Dancing David,
I have read your criticisms of the ganzfled protocol on the other thread and I believe you to be in error. With regard to your claim that a 25% hit rate is not expected by chance because of biased matchings of receivers words to targets I shall say this:
The exact same scenario occurs during games of roulette. People have their favourite numbers - 0, 13 (unlucky for some), birthdays etc. This creates exactly the same bias as you are talking about. In other words, the "target" numbers will not be chosen by the gamblers at a consistent frequency. However we all know that the gambler cannot beat the odds. Just because there exists a bias in the selection of the winning number by the gambler, the chances that each number will come up remains consistent.
Then you did not understand the critique at all :
1. Matching words that match more than one picture, in the case where pictures are not tested against random reciever strings it is very possible that there could be more than one picture in a set that matches a given word spoken by the reciever. In this case the odds of a word spoken by the reciever matching one of the pictures is higher than 25%.
2. If you have many pictures that are unrated for random reciever word string match then you can have an effect where say you have four pictures that have a 50% chance of matching a 'random reciever string', that means that there is then a fifty percent chance of a hot being scored for any random string uttered by the reciever. This is a serious methodoligical falw that could be controlled for and in fact can be used to the benefit of psi research. So again this destroys the false assumption of a twenty five percent rate and makes it much higher.
The point is that before you assume that the pictures will match a random reciever word string at a twenty five per cent rate you need to determine the actual match rate.
That is why it is very important that the 'match criteria for a hit' be determined in a fashion described as 'objective', in that there should be a list of words for each picture and only those word will generate a hit, then sets should be devised such that each picture has unique 'match criteria for a hit'. This would eleiminate most of the flaw that i see
But you can't just assert that there is a twenty five percent match rate with out these criteria.
Yes, this is known factors. I'd like to see you try to control for an effect that is not known!
The factors that I discuss are known factors, this happens in research all the time, there are times where even the way that the experimenter presents the routine can effect the outcome, which is why you have video tape , trainings and scoring.
Assumiong that the match rate will be twenty five percent for a trial is a mistake and it can be fixed very easily.
That's funny, because this is the tired old criticism of parapsychology!
In science it is imporatnt to think about confounding factors before you design your experiment, and once they are suggested then that is just "Oh too bad." This is true on about eighty percent of psychology research, experimenter does a study, people point out potentail flaws, study is redone or abandoned, that is science.
How do you control for something when you have no idea what it is you are supposed to be controlling for! This is my objection to this criticism of psi research. The criticism asserts that psi experiments could be down to an unknown factor that you haven't accounted for.
No I don't say that you have to control for tidal forces, but when you design your research you have to be open to thinking about the flaws in your methodology. I am not saying that psi research is flawed I am saying that the methods of the studies I have seen are flawed.
Believe me, this is true of much research in the social science.
The flaws you point out do not exist. See above.
The flaw is in the bold assumption that there will be a twenty five percent match rate in a given set. there are reasons that this might not be the case. The methodology could be very easy to implement and there are a number of ways to do it.
1. rate the random match for each picture.
2. set up unique criteria for each picture in a set.
Either one will do, but you can't assume a twenty five percent match.
Which is why they have used a distribution free method of analysis. If they used a parametric t-test then you would have a point.
Doesn't matter if the standard deviation is higher than the effect than there is no effect.
I have no idea what you are saying here
Then you didn't read your own study or you didn't ubderstand it , part of the way that they generate the effect is by looking for responses prior to the stimulation.
At this time there is no mechanism to account for it. I find it interesting but distatseful as it strechs the times frame to make the effect appear.
Yes I can sat that because the elimination of the data was performed in a random way since all control epochs were generated randomly. The elimination was done before any analysis is made.
You misunderstand the nature of good sampling technique this is called an artifact of sampling, it can not be assumed to be random and it could have easily been eliminated. It is not a deadly flaw but it would not be used in any other field than psi research.
Since it's not a flaw this would have been a waste of time.
Your above comment shows that you have little understanding of sampling technique. It would not have wasted anytime at all, it would have allowed for clearer data sets and would not have had the artifact in the elimination of data that you are just asserting is randomly eliminated.
This is exactly the same as saying that if they conducted an extra 50 light flashes and added this data to what they had it may have eliminated the effect. I think it is you who is guilty of wishful thinking. Remember that the elimination of the data was random (since all control epochs were randomly generated) and occured before any analysis was made.
You so silly. You obviously don't know about the perils of sampling error, many a fine theory has gone down the tubes because of sampling error, it is a real possibility. Sorry, that is the way it is. I hold parapsychology to the same standard I hold any social science, you wouldn't understand that would you?
Sampling error is real thing that produces many a strange artifact of science. Oh Too Bad!
You also don't understand that the sample they produced is not a random sample any more.
Wow, the first thing we agree on! Perhaps they did keep records of this.
In other words the attitude of the researchers/experimenters must be known before and during trial…. That should be the next step in PSI research, an agreement of publishing accompanying attitude (yes, no skeptic is going to sign ‘this is absolute crap!’ 