Twitter

[kenberg]

 

 

My comment was about Musk's purchase of Twitter - that it is not nearly as big of deal as the bias-confirmation algorithms of Facebook, which is part of social media in general. I apologize if I broke your train of thought.

 

This could lead to a very interesting discussion. All of us have biases, at least that is my biased view, and no doubt that affects our view of events. If we are speaking one-on-one with a person, possibly these biases can be acknowledged, navigated, dealt with. With a bot? Nope.Whatever the merits of bots, having at least a partially open mind isn't one of them.

 

We have all had frustrations with bots. Here is one of mine, nothing to do with politics. I had a routine doctor's appointment, I got an email saying that for (their) convenience I should send them photos of my insurance cards and my driver's license and they gave me instructions on how to do this via smartphone. Well, I had lost my cellphone and had not at that time bought another but I figured no problem. I scanned the stuff with my printer/scanner, I downloaded it as a jpeg onto the computer, I opened the app or whatever that they said to use, it said I could attach the file so I attached it, I sent it, or rather I clicked on send. Nope. By smartphone yes, any other way no, it won't take it. The staff at the doctor's office, they were humans, had no trouble accepting my cards.

 

Five years or so ago, in my late 70s, I was pleased to be able to help some of my fellow senior citizens cope with computer issues that flustered them Now I often just throw up my hands in despair. I really like humans.

 

Any algorithm will be biased. With humans you can at least have a hope of negotiating. With bots, no.

 

I suppose there are studies of the bias-confirmation algorithms of FB. Do you have a reference? Or examples?

[y66]

re: bias -- Matt Yglesias puts it thusly: "All news is bad news. And the really bad news is that it's your fault. (June 2021)"

 

Bruce Sacerdote, Ranjan Sehgal, and Molly Cook published a research paper at the beginning of the year showing that media coverage of Covid-19 was biased toward negative stories.

 

In turn, the paper generated media coverage which I saw processed largely in terms of partisan politics, with conservatives assimilating it to larger critiques of ideological bias and progressives snarking that of course you emphasize the bad news about a virus that kills millions of people globally.

 

But the study was more careful and more interesting than that. They note, for example, that scientists developed and regulators approved a number of different safe and effective vaccines on an unprecedented timeline. That’s good news. And the worse you think the pandemic is, the better the vaccine news is. And when they say that coverage accentuated the negative, that’s what they mean — not that the press forgot to mention the upside of mass death, but that stories about progress in combatting the illness got underplayed. It would be as if newspapers in World War II decided they weren’t that interested in Allied victories.

 

It’s also not a partisan story. The authors aren’t following the convention where “the media” is taken to exclude incredibly popular and influential media outlets simply because they slant right. Fox News and the New York Post are in the database. What they find is that “the most influential U.S. news sources are outliers in terms of the negative tone of their coronavirus stories and their choices of stories covered,” and “we are unable to explain these patterns using differential political views of their audiences or time patterns in infection rates.”

 

In other words, conservative and liberal outlets alike emphasized the negative. The intra-media difference is that the biggest and most influential outlets were more negative. And within those outlets, “the most popular stories … have high levels of negativity for all types of articles.”

 

In other words, the negativity bias in Covid coverage didn’t come from the liberal media being out to get Trump. It came from you, the reading and viewing public, who strongly prefer to consume negativity-inflected stories on all types of topics, creating a situation where the biggest and most influential media brands have gotten really good at delivering the negativity that the people crave.

 

I think a lot of people on the American right sincerely believed that gloom-inflected media coverage of the pandemic in 2020 was designed to stick it to Donald Trump.

 

And certainly, there was an aspect of partisan contestation around the pandemic.

 

When you have the President of the United States saying “it will all work out well. In particular, on behalf of the American People, I want to thank President Xi!” (January 24), “the 15 [cases] within a couple of days is going to be down close to zero” (February 26), “we’re doing a great job with it and it will go away” (March 10), two things happen. First, anyone offering a more realistic assessment of the situation becomes de facto anti-Trump because Trump is childish and reacts to unwelcome factual information as criticism. Second, negativity per se becomes anti-Trump resistance, so you get stories like “Georgia’s Experiment in Human Sacrifice” published in late April.

 

But on the conservative theory of this, Biden’s inauguration should have prompted a new era of joyful media optimism. In fact, we’ve seen the reverse. In February, a lot of prominent stories were warning that vaccines might not block Covid transmission. In March, a widely discussed New York Times story darkly warned that “experts now believe” herd immunity would never arise in the United States, based on a logic that seemed to completely ignore acquired immunity on the part of people who get sick.

 

This is a story about how fewer people are dying of Covid-19:

 

https://www.nytimes.com/2021/05/31/us/coronavirus-deaths-vaccine.html

 

The Pandemic in the U.S. Has Vastly Improved. For These Families, the Worst Has Just Begun.

 

Families losing a loved one to the coronavirus now described a surreal, lonely kind of grief, as the threat from the pandemic lessens in the United States.

This story, obviously, is not false. Even with the daily death toll far lower than what it was just a few weeks ago, there are still hundreds of people dying. That’s a very sad event for their families. If you or someone you love is personally gravely ill right now, then it’s true that “the worst has just begun” for you. But why the interest in such a negative framing? Is it remotely plausible it’s a deliberate effort to sabotage Joe Biden? Of course not. It’s business. This is what the people want.

[y66]

re: "It's business"

 

From What to Know About Tucker Carlson’s Rise by Nicholas Confessore at NYT:

 

Here is the “Tucker Carlson Tonight” playbook: Go straight for the third rail, be it race, immigration or another hot-button issue; harvest the inevitable backlash; return the next evening to skewer critics for how they responded. Then, do it all again. This feedback loop drove up ratings and boosted loyalty to Fox and Mr. Carlson.

 

What it did not do was endear Mr. Carlson to advertisers. As blue-chip sponsors fled, Fox filled the space with in-house promos — using Mr. Carlson’s popularity to push other Fox shows — and direct-to-consumer brands like MyPillow, whose chief executive is a major promoter of Mr. Trump’s stolen-election lie.

 

Last May, after promoting the white supremacist “replacement” theory, Mr. Carlson had half as many advertisers as in December 2018. But he brought in almost twice as much money.

[Winstonm]

This could lead to a very interesting discussion. All of us have biases, at least that is my biased view, and no doubt that affects our view of events. If we are speaking one-on-one with a person, possibly these biases can be acknowledged, navigated, dealt with. With a bot? Nope.Whatever the merits of bots, having at least a partially open mind isn't one of them.

 

We have all had frustrations with bots. Here is one of mine, nothing to do with politics. I had a routine doctor's appointment, I got an email saying that for (their) convenience I should send them photos of my insurance cards and my driver's license and they gave me instructions on how to do this via smartphone. Well, I had lost my cellphone and had not at that time bought another but I figured no problem. I scanned the stuff with my printer/scanner, I downloaded it as a jpeg onto the computer, I opened the app or whatever that they said to use, it said I could attach the file so I attached it, I sent it, or rather I clicked on send. Nope. By smartphone yes, any other way no, it won't take it. The staff at the doctor's office, they were humans, had no trouble accepting my cards.

 

Five years or so ago, in my late 70s, I was pleased to be able to help some of my fellow senior citizens cope with computer issues that flustered them Now I often just throw up my hands in despair. I really like humans.

 

Any algorithm will be biased. With humans you can at least have a hope of negotiating. With bots, no.

 

I suppose there are studies of the bias-confirmation algorithms of FB. Do you have a reference? Or examples?

Ken,

I have no ready links but have read about the FB algorithms that they are designed to keep the user engaged by delivering articles based on your click (reading) interests. If you click on QAnon nonsense and continue that trend that is the type of articles FB will post on your account and you end up in a biased bubble.

 

PS: Here is something I looked up: https://www.scientificamerican.com/article/biases-make-people-vulnerable-to-misinformation-spread-by-social-media/

Edited April 30, 2022 by Winstonm

[kenberg]

Ken,

I have no ready links but have read about the FB algorithms that they are designed to keep the user engaged by delivering articles based on your click (reading) interests. If you click on QAnon nonsense and continue that trend that is the type of articles FB will post on your account and you end up in a biased bubble.

 

PS: Here is something I looked up: https://www.scientif...y-social-media/

 

The Scientific American article is interesting enough although the title "Biases make people vulnerable to misinformation spread by social media" is hardly one of those "Oh my, who would have thought that" things. in the closing paragraph, they point out, again obviously, that technology alone will not solve the problem.

 

What will solve the problem? There is no clear answer to that but I think a good starting place is to recognize that we all filter information through our own view of the world. We have more influence on ourselves than we do on other people so we could work on ourselves. Be skeptical of statistical arguments, especially of statistical arguments that support our views.

 

There was an article in WaPo the other day that used perfectly good statistics (afaik) but here is a quote: "A recent New England Journal of Medicine study found that while omicron was dominant, effectiveness against symptomatic disease after two Pfizer doses was about 65 percent at two to four weeks, dropping to under 9 percent after 25 or more weeks. " Ok, but 65% effective means what? Some number was divided by some other number and the result was 0.65. What was the numerator, what was the denominator? The author is by no means a scammer or troll or whatever. But before I make any decisions based on this 65% I would like to know what quantity is 65 % of what other quantity. But of course I do not actually plan on doing that. I plan to get my second booster next week, with 65% meaning whatever it means, because I have come to the possibly biased opinion that it's a good idea. I can't prove it's a good idea, I don't really understand the 65 %, but I will be getting my second booster. Most people make most decisions, including important ones, in the way that I am making mine. It seems right to us.

Skepticism is an extremely useful trait. But that includes skepticism about skepticism.

Social media presents us with information. Some of it is correct. I do not think this is an easy problem to deal with.

[Winstonm]

The Scientific American article is interesting enough although the title "Biases make people vulnerable to misinformation spread by social media" is hardly one of those "Oh my, who would have thought that" things. in the closing paragraph, they point out, again obviously, that technology alone will not solve the problem.

 

What will solve the problem? There is no clear answer to that but I think a good starting place is to recognize that we all filter information through our own view of the world. We have more influence on ourselves than we do on other people so we could work on ourselves. Be skeptical of statistical arguments, especially of statistical arguments that support our views.

 

There was an article in WaPo the other day that used perfectly good statistics (afaik) but here is a quote: "A recent New England Journal of Medicine study found that while omicron was dominant, effectiveness against symptomatic disease after two Pfizer doses was about 65 percent at two to four weeks, dropping to under 9 percent after 25 or more weeks. " Ok, but 65% effective means what? Some number was divided by some other number and the result was 0.65. What was the numerator, what was the denominator? The author is by no means a scammer or troll or whatever. But before I make any decisions based on this 65% I would like to know what quantity is 65 % of what other quantity. But of course I do not actually plan on doing that. I plan to get my second booster next week, with 65% meaning whatever it means, because I have come to the possibly biased opinion that it's a good idea. I can't prove it's a good idea, I don't really understand the 65 %, but I will be getting my second booster. Most people make most decisions, including important ones, in the way that I am making mine. It seems right to us.

Skepticism is an extremely useful trait. But that includes skepticism about skepticism.

Social media presents us with information. Some of it is correct. I do not think this is an easy problem to deal with.

 

The issue I have is that the media giants like FB, at least in the past, have not controlled input. If someone, for example, had right-wing leanings and clicked on news article about right-wing interests, that client would eventually be flooded with all sorts of misinformation and basically, propaganda. There would be no balance, no need to scroll around to find other articles similar - it's just one after another after another. That is the bubble that is talked about. It can be either political party, or no political party. It is based on interests and clicks. The object is to keep you on your device or computer, clicking as long as possible, regardless of content or outcome.

 

Really, it is the end result of unbridled capitalistic goals instead of those goal modified by personal morals. In a sense, it is still the 1990s but not a blatant as Forbes' "he who dies with the most toys wins." That part is now written in personal diaries only, or whispered at CPAC.

 

 

[kenberg]

Winston,

You, I, and probably most people agree that there is a huge problem with some bot somewhere seeing that we clicked on something and then passing our name on to his fellow bots who want us to buy or think or do something, and so they send us a bunch of crap. And there are other slick moves. Just yesterday Becky got a FB request from a woman she knew some time ago before she had moved to Delaware, asking her to friend her on FB. Becky said sure. Uh oh. Apparently, the friend's account had been hacked. After the friending, Becky then received a bunch of crap and links to get more crap. So don't go to the links, do some deletes and hope it all works, It apparently has. Me, I just stay off of FB. We are both pretty careful but these bots are clever.

 

Last night we watched an old movie, Clash By Night. It sort of pushed against the censorship standards of 1952 and I thought of the comparison with algorithmic censorship today. In the movie Mae (Barbara Stanwyck) is married to Jerry (Paul Douglas) and is about to begin an affair with Earl (Robert Ryan). What are the censorship issues? Well, in one scene we see Jerry in bed sleeping, Mae is at the window in their bedroom contemplating what to do. We can tell from the pillows that she had been in bed beside Jerry. The censorship rules at the time were that you could not show a man and a woman lying in the same bed, even if they are man and wife and they are snoring. But Mae was out of bed contemplating an affair so this was ok. I see this as robotic censorship even if done by humans. Yes, Mae can be contemplating, and later having, an affair, this is ok as long as you do not show Mae and Jerry in the same bed sleeping. And oh, we know the end result will not be that Mae and Jerry get a divorce, this was not an approved solution in 1952. Maybe one of them will kill the other, that would be ok, but divorce is out. Well, divorce is almost put. Earl gets a divorce during the movie, just before he begins his affair with Mae. Earl's wife has been gone for a couple of years having various affairs, finally Earl gets a divorce, but he makes it clear that he is totally crushed. Not by his wife leaving and having affairs, rather it is due to his finally getting a divorce. He is crushed by the divorce. He is required to be crushed, or the censor will nix the movie. Robotic rules lead to idiocy. The movie is ok, not great but not bad, Marilyn Monroe has a supporting role. The movie was one year before Gentlemen Prefer Blonds.

 

We have to deal with problems created by Twitter, FB, and such, but somehow the solution, if there is one, has to come from society itself. The robots can't do this for us. They can help. Or perhaps they can help.

[y66]

From Lunch with Jonathan Haidt at FT

 

On the morning of the crisp, sunny April day I am meeting him, a long essay that Haidt has been working on for three months, “Why the Past 10 Years of American Life Have Been Uniquely Stupid”, has been published in The Atlantic. In the essay, which he calls “the most important thing I’ve ever written”, Haidt argues that social media is having a devastating impact on society. He invokes the parable of the Tower of Babel, in which God, “offended by the hubris of humanity”, makes the people unable to communicate.

 

“Our institutions are malfunctioning because of the way that social media amplifies performance and moralism and mob dynamics, which brings the normal process of dissent to a grinding halt,” he tells me.

 

Haidt, conspicuously, does not use the term “cancel”, but does talk about social media having armed us with “darts”, which he describes as “attempt to shame or punish someone publicly while broadcasting one’s own virtue, brilliance or tribal loyalties”. These darts cause “pain but no fatalities”, yet are enough to have had a chilling effect on discourse. He criticises both left and right in the essay, but “it was the “younger progressive activists who did most of the shooting”, he writes, giving institutions a “chronic fear of getting darted”, thus making them “stupider”.

 

While some feel Haidt focuses too much of his fire on the left, he has only ever voted Democrat. “I cannot imagine voting Republican because the Republican party has completely lost all sense of constitutional responsibility and has lost all touch with conservatism. I have a lot of respect for liberalism, but there’s a lot of illiberalism on the left, and I have a lot of respect for conservatism, but there’s not much conservatism left on the right,” he says.

I put it to Haidt that the Babel analogy seems to presuppose a time when there was some kind of unity of knowledge, which has surely never been the case. “I know that there never was a golden age in which we all loved each other and spoke the same language,” he clarifies. “But there’s a difference between normal divisions of conflict and hatred, versus the complete insanity that we’re going through now, in which companies and states are at war with each other . . . There is a qualitative difference.”

 

As well as the impending downfall of liberal democracy, what concerns Haidt most about social media is its impact on young teenagers. He cites data showing rates of anxiety, depression and self-harm suddenly surging in the early 2010s, and “correlational and experimental studies” suggesting a link with the use of the platforms.

 

So what is to be done — do we need government intervention? Haidt, reluctantly, thinks we do. He does not believe in more content moderation, but does think children should not be allowed on to these platforms until they are 16, and that companies should be held responsible for enforcing this. He also thinks social media companies should have to carry out user verification in a similar manner to the way banks do know-your-customer checks.

 

But there are simpler, non-governmental interventions that could help too. “If everybody would just cut their social media use by 50 per cent, they would generally be happier and our problem would be ameliorated because the whole thing is built on us contributing content.”

 

I later email Haidt to ask his thoughts on Elon Musk buying Twitter — could this be a good thing for free speech? Haidt tells me people are “barking up the wrong tree” by focusing on this: the problem with the platform is not the content but the “enhanced virality” that spreads it, rewarding the inflammatory, the false and the stupid. Serial tweeter Musk, it strikes me, is unlikely to help with that.

 

After our latkes, we move on to smoked sable — black cod — and bagels, which have arrived on a three-tiered stand, the fish at the top, the bagels at the bottom, with a plate of sliced cucumber, tomatoes, capers and dill in the middle.

[kenberg]

As to "children should not be allowed on to these platforms until they are 16", I am not sure. And that means not sure. Maybe it's a good idea but I am not sure.

When I was 12 or 13 I found a book in the library that explained how I could make my own fireworks. I was not really forbidden to read it but definitely forbidden to act on it. Mostly I read what I wanted to read. But life was simpler for kids in 1952, no doubt about that.

We all need to give our trust slowly, this should be learned early, so I think young people need to be encouraged to read varied material and read it with a questioning mind.

But there is cyber-bullying and it needs to be dealt with. Stopping a 16-year-old from doing what he wants to do is difficult. Some break into houses, some beat up other kids for the fun of it, and cyber-bullying should be thought of as being at that level of unacceptability or maybe even worse.

. “I know that there never was a golden age in which we all loved each other and spoke the same language,” he clarifies. “But there’s a difference between normal divisions of conflict and hatred, versus the complete insanity that we’re going through now, in which companies and states are at war with each other . . . There is a qualitative difference.”

Yep. that's how I see it.

[barmar]

Interesting topic. Where to start.

 

Just to pick up this Matt Yglesias person on one thing. Twitter is not at present a private company. One of the things Mr Musk is considering is taking a listed company private, and all that entails

I believe the distinction he was making is private vs. government, not privately-owned versus traded on a stock exchange.

[pilowsky]

There was an article in WaPo the other day that used perfectly good statistics (afaik) but here is a quote: "A recent New England Journal of Medicine study found that while omicron was dominant, effectiveness against symptomatic disease after two Pfizer doses was about 65 percent at two to four weeks, dropping to under 9 percent after 25 or more weeks. " Ok, but 65% effective means what? Some number was divided by some other number and the result was 0.65. What was the numerator, what was the denominator? The author is by no means a scammer or troll or whatever. But before I make any decisions based on this 65% I would like to know what quantity is 65 % of what other quantity. But of course I do not actually plan on doing that. I plan to get my second booster next week, with 65% meaning whatever it means, because I have come to the possibly biased opinion that it's a good idea. I can't prove it's a good idea, I don't really understand the 65 %, but I will be getting my second booster. Most people make most decisions, including important ones, in the way that I am making mine. It seems right to us.

 

 

The 65% refers to "effectiveness against symptomatic disease."

This is quite different from disease that causes hospitalisation|ventilation|death.

A symptom is something that the patient reports. "I feel unwell" "I can't smell" are symptoms.

A sign is something that can be observed - "Ken has a fever/red eyes etc."

 

This means that a person who has extremely mild|vague symptoms and possibly no signs might feel "not quite right" and get a test that comes back positive for sars-cov-2.

 

The authors of the NEJM article conclude that (inter alia): "Our findings indicate that two doses of vaccination with BNT162b2 or ChAdOx1 nCoV-19 are insufficient to give adequate levels of protection against infection with the omicron variant and mild disease. Boosting with BNT162b2 or mRNA-1273 provided a substantial increase in protection against mild disease".

Suggesting that a booster is a good idea. Four doses is even better.

 

The "mild disease" rider is critical. The survey doesn't discuss the effectiveness vs serious disease (needing hospitalisation/ventilation etc).

We still don't know what "mild disease" means in terms of long-term sequelae. My recollection of the heart disease data after quitting smoking is that the average population risk of heart attack returns to baseline after 3-4 years (old data - not sure about recent work).

Similarly, there may be long-term sequelae of a mild case of COVID-19 - we don't know what this will be.

Long term consequences of viral infections are very common e.g von Economo's disease (as discussed by Oliver Sacks), depression following infectious mononucleosis and many others.

The relationship between initial disease severity and sequelae is also unclear.

 

In NSW we have had around 1M positive cases (popn ~8.5M) in the past 2 months alone. This is likely a gross underestimate since antigen tests give false negative results ~50% of the time. It may be that pretty much everyone has now been exposed if 1 in 8 people were positive in the last 6-8 weeks. This is probably the reason why many jurisdictions have 'given up' as evidenced by the almost total lack of mask wearing.

 

In the end, survey results only tell us about the whole population.

The survey results published in the NEJM seem to confirm the information that is out there: best protection comes from having 4 doses at appropriate intervals.

 

In Australia at the moment, despite having a superb double-dose vaccination rate (but not quadruple) there is a surge in positive cases and our deaths per million cases is the same as the USA.

 

Looking at the Johns Hopkins dashboard the figures today are:

 

Australia vs USA

Deaths last 28 days: ____867_______________11776

Population: _____ _____ 25687000_____ 329484000

Deaths/million: _____ ____34_____ __________36

[kenberg]

This will be a "yes but" reply.

 

The 65% refers to "effectiveness against symptomatic disease."<br style="color: rgb(28, 40, 55); font-size: 13px; background-color: rgb(248, 248, 248);">This is quite different from disease that causes hospitalisation|ventilation|death.<br style="color: rgb(28, 40, 55); font-size: 13px; background-color: rgb(248, 248, 248);">A symptom is something that the patient reports. "I feel unwell" "I can't smell" are symptoms.<br style="color: rgb(28, 40, 55); font-size: 13px; background-color: rgb(248, 248, 248);">A sign is something that can be observed - "Ken has a fever/red eyes etc."

It seems 65% effective cannot mean that 65% of the vaccinated never had and never will have symptoms, that would require predicting the future. Otoh, I doubt it means that a survey was done and, on the day of the survey, 65% of those interviewed were not having symptoms that day. That would mean 35% were having symptoms on that day and that seems extreme. I suppose it is somewhere in between.

It all gets tricky. Playing with the robots on Saturday I had a 75% game. Playing with the robots on Sunday I had a 48% game. And of course these percentages are averages of the percentages on the various hands.

But here is what I was really getting at. Maybe, if I looked at the studies hard enough, I could work out what that 65% really means. But I didn't. And few have. Still. I plan to get my second Pfizer booster this week. Someone, maybe many someones, said that every important decision is made on the basis of incomplete information. We decide, we act, we move on with our lives.

Somehow, with social media the way it is, people are not merely acting on incomplete information they are acting on totally false information. How do we address that? I know enough to know that I don't know much, but I feel I have a fairly decent ability to choose carefully. My first two shots were Pfizer. I could have gotten my first booster using Moderna but I waited a few days so I could get another Pfizer. Was this my best choice? Beats me, but it was my choice and I think it was reasonable enough. Becky had a similar choice to make when she got her first shot, she could get J&J right away or she could get Pfizer by waiting a week. She waited. Neither of us considered drinking bleach.

Going up a few posts, Y66 cites an interview with Jonathan Haidt where he says "there is a qualitative difference". And I responded "Yep, that's how I see it".

The trouble is that I don't really see how to deal with this qualitative difference. Making a reasonable but perhaps not optimal choice is one thing, making a totally nutty choice is very much another thing.

[thepossum]

I believe the distinction he was making is private vs. government, not privately-owned versus traded on a stock exchange.

 

They need to be more careful with their constant sloppy language and arrogance that goes with it

 

Organisations like Twitter have responsibilities as well as rights. All of them do

 

I could say that about the whole country and State in which many reside

 

They are huge international utility like corporations throwing their weight around a bit too much

 

Many of us have been on the receiving end of their arrogance and ignorance. I was even defamed by them many times etc

 

They basically make their own laws. Summary injustice etc

 

Many people including the likes of Iglesias and others keep defending what they do. It's frightening

 

You don't like it find another platform. It's cultural arrogance

 

What's even worse are the people who say if you don't like it go and set up your own

 

 

Nobody is undermining the achievement of certain creative types and the huge global corporations that grew from their idea and unique opportunity. Maybe not everyone can set up an

Intel, a Microsoft, an Amazon Web Service, a Twitter,a Facebook, a Hewlett Packard etc. I believe they would sue you for stealing their idea

 

Time to get out of the World Series mentality and think about the real world

[cherdano]

This will be a "yes but" reply.

 

[/size][/color]

It seems 65% effective cannot mean that 65% of the vaccinated never had and never will have symptoms, that would require predicting the future. Otoh, I doubt it means that a survey was done and, on the day of the survey, 65% of those interviewed were not having symptoms that day. That would mean 35% were having symptoms on that day and that seems extreme. I suppose it is somewhere in between.

65% vaccine effectiveness means that (over the course of the study period) a vaccinated individual has a 65% lower chance of catching covid than an unvaccinated individual.

 

Of course, as the virus keeps circulating, the unvaccinated get more immunity-from-infection, and the measured vaccine effectiveness will drop. (If I understand correctly, Hazard ratios are a measure that is immune (sorry!) to this effect, by continuously dropping the previously infected from your study population counts.)

[helene_t]

If I understand correctly, Hazard ratios are a measure that is immune (sorry!) to this effect, by continuously dropping the previously infected from your study population counts.

Not necessarily, you can also report hazard ratios from a model that allows reinfection.

 

Maybe people who are known to have been infected are dropped. People who may or may not have been infected (no antibody test available) should ideally be "partially" dropped but I don't think that epidemiologists do this so often.

 

On the other hand, previous infection (whether known or unknown) could also give some additional immunity to vaccinated people, so it may not skew the results so much.

 

Disclaimer: I have little experience with modelling of infections diseases, I work mostly with cancer and arthritis.

[cherdano]

What I meant is the following. If someone gets infected during the study period (vaccinated or not), we may not want to count them among the study population for the remainder of the study. But this means our denominators (for the size of the vaccinated/unvaccinated groups) is changing constantly. If I understood correctly, hazard ratios are the normal way of dealing with this.

 

Is that correct? I guess it's used more commonly for comparing death rates, as most people can die only once.

[kenberg]

I could have, and ok I should have, looked up just what the effective rate was. Even there it gets a bit confusing. Apparently there is both an effective rating and an efficacy rating, the difference being that efficacy apple when they have more control. If they just look at how the vaccinated fare against the unvaccinated, that might not all be due to the vaccine. I now have had four shots. But I also still wear a mask when I grocery shop and I avoid crowded areas. To really get a handle on effectiveness, it seems you want to compare a group of people who are generally cautions and get the vaccine against a group that re generally cautious and don't get the vaccine. The problem being that the second group might be the empty set.

 

I acknowledge my laziness in not looking into details, I still haven't. my general point still seems right to me. News reports cite statistics, probably they are real statistics if you go to a serious news source, but exactly what the numbers mean is apt to take some thought.

 

And, like just about everyone, I don't give it enough thought. It seemed clear Pfizer was good, it seemed clear i should wear a mask and avoid crowds, so that's the plan.

[helene_t]

What I meant is the following. If someone gets infected during the study period (vaccinated or not), we may not want to count them among the study population for the remainder of the study. But this means our denominators (for the size of the vaccinated/unvaccinated groups) is changing constantly. If I understood correctly, hazard ratios are the normal way of dealing with this.

 

Is that correct? I guess it's used more commonly for comparing death rates, as most people can die only once.

You are probably right, they might have called it incidence ratios rather than hazard ratios if they didn't remove infected patients. But would they know if someone got infected without showing symptoms?