted...

Yo just get the pfizer one.

Who cares if I need to get another shot?! I for one enjoy the free bag of mircrochips for you to munch on.

HAHAHAHAHAHA what a completely original joke that nobody has ever said before.

Joeydollaz posted...

JnJ is the real only one that is ORIGINAL in terms of process,

the other two are alot alike.

J&J is basically just AstraZeneca without a second shot.

---

I've seen early results for Pfizer showing a 33% drop in 6 months, its still too early to tell if this will continue at a steady rate or flatten out. Supposedly a third shot could eliminate the need for yearly boosters but again its too early to tell.

---

ReignFury posted...

I've seen early results for Pfizer showing a 33% drop in 6 months, its still too early to tell if this will continue at a steady rate or flatten out. Supposedly a third shot could eliminate the need for yearly boosters but again its too early to tell.

Too early to tell? I thought all the testing for the vaccine was done. They didnt know this?

---

Jx1010 posted...

Too early to tell? I thought all the testing for the vaccine was done. They didnt know this?

You can't do much long term testing until sufficient time has passed.

You can't have 2 year data for something that's only existed for 1 year.

---

Jx1010 posted...

Too early to tell? I thought all the testing for the vaccine was done. They didnt know this?

They dont know how long the antibodies last longterm because the first vaccine was 10 months ago

---

DarkRoast posted...

In one single data point for one month

This is extremely misleading.

Here's their Hazard Ratios plotted across time with 95% confidence intervals:


Which you could get from the table, but your focus on the p-value without context made it hard to see.

Errorbars are larger with J&J because they have fewer patients in that group, but they certainly have enough to do the estimation.

---