Jump to content

Prometheus

Senior Members
  • Content Count

    1601
  • Joined

  • Last visited

  • Days Won

    9

Prometheus last won the day on January 22

Prometheus had the most liked content!

Community Reputation

524 Glorious Leader

2 Followers

About Prometheus

  • Rank
    Primate

Profile Information

  • Interests
    Building statistical models for Raman spectroscopy.

Recent Profile Visitors

19222 profile views
  1. Khan academy made a video giving a good explanation of how estimates are made:
  2. Well the papers above were all for users of phones themselves - i.e. you need the phone glued to your ear. There was nothing looking at the more general environment - but if the effect size was small for the users themselves then it's going to be even smaller for for someone at any distance.
  3. Better sell your car, put away any and all tools, and never light a fire or wear buttons (not being facetious but all these are measurable causes of death - the last by choking).
  4. Without digging into the details it looks reasonable quality for what it is. Case-control studies, while useful, are relatively low on the hierarchy of medical evidence. A more comprehensive meta-analysis found an association between long-term use and low-grade gliomas - but acknowledges that most of the evidence is low quality. As to whether you want to worry about it that's up to you - everyone has a different threshold for how much risk they are willing to accept. I'd say there are far more concerning things in the world if you want something to worry about, and that being able to speak to my parents on the phone is worth the possible tiny increase in my glioma risk.
  5. Ultimately why want to be able to say something about the efficacy of the intervention. I think a retrospective cohort design with statistical control of the confounding variables is your best choice. By the hierarchical structure i mean that the schools themselves might be considered confounding variables (some schools are better able to implement the programme than others for whatever reasons). Also, since you are looking at several outcomes you need to account for an inflated false discovery rate (the more tests you do, the more likely you are to erroneously find something 'statistically' significant) - something like a Bonferroni correction. The biggest problem i can see is that schools self selected for the study. Could it be that these schools just cared more about their students nutritional health? Or did they had more resources to start with? You've already explored this problem by considering the student's pre-intervention scores. You could take the change in scores of each pupil to be your outcome, instead of the post-intervention score.
  6. This sounds like the poisoned drinks problem. Unfortunately it won't work here. In the idealised case where we there is say, exactly 1 in every 100 people infected then it could apply. But for every 100 hundred people we take there is no guarantee of the number of infected people. Sometimes there are none. Most times there will be one. Occasionally there will be 10. It is a random variable itself. Also, the tests themselves have a number of false positives and false negatives which will likely be significant.
  7. Sounds like one of those studies that could end up with a complex design. Here are some questions: Was this an intervention set-up specifically to assess whether it makes a difference (if so that would have been the time to consult a statistician), or was it some kind of programme and retrospectively you have decided to try to measure it's effects? Were the schools selected for intervention done so by truly random means? (and not by district, or the first letter of the schools name etc.). Was it exactly the same intervention (at least on paper) for the same length of time on the same age groups? Is the base unit you are interested in the school or the children? If the latter you'll need to take into account the hierarchical structure in the data from the schools.
  8. Thought this was an excellent video introducing the maths of exponential growth and logistic growth and applying to the virus. https://www.youtube.com/watch?v=Kas0tIxDvrg
  9. High blood pressure is a risk factor for several bad things. It can be caused by many things (disease, lifestyle, genetics). It is aggressively treated because it is one of the most easily modifiable risk factors for a range of bad outcomes. If you are ever getting aggressive chest pain please call an ambulance.
  10. Not an odd question at all, but maybe a bit niche. I was recently reading something about our tendency to distort probabilities in games - people interpret a 99% chance as being certain and go ape when it doesn't happen. Also read something about rock, paper, scissors and how even in such a simple game psychology dictates optimal plays. I think there's a great body of literature out there, get onto google and have a look. I found this which seems a reasonable start.
  11. We only know certain biomarkers correspond to certain mental states because people were asked how they feeling at the time they were taken: you still needed to trust someone was truthfully reporting their mental state at some point. We still don't directly experience someone else's mentation. Anyway, all this is just dancing around the question of why the substrate matters. Humans have a mental state because of biology. Why does that preclude AI having mental states?
  12. Are you stating that emotions can only be felt as the result of neurotransmitters, hormones and other chemicals in the brain? It's a bit of a black swan situation isn't it. All we've seen are white swans so far. The problem is, when it comes to the mental states of other beings, we are colour blind.
  13. Every year there's are more stories about an extinction level meteorite hitting the Earth. One day one of those stories will be 'right'. But kudos to Sylvia Browne for at least making quite an, oddly, specific prediction.
  14. I guess a question is when, if ever, it is useful to regard the planet a living organism. Could it help motivate environmental efforts for instance, or is the risk of new age beliefs compromising scientific understanding among lay people too great?
  15. Humans aren't born able to recognise faces, but learn. We are born with the instinct to track faces though, greatly helping the learning process. It's not too hard to imagine AI able to track faces - i imagine your smartphone can already do it - and thereafter learn to distinguish individual faces. I'm sure dogs have some understanding of humans and that it's quite unlike our own understanding (rooted in smell for instance). I can well imagine AGI not understanding humans the way humans do, but having some understanding, unless the emulation pathway is successful.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.