Forums


Go Back   The TrekEarth Forums > TrekEarth Forum Home > General

Notices

Reply
 
Thread Tools Display Modes
  #21  
Old 01-18-2008, 08:24 AM
Loic_bzh Loic_bzh is offline
TE Expert
 
Join Date: Feb 2006
Location: Paris, France
Posts: 1,571
Default Re: Some TE statistics...

Just-A-Guy wrote:
__________________________________________________ _____________
"Of course the average is near 1"

Why is it natural for this value to be around 1 ? Without reciprocation it could be any value. he fact that it is centered around 1 suggests a stronf influence of reciprocal point swapping.

__________________________________________________ _____________


Err... ok, I'm not a sats pro but since for every critique there a giver and a receiver, the global average is not around 1 but IS 1. Am I wrong?
Reply With Quote
  #22  
Old 01-18-2008, 08:54 AM
ktanska ktanska is offline
TE Expert
 
Join Date: Aug 2005
Posts: 1,055
Default Re: Some TE statistics...

> No, it really means that for 95% of TE members the real value lies
> within this range. Only for 5% of the members (like me) this value is
> larger than 1.08 or smaller than 0.96

No it does not. Alexander's second analysis gives a closer idea on how big is the real correlation between points and critiques given:

> The other approach is to plot the received points against the given
> critiques. When I did that with my selection, I found that the linear
> regression model can be applied to the plot with the R-squared value of
> 0.88, which is again extremely high for a random selection and signifies
> a strong linear relationship between received points and given critiques.

But also this neglects the fact that both of these factors have also strong correlation with the number of uploaded photos and time since joining TE. I agree that there's strong correlation but the strength is not that easy to calculate.
Reply With Quote
  #23  
Old 01-18-2008, 08:58 AM
ktanska ktanska is offline
TE Expert
 
Join Date: Aug 2005
Posts: 1,055
Default Re: Some TE statistics...

> Err... ok, I'm not a sats pro but since for every critique there a
> giver and a receiver, the global average is not around 1 but IS 1. Am I wrong?

You are very close to right. Except that Alexander's analysis excluded members with points less than 629 points. And there are some qritiques where 0 or 1 point is given instead of 2. And some points disappear when photos are deleted.

Kari
Reply With Quote
  #24  
Old 01-18-2008, 08:58 AM
Curioso Curioso is offline
TE Junior Member
 
Join Date: Jun 2003
Posts: 497
Default Re: Some TE statistics...

I found 2.75 for me...

I like to the idea of be part of the 5% remaining... Funny game !
Reply With Quote
  #25  
Old 01-18-2008, 09:14 AM
Loic_bzh Loic_bzh is offline
TE Expert
 
Join Date: Feb 2006
Location: Paris, France
Posts: 1,571
Default Re: Some TE statistics...

Curiosowrote:
____________________________
I found 2.75 for me...

I like to the idea of be part of the 5% remaining... Funny game !

____________________________

That's just because you're posting -good- photos without writing critiques anymore (and since you're a long time member, you're known by some active members).

Definitly not my vision of TE but everybody is free to use TE his own way...
Reply With Quote
  #26  
Old 01-18-2008, 09:26 AM
Just-A-Guy Just-A-Guy is offline
TE Newbie
 
Join Date: Jan 2007
Posts: 75
Default Re: Some TE statistics...

I've done this analysis in march 2007 with both variables divided by the number of posted photo, and still found a good correlation with a coefficient of 0.81 (this analysos was for members with 10000+ points at the time)
Reply With Quote
  #27  
Old 01-18-2008, 09:29 AM
pasternak pasternak is offline
TE Junior Member
 
Join Date: Apr 2006
Posts: 435
Default Re: Some TE statistics...

Yes, theoretically the global average IS 1 (if we exclude the critiques rated 0 or 1), the fact that in my analysis the average is 1.02 and not exactly 1 means that the selection of 116 members was not, strictly speaking, representative enough. The more members you include in your analysis, the closer the average will be to 1, and I didn't doubt that from the beginning; what I wanted to show is that the confidence interval (or standard deviation, if one prefers) is small. This shows that the TE model is to reciprocate critiques, as opposed to the model with "givers" who only write critiques and "receivers" who only receive them; in the latter case, the confidence interval would be much wider.
Reply With Quote
  #28  
Old 01-18-2008, 09:41 AM
pasternak pasternak is offline
TE Junior Member
 
Join Date: Apr 2006
Posts: 435
Default Re: Some TE statistics...

Yes, you are right, strictly speaking, my results are true only for the members with 629 points or higher, and even then, if I'd include more members in the analysis, the results would be slightly different. What I wanted to show is the common trend though, not exact numbers. Also, 95% confidence interval doesn't imply that 95% of members in my selection had CR values in 0.96 - 1.08 range, it is more complex. Of course, there are members having CR higher and lower, and even in my selection, there is one member with an incredible CR of 2.99, and also two members with CR less than 0.5; but statistically speaking, the situation is as I described (at least if you believe Excel 2003... ;)))
Reply With Quote
  #29  
Old 01-18-2008, 09:55 AM
ktanska ktanska is offline
TE Expert
 
Join Date: Aug 2005
Posts: 1,055
Default Re: Some TE statistics...

> I've done this analysis in march 2007 with both variables divided by the
> number of posted photo.

Ok then, that's good enough for me.
Kari
Reply With Quote
  #30  
Old 01-18-2008, 10:50 AM
Davids Davids is offline
TE Senior Member
 
Join Date: Oct 2004
Posts: 583
Default Re: Some TE statistics...

There are some members who are only "givers" . . . they only write critiques, but post no photo's. Still, for every giver is one receiver . . . but if you'd calculate the CR of a pure giver, it will always be 0, no matter how much critiques the member has written . . . .
A pure receiver will also have 0.
As soon as you include one of these pure givers/recievers in your calculation, the outcome would not be 1 . . . if the number of members included in the calculation is lower than the total amount of members. (but most likely to be within you standard deviation)

even you include all members the outcome would not be 1, because then the outcome would be polluted by members who use the full range of smiles (like me).
In my case the 2 would have to be replaced by something like 1,3.
Once you change the 2 in Pg (points given), the outcome of all members should be 1

A note on the side;
Smiles and points are not the same BTW, if you write a critique with a smile, the receiver could get points -depending on the color- but if you write a comment with a smile, the receiver will not get the points.

Now I need to go out and buy a dog and sausages, because I need to check something. So far I have noticed that the chance of getting points in return if you give a smile with one point are very small.
If the dog does NOT react when I give him ONE sausage, I will understand. . . .
I think I'll by a golden receiver.

Have a nice weekend to all of you :)
Reply With Quote
 
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT. The time now is 10:50 AM.

Powered by vBulletin® Version 3.8.4
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.


explore TREKEARTH