What would be an interesting follow up study is: if a woman wears semen cologne from one man, are other men repelled by it? And vice versa, if a man has another woman’s scent, is a second woman repelled?
Likely more difficult to get participants for though 😂 But perhaps it could be designed another way.
That's interesting, and it made me wonder, are there pheromones that are repellant rather than attractive in the animal kingdom?
The answer is yes, e.g. pea aphids release an "alarm pheromone" that makes the other aphids fall off the plant when there is a predator or parasite around.
Can you make your data available? And give the likelihood function? What models are you using for your hypotheses?
I find it strange to give Bayes factors as if one was doing null hypothesis testing. After all, the null probably just wrong, even if the effect is irrelevantly small. The frequentist methods that most people use to get the p value here model the distributions as normal distributions with unknown means. Thus the Bayesian approach should also do this, and get the likelihood function L(d) giving the probability of the data assuming the metahypothesis that the two normal distributions have means whose difference is d ("metahypothesis" because this is really the integral over the hypotheses parameterized by mean1,mean2 restricting to mean2 - mean1 = d).
This way we can see the entire distribution of how we should update.
Thank you for your comment and I would be happy to learn from your analysis. It sounds like you are saying that a parameter estimation approach is more appropriate than a decision rule here, which I agree with.
Do you see significant autocorrelation in the time series of heart rates and distances, taken in the order in which the participants were subjected to the test? Were people waiting in line standing or sitting before walking into the room for hugging? I wonder whether the lumping of all controls at the beginning of the experiment may have an effect. If the first 10 people just walked in or maybe waited a little while standing while the other 10 realized that they were facing a longer wait and sat down then maybe the heart rate is explained by orthostatic response?
They were summoned to the testing room from a common area at a conference where they may have been talking, walking, sitting, working, etc. Note that the heart rate measurement is the difference in heart rate within a 10 second window, ie heart rate after 10s hug minus heart rate immediately before the hug.
"Seeds" of Science indeed :-)
Joking aside, I appreciate that someone is willing to test these things!
hehe ;)
What would be an interesting follow up study is: if a woman wears semen cologne from one man, are other men repelled by it? And vice versa, if a man has another woman’s scent, is a second woman repelled?
Likely more difficult to get participants for though 😂 But perhaps it could be designed another way.
That's interesting, and it made me wonder, are there pheromones that are repellant rather than attractive in the animal kingdom?
The answer is yes, e.g. pea aphids release an "alarm pheromone" that makes the other aphids fall off the plant when there is a predator or parasite around.
Can you make your data available? And give the likelihood function? What models are you using for your hypotheses?
I find it strange to give Bayes factors as if one was doing null hypothesis testing. After all, the null probably just wrong, even if the effect is irrelevantly small. The frequentist methods that most people use to get the p value here model the distributions as normal distributions with unknown means. Thus the Bayesian approach should also do this, and get the likelihood function L(d) giving the probability of the data assuming the metahypothesis that the two normal distributions have means whose difference is d ("metahypothesis" because this is really the integral over the hypotheses parameterized by mean1,mean2 restricting to mean2 - mean1 = d).
This way we can see the entire distribution of how we should update.
Thank you for your comment and I would be happy to learn from your analysis. It sounds like you are saying that a parameter estimation approach is more appropriate than a decision rule here, which I agree with.
You can find the anonymized vabbing data here: https://docs.google.com/spreadsheets/d/1vinOyL-9XDhhC7ybQAxnC4556WghehAnNGwagFwLopo/edit?usp=sharing
I used this function to calculate the Bayes Factor of the alternative hypothesis: https://pingouin-stats.org/build/html/generated/pingouin.ttest.html
The documentation suggests that they use Jeffreys–Zellner–Siow (JZS) priors.
Do you see significant autocorrelation in the time series of heart rates and distances, taken in the order in which the participants were subjected to the test? Were people waiting in line standing or sitting before walking into the room for hugging? I wonder whether the lumping of all controls at the beginning of the experiment may have an effect. If the first 10 people just walked in or maybe waited a little while standing while the other 10 realized that they were facing a longer wait and sat down then maybe the heart rate is explained by orthostatic response?
They were summoned to the testing room from a common area at a conference where they may have been talking, walking, sitting, working, etc. Note that the heart rate measurement is the difference in heart rate within a 10 second window, ie heart rate after 10s hug minus heart rate immediately before the hug.