SoylentNews is people
SoylentNews
If you like your coffee black, you may be someone who prefers strong flavours, takes good care of their health, or just wants to drink their coffee the way it’s supposed to be drunk.
At least, that’s according to a new study published in the journal Appetite, which found a correlation between a love of black coffee and sadist or psychopathic tendencies.
The research surveyed more than 1,000 adults, asking them to give their food and flavour preferences. The participants then took a series of personality tests assessing antisocial personality traits, such as sadism, narcissism and psychopathy.
The study, carried out by researchers at the University of Innsbruck, found that a preference for bitter flavours was linked to psychopathic behaviour.
The study missed a key, deciding factor: the coffee that psychopaths drink black is instant.
(Score: 2) by moondrake on Friday June 23 2017, @02:30PM (6 children)
So after you wrote this, I read it.
Psychology is not my field, and I am somewhat prejudiced in that I generally do not like the methodology in psychology (but some of that is probably inevitable when working with human behavior).
While I see some short-comings in their approach, they acknowledge that for the most part. In the end, we are just left with a weak correlation between preference of what people perceive as bitter, and questions that attempt to judge whether such people are more likely to have traits that are slightly socially frowned upon (its a very long way to true psychopaths I think). The statistics seem reasonable (though I disagree with their hypothesis formulation). What p-value hacking are you talking about? If this where my field, and if I have not missed something, I say no reason why such a study should not be published (whether I would fund it is another matter, but they seem to have not gotten specific funding for this, but I agree that they are probably employed on tax payer money).
So enlighten us, what is so bad about this study?
(Score: 1) by khallow on Friday June 23 2017, @04:20PM (4 children)
The problem is that if you look at enough combinations, you will find such weak correlations by random chance. If you look at one potential combination and find a correlation that has a 1 in 20 chance of appearing randomly, then you might have a significant correlation. If you look at 1000 combinations, you'll expect find 50 such 1 in 20 spurious correlations merely by random chance. At that point, merely finding correlations is no longer good enough.
(Score: 0) by Anonymous Coward on Saturday June 24 2017, @12:13AM (3 children)
It was never good enough. Imagine if Kepler stopped at "orbital velocity of Mars is correlated with distance from sun, and also the color of the leaves". Or Newton stopped at "acceleration due to gravity is correlated with the distance between the objects, and also their albedo". I really do not care if you found a correlation, there are unlimited number of real, actual correlations that are of no use to anyone. Then on top of this they are too cheap to make sure they find those, but p-hack into "fake"/transient ones. If you find a correlation and think it is interesting, do the next step and figure out what process could explain the actual relationship. With research like this they just stop at the correlations and get stuck there for decades. It is just collecting correlations.
(Score: 0) by Anonymous Coward on Saturday June 24 2017, @12:28AM
I mean appreciate these papers for what they are: homework assignments you need to do to graduate.
(Score: 2) by maxwell demon on Saturday June 24 2017, @07:20AM (1 child)
Planet speed is indeed strongly correlated with brightness: The closer the planet is to the sun, the more light it receives from the sun.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 0) by Anonymous Coward on Saturday June 24 2017, @03:47PM
Yes I was trying to use non-negligible correlations as an example.
(Score: 0) by Anonymous Coward on Saturday June 24 2017, @12:19AM
I really don't remember from earlier since it was so standard. I will guess the usual though:
1) not blinded
2) claiming x is related to y and not showing a scatterplot of x vs y
2) using dynamite charts or the equivalent tables ( I do remember they didn't even bother to make plots)
3) misinterpreting statistical significance in at least one of hundreds of ways
4) not coming up with any real model for what is going on, just rejecting the hypothesis of "no correlation" for that specific group of people since there was no well defined population (I think I remember it was mechanical turk samples...)
This is just run of the mill blab.