Perhaps you didn’t notice, but the paper is gated. It’s not possible for me or most people to check the paper. The description doesn’t mention the other two studies. The study described doesn’t sound like a strong result. I never suggested it wasn’t statistically significant. If it wasn’t, it shouldn’t be used to adjust one’s views at all. I assumed it had achieved significance.
It’s also odd for you to criticize me and then ultimately come to a conclusion that could be interpreted as identical to my own or close to it. What do you mean by “too soon to be confident of what the results mean?”. That could be interpreted as adjust your prior by 3% which was my interpretation. If you think a number higher than 15% is warranted then that’s an odd phrasing to choose which makes it sound like we’re not that far apart. Given that I was going by one study and you have three to look at it, it shouldn’t be surprising that you would recommend a greater adjustment of ones prior. Going by just the facial expression study, what adjustment would you recommend? Do you think this adjustment is large enough for most people to know what to do with it? What adjustment to ones prior do you recommend after reviewing all three?
While the scientific publication paywall is a pain (and inappropriate especially for publically funded research) it is not an impossibility to get the article—and as pianoforte611 already mentioned, secondary citations or descriptions to primary sources may not provide enough information to evaluate the source.
How to get articles: I’ve seen numerous cases here at LW where a request for a copy of a paywalled publication is quickly met with a link or an email from someone who has access.
The twitter hashtag #icanhazpdf also serves this purpose: tweet with the hashtag including a link or DOI to the article you are requesting, include your email address in the tweet, and delete your request after you get the pdf. You can use a temporary read-only email address (e.g. slippery.email) if you are concerned about anonymity/privacy.
On this instance feel free to send me a private message with your contact details and I will send you a pdf—I already downloaded a copy.
Edited to add: it’s also entirely legitimate to email the author of a published article and request an electronic copy of the article. There’s no need to explain why you want it and you need not be an academic “insider”, just be clear which article you are requesting. This is an example I received yesterday: “Dear {author}, I am interested in your recent article {full citation} but do not have subscription access. Would you be able to send me an electronic copy? Many thanks”
Sorry for assuming you had easy access to the paper. Given that you don’t, you are of course free to decide whether the pop science report warrants further investigation. However to authoritatively criticize and speculate on the details of a paper you haven’t read, I think lowers the quality of discussion here.
I’m not a Bayesian but nevertheless, I don’t agree that my conclusion is similar to yours. Prima facie, the effect itself seems fairly robust across the five experiments, but their theory as to why (which they did go reasonably far to test), still needs more experiments to be established. This is not a bug, and that does not make it a low quality paper. This is how science works. There may be more subtle problems that I (not being a statistician, or a psychologist) may have missed, but those can’t be known without delving into the details.
Perhaps you didn’t notice, but the paper is gated. It’s not possible for me or most people to check the paper. The description doesn’t mention the other two studies. The study described doesn’t sound like a strong result. I never suggested it wasn’t statistically significant. If it wasn’t, it shouldn’t be used to adjust one’s views at all. I assumed it had achieved significance.
It’s also odd for you to criticize me and then ultimately come to a conclusion that could be interpreted as identical to my own or close to it. What do you mean by “too soon to be confident of what the results mean?”. That could be interpreted as adjust your prior by 3% which was my interpretation. If you think a number higher than 15% is warranted then that’s an odd phrasing to choose which makes it sound like we’re not that far apart. Given that I was going by one study and you have three to look at it, it shouldn’t be surprising that you would recommend a greater adjustment of ones prior. Going by just the facial expression study, what adjustment would you recommend? Do you think this adjustment is large enough for most people to know what to do with it? What adjustment to ones prior do you recommend after reviewing all three?
While the scientific publication paywall is a pain (and inappropriate especially for publically funded research) it is not an impossibility to get the article—and as pianoforte611 already mentioned, secondary citations or descriptions to primary sources may not provide enough information to evaluate the source.
How to get articles: I’ve seen numerous cases here at LW where a request for a copy of a paywalled publication is quickly met with a link or an email from someone who has access.
The twitter hashtag #icanhazpdf also serves this purpose: tweet with the hashtag including a link or DOI to the article you are requesting, include your email address in the tweet, and delete your request after you get the pdf. You can use a temporary read-only email address (e.g. slippery.email) if you are concerned about anonymity/privacy.
On this instance feel free to send me a private message with your contact details and I will send you a pdf—I already downloaded a copy.
Edited to add: it’s also entirely legitimate to email the author of a published article and request an electronic copy of the article. There’s no need to explain why you want it and you need not be an academic “insider”, just be clear which article you are requesting. This is an example I received yesterday: “Dear {author}, I am interested in your recent article {full citation} but do not have subscription access. Would you be able to send me an electronic copy? Many thanks”
Choking Under Social Pressure: Social Monitoring Among the Lonely, Megan L. Knowles, Gale M. Lucas, Roy F. Baumeister, and Wendi L. Gardner
Most people in the general population can’t check the paper but on LW, I don’t think that’s the case. If you don’t have access to a university network http://lesswrong.com/lw/ji3/lesswrong_help_desk_free_paper_downloads_and_more/ explores a variety of ways to access papers.
This link is often useful for obtaining paywalled papers.
Sorry for assuming you had easy access to the paper. Given that you don’t, you are of course free to decide whether the pop science report warrants further investigation. However to authoritatively criticize and speculate on the details of a paper you haven’t read, I think lowers the quality of discussion here.
I’m not a Bayesian but nevertheless, I don’t agree that my conclusion is similar to yours. Prima facie, the effect itself seems fairly robust across the five experiments, but their theory as to why (which they did go reasonably far to test), still needs more experiments to be established. This is not a bug, and that does not make it a low quality paper. This is how science works. There may be more subtle problems that I (not being a statistician, or a psychologist) may have missed, but those can’t be known without delving into the details.