Help with statistics on successful experiment

RV Protocol

GO VIEW! @TKR's dojo

Project Home Page

RV webForum (here!)

Email: @Yahoo|@Google

Help make it possible for everyone!

TKR is Funded by:
PJ of Dojo Psi

Simple Session

CS Labs
Ingo Swann
Joe McMoneagle

Baylands Nature Preserve
Pete's Harbor Restaurant
Lucid Dream Example
Radio Listening Post
Test-Bed Trials
Video ABC Put to the Test

P.I.A.

Greg K's ARV Trials

RV Roast (alt)

RV Library

[Benton]

Hi folks.  Its been years since my last statistics class...many years.... and I need help. 

At the Rhine Center we have a group of remote viewing enthusiasts that meets monthly to discuss and practice RV.  Lately we have started doing ranking in our analysis of session data.  We have a small group of viewers view, then we give the session data to a small team of analysts who rank 4 photos from a target pool based on the session data.  One of those photos is the actual task.  Everyone is blind to the target, because the actual target has not been chosen yet.  The viewers themselves never see the 3 photos that are not the target.

So the analysis group looks at the target pool, and looks at the session data, and ranks the photos, 1 for most like the session data, 2 for next most like it, etc.  Then we put the photos back in their envelopes, shuffle them, lay them out on the table and toss a coin twice.  There are 4 possible combinations of coin toss, 1 for HH, 2 for HT, 3 for TH and 4 for TT.  That number determines which envelope is the task envelope.  We get the entire team of viewers and analysts in the room and we open the one envelope that was chosen by the coin tosses.  Then we reveal the analysts rank of that photo.

Here's the deal though.  We have done this type of judging 5 times and only 5 times, and we have 4 #1s and 1 #2.  I know its a small sample size but I see a pattern and I want to be able to discuss the significance competently.

Can someone provide me websites or journal articles on statistical analysis using rank order judging? 

Any help is appreciated.

[sonny5085]

Sounds very successful !

Just curious about the 4 photo targets out of the target pool, are they selected out of the target pool before session data presented to the judges.
It sounds like your saying that the session data from the viewers, than judges determines the 4 photos out of the pool (what size is that pool) then put into (4 envelops) then the coin toss determines the actual target out of the 4 ? with the 2 coin toss heads, heads, tails, tails, ect.
envelopes.

Im not sure how that odds are figured, at any rate, sounds like your really doing well !

[Benton]

thanks for the positive feedback.

I walk into the meeting with 4 photos I have chosen, to make sure they are relatively dis-similar.  A random code number is chosen, like BR549, and the viewers are asked to describe the photograph in the envelope marked BR549.  But the actual photo of the 4 has NOT been chosen yet. 

So they do their sessions, and their session data is given to the analysis team who also sees the 4 photos.  The viewers of course did not see the photos and only got the BR549.  The analysts rank the 4 photos by the session data from most like the data to least like the data. 

Then they put the pictures back in individual envelopes and I lay them on a table and toss a coin to choose the envelope.  Again, I am not looking at any photos.  I take the label BR549 and stick it on that envelope determined by the coin tosses. 

I hold up that one envelope with the label BR549 on it, and tell the group "This is the target BR549" and I open it, and there is the photo ranked most like the session data.

Its more complicated than that, but that is how the session goes and targets are chosen.  I hope that clears up earlier vagueness in my description.

[sonny5085]

Ok ! got it !

4 photo s possible (not sure since this is totally blind and has a possiblity to be any 4 pictures in the universe to choose from)
1 out of 4  that you choose to be dis similar from each other but still just 4 pictures how does one determine odds for that?

Then the sessions match up to the best match out of the 4 targets judges ranking with 1 being the best outcome,for each session that is judged to match photo.

Then the coin toss to determine the envelope that gets chosen as the target ! (dont knw what the odds are then?)
if the target envelope chosen has the judges analyzed that particular photo with viewers sessions graded as 1,1,1,1,1 then all your viewers nailed the experiment !

Is that right ?

[Gene]

Wow Benton,

While I have no ability to help you with the needed math/ statistical skills, I really want to commend you on the airtight protocol of all this. I wonder if Daz might not want to offer a little space in the next issue of Eight Martinis as it really shows it doesn't take a team of scientists working in a lab to produce things in protocol.

Again I thought this all quite stunning.
Gene

[Benton]

... your viewers nailed the experiment !

Is that right ?

The viewers sessions were judged as a group, so all the paperwork of all the viewers on the team was spread out on the table.  Then the 4 potential targets photos were spread out too, and the analyst team reviewed the session data, as a whole, and looked for patterns and motifs and similar pieces and parts, and then compared that info with the pictures and rank ordered the photos according to the whole pile of session data. 

This is a type of analysis used by early RV researchers.  The ranking process gives you easier statistical analysis (which I need more info on), as opposed to more subjective assessments.  You know how it goes, you do a session, you get feedback, and you say "oh, yeah, I saw that, and that, and I forgot to write down that..... " but with this method, you have to PICK a picture, much of the subjectivity is done away with.  If there was NO PSI involved, you'd expect to see an even distribution of 1, 2, 3, and 4s.  We got 1, 1, 1, 2, 1. 

There was an old BBC documentary with Keith Harary where they used this method, find it on YouTube "In Search of the Dead" which is a great title for something about ESP, ironic anyway.

I am not sure WHY our little method is working ( that's not true of course, I am convinced I know why...LOL) , but it sure amazing to see it happen.  Now having said all this, next trial will fall flat.  But hey, that is science.  And if it DOES fail, that tells us something too.  Anyway, LOTS of trials are needed and hopefully we can do that.

[PJ]

Awesome!  RV is so much fun with others! Benton I just realized, we have a "Meet-Up Groups!" link set on the left of this screen now, we should add yours if you have some kind of link for it!

Is there any way you can arrange the tasking so that you are not informed of the potentials? I realize that there are four options and nobody knows which option will be chosen until afterward. That is actually the best-practice protocol in the lab right now for target selection (after viewing, after judging), so congratulations for choosing that one, nobody can argue that for sure.

They could argue that the tasker--even with 4 options--shouldn't be present with the viewer(s) or analyst(s) though, that's more like a methods training protocol than an actual remote viewing protocol. If you couldn't change this, you could at least further-randomize it and then fully blind it to yourself, that would cover it.

Such as choosing 4 options, put each in an envelope, put those in a bigger envelope, and then do that a whole bunch of times, and then use a die roll to choose the starting set (ideally, the viewers or analysts doing it without your presence, and then putting both the selected ones and the non-selected ones somewhere not visible to you until the experiment is over (until the target is being chosen), so even subconsciously by envelope visual, you could not be aware of which set it might be).  

I realize this is a bit fierce, but RV by its nature is pretty much obsessed with preventing any, even the slightest, non-psi based target info, so this would just help it to be much better for that element  -- if you have zero way of knowing any of the options currently in focus, then your presence during viewing or analyzing is not a factor. (Let's assume it is truly not a factor, in which case, this extra effort should make no difference to the viewing results at all, but will fit much better with RV's protocol.)  

Meanwhile, did you figure out the stats approach for the ranking-analysis?

By the way. I would expect any group of people with sincere intentions to demonstrate psi. I would expect a social group having fun with it to probably do better. And of course, having the benefit of Benton in my opinion would make it all even more ideal.  

PJ

[PatMcDonald]

The short answer is "no" Benton I do not have a link to a home course on statistics BUT...

... it does strike me that, only 5 is not any sort of data sample type. So, I suggest you;-

1) Perform at least 100 trials to give some sort of percentile data set.

2) While doing that, take a crash course in statistics.

Fact - if you do 100 basic coin tosses, probability is that you will get at least seven straight heads or seven straight tails.

Your results, while encouraging, will not be taken seriously with such a small data set for that reason.

Just trying to be realistic here.

[Benton]

I got in touch with a lot of folks including Ed May who sent me some info, referred me to some articles. 

Yes, sample size is too small.  I am experimenting on my own.  In the group, we are trying new stuff but I am always working to keep the parts that seem most important to the process as part of the mix. 

I must say having a group to meet with and try stuff out with is very helpful.  Joe McMoneagle's recent article in 8 Martinis is correct in that regard.

Thanks for the feedback here, it is very helpful.

Username:
Password: