DPChallenge: A Digital Photography Contest You are not logged in. (log in or register) 
 

DPChallenge Forums >> Challenge Results >> Voting Results
Pages:  
Showing posts 1 - 17 of 17, (reverse)
AuthorThread
 12/13/2004 01:14:27 PM · #1
A question to people more familiar with statistical analysis. What does an irregular bell curve in the voting results for a picture imply? Does it imply people aren't voting correctly? Should a fairly normal looking bell shape be expected? I'm just curious because sometimes when perusing the results the curves are irregular so I was wondering what that implied.
 12/13/2004 01:16:45 PM · #2
It could mean an emotionally charged or personal subject/technique that voters would either love or hate with not much middle ground.
 12/13/2004 01:18:20 PM · #3
To my untrained eye, I would suggest an irregular bellcurve is mostly indicative of the basic opinion of the voters. A bell curve weighted toward the 10s, a well-liked shot, weighted to "1"s, not so liked. If it isn't even a curve, a very polarizing image. I don't think the votes we get here, on any one picture, form enough of a sample a true bell curve would require. I would bet, that all of the votes on this site, would end up as a bell curve. But any one picture, probably will have a curve-ish pattern.
 12/13/2004 01:30:19 PM · #4
Really? You don't think there are enough votes to constitute a decent curve? From my recollection you only need like 30 samples for a rough significant value. Images normally get in excess of 300 votes. I would have thought that to be a decent sample size.

I see what scalvert is saying. If the bell is distributed normally it suggests people are voting too emotionally. I wonder what else it implies.
 12/13/2004 01:39:06 PM · #5
Every photo has the bell curve displayed (on its side) next to the photo.

Most have a relatively normal distribution, perhaps weighted towards one end based on quality. I actually would strive for the inverted curve, with spikes at either end, indicating a picture which "connected" strongly with voters, love it or hate it.
 12/13/2004 01:55:00 PM · #6
Sometimes, I get more of a French Horn curve.
 12/13/2004 01:58:05 PM · #7
huh? scalvert, could you ASCII illustrate that please?
 12/13/2004 02:03:18 PM · #8
haha scalvert

dsb_mac: i alread forgot some of the things in the statistics class i took last year, but i don't think you can generalize what will constitute a normal distribution and what won't. if people voted on a scale of 1-1000, i think 30 votes may not do as good of a job of representing the general consensus/opinion as a scale of 1-2 would for future voters. hope that made sense (and i'm not sure if it's theoretically correct, but seems like it to me)

edit: upon reflection, maybe i'm focusing on the wrong variable. i think it is more related to what percent of the population is represented - ie: 30 people out of 30,000,000 is statistically insignificant, but 30 out of 31 is gonna give you a pretty darn good idea of the end result.

edit again: i'm not even sure if i answered your initial question. maybe it's hidden in there somewhere.. bah!

Message edited by author 2004-12-13 14:06:03.
 12/13/2004 02:09:37 PM · #9
Originally posted by vontom:
To my untrained eye, I would suggest an irregular bellcurve is mostly indicative of the basic opinion of the voters. A bell curve weighted toward the 10s, a well-liked shot, weighted to "1"s, not so liked. If it isn't even a curve, a very polarizing image. I don't think the votes we get here, on any one picture, form enough of a sample a true bell curve would require. I would bet, that all of the votes on this site, would end up as a bell curve. But any one picture, probably will have a curve-ish pattern.



This is indicative of all the votes cast on all challenges based on some sampling I did. "Your mileage will vary."

Because of the laws of statistics, most curves for a single photo will be "bell shaped" but skewed with a mean toward the 10s for a very good photo or a mean toward the 1s for an unpopular photo. In my opinion, a photo with two peaks ... possibly with one toward the 1s and one toward the 10s ... might be indicative of a well executed shot of a risky subject or a politically charged shot with polarized voters.

Oh, and 200-300 votes will be statistically significant.

With Love and Affection -- The Math Guy

Message edited by author 2004-12-13 14:11:53.
 12/13/2004 02:22:51 PM · #10
Originally posted by dsb_mac:
could you ASCII illustrate that please?


A chart that looks like this. Such a chart just means I confused the voters.
 12/13/2004 02:50:51 PM · #11
Umm... yeah. DQ, thanks for proving my point. The average for all votes is a nice bell curve. I would also like to say, in no way did I suggest that the votes were not "statistically significant" in the very loos concept we have here. I said that the number of votes any given photo receives is not enough to guarantee a true bell curve for every picture (in particular the polarizing ones).
 12/14/2004 04:41:55 AM · #12
interesting graph, dq's and scalvert's. I noticed the graph was flared or fatter at the ends than a normal curve. showing the love it or hate it effect occurs more often than chance would imply.
 12/08/2005 06:12:35 PM · #13
I wondered if anyone had thought abuot have a graph available from the profile page to see how a person's average score changes over time. It would be a great advert for how dpchallenge improves its end users skills, if nothing else. I was just looking at librodo's profile and if you look back, his scores are definitely improving over time. Just thought it would be nice to see in one graph.
 12/08/2005 06:19:11 PM · #14
You only need 42 RANDOM samples to be 95% confident that the results of the sample would accurately reflect the population within one standard deviation. It's about the only thing I remember from statistics, but used extensively during my former career as an internal auditor.

The key is random - since the pix come up randomly for each voter that helps. The reason you see sometimes see your score move wildly within the first couple of days is that the voters themselves have predispositions and could be voting strongly one way or another based on their interpretation of the challenge. I've rarely seen my score move more than .2 once at least 150 votes have been cast.

12/08/2005 06:58:24 PM · #15
Originally posted by A1275:
The key is random - since the pix come up randomly for each voter that helps.

Currently, people still have the ability to "cherry-pick" from the thumbnails page which photos they vote on, so the randomness factor is partially compromised.

Message edited by author 2005-12-08 18:59:28.
 12/08/2005 07:02:16 PM · #16
Originally posted by GeneralE:
Currently, people still have the ability to "cherry-pick" from the thumbnails page which photos they vote on, so the randomness factor is partially compromised.


Exactly right - that's why I think it takes about 100-150 votes to get the 42 purely random - at both ends - sample. Especially the early voters who vote only those they like the best, and then the voters that assign tiers of shots on a first pass before they go back and assign true values, and any other non-random voting patterns I'm forgetting about.
 12/08/2005 07:06:18 PM · #17
I'm actually quite surprised they haven't removed the thumbnails yet. Ah well.
Pages:  
Current Server Time: 10/22/2025 04:55:22 AM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 10/22/2025 04:55:22 AM EDT.