if you are arguing that all the points he was successful on can be determined by age, I think you are not seriously investigating the nature of all his claims (eg - social history, character/disposition, etc)
No, I'm not saying all his hits were determined by age, I've covered a number of factors. But there is an average of 15 predictions per person, so even 3 correct age-based predictions per person would raise the probability by 20%.
you seem to forget that when they switched readings (unbeknown to either the astrologer or the person on whom the reading was made) the accuracy slumped to something like 30%
Not at all. Certainly he seemed to hit on some fairly strong points. But again, there's not enough information in the video to know exactly what the hits and misses were. If the predictions were largely based upon the given information this is not at all surprising. Using the 20% figure from above, given an age difference and 3 such guesses we can expect the switch to change the accuracy by 40%.
so you are arguing that about 70% of people identify themselves as a martial artist/dancer?
Where do you get that idea from? My point is that the generalization and the approach used raises the probability of a hit as people tend to personalize the information given. His specific prediction was that the person would be either a martial artist or a dancer. This phrasing suggests identity, such that if you asked the person they would respond "I was a martial artist" or "I was a dancer". This doesn't seem to be the case. The woman responds that she "did study taekwondo". So whether or not this should be a hit is debatable. How accurate does the prediction have to be before a broadening interpretation significantly affects the probability? Is it the same hit if the woman studied taekwondo for 6 months or 6 years? Would it be a hit if the woman had a daughter in ballet? Or what if she just really liked kung-fu movies? This type of generalization is something that "psychics" use all the time, any vague agreement is taken as a hit.
But there's another point here as well. The astrologer combines two predictions into one, "there is a likelihood that this person was a dancer
or a martial artist". Here you see that agreement with either prediction was counted as a single hit. Where, in actuality, this is two guesses. Even if we take martial arts as a hit, he misses on the prediction that she was a dancer. Two predictions bundled together like that unfairly increase his odds of a hit.
I certanly don't see that idea present in any broad scale advertising campaigns ....
You must be joking. It's ubiquitous. Almost all advertising attempts to get you to make a personal identification with the product.
there are so many factors ... much like there are so many factors that can determine why a child doesn't develop cancer when (given their environmental/genetic background) there is a 70% chance they will
No one can make such a prediction for an individual. Those statistics are only relevant for populations. It is meaningless to translate them to an individual and say, "you have a 70% chance of getting cancer". All you can say is that an individual's profile fits in a population with a 70% cancer rate and therefore it would be a good idea to get screened more often.
Mere percentages of hits vs misses don't really provide a very good analysis in this scenario in any case because it really depends on what is being predicted. For example (grabbing a quick statistic), in America 80% of women from the ages of 40 to 44 have children. Therefore the baseline for this guess being a hit for a woman in this age bracket is 80%, making that which seems like a very accurate result due purely to chance.
Since you seem to enjoy referring to demographics and marketing we can find some quick baselines from that industry. "When researchers asked students to rate their ability to get along with others, 60 percent rated themselves in the top 10 percent. Ninety-four percent of university professors say they are doing a better job than their average colleague. Most men think they are good-looking."
Looking to the weight prediction, according to the CDC 66% of adult Americans are overweight and the percentages rise to around 70% with age. Add to that, for instance, that almost 50% of women who are not overweight think they are and you get a baseline for a positive response regarding a prediction that a woman is overweight of 83%.
You can begin to see then how such "predictions" have a chance base line far above 50%, which makes his 77% accuracy, even taken at face value, far less significant. IMO, this is merely a demonstration of cold-reading. His accuracy is entirely explainable within the scope of the data available to him and not the least bit surprising.
~Raithere
