Statistics, Damned Statistics, and the 'Visio Populi'

By Ctein

I don't get much of an opportunity to find out which of my photos are most popular among my audience. The reason is that almost all my sales are high-end prints, so I rarely sell many of each. The "best sellers" sell a handful of copies; most sell just a few. It's not a good sample set, statistically.

I have sold scores, even hundreds of prints in the four special print sales I've run over the years through Photo Technique magazine and The Online Photographer. The selection set is so constrained, though, that it still doesn't tell me anything interesting.

Recently I got to have some fun with numbers. I put up a web page with 14 photographs for my Contributors to choose their prints from.

Almost none of the Contributors had seen any of these photographs in the flesh and nobody had seen all of them, so people really were picking based on the online images. The selections include a few photographs that Contributers suggested, but there were only a very few suggestions, so it doesn't much skew the results.

Forty-five of my Contributors selected their prints, which was a large enough sample to run some numbers with. The results entertained me. I am perhaps too easily entertained, but I hope you'll also get some entertainment value out of seeing how close your tastes are to my collective audience's. I don't know what you'll find. Even more interesting, I suppose, would be if any of you who aren't especially enamored of my work wind up ranking the photographs much as my audience does. That's a particular and most rare talent beneficial to curators, gallery owners and editors—being able to anticipate the taste of the visio populi, even when it isn't to your personal taste.

Including a dye transfer ("Sealing Wax Palm") throws off the results a bit, as I gave Contributors the choice of picking either the dye transfer or two digital prints; a bit of apples vs. oranges muddles the results. Fourteen dye transfer prints were chosen in lieu of two digital prints. Amazingly, that is about the percentage I would've guessed. A fluke.

The rest selected 100 digital prints; here they are listed in order of popularity:

Columbia at Dusk—15
Floodlit Colombia—13
Alcoa Building—12
Apollo-Soyuz in Floods—9
Reflections in Montreal Centre—8
Grasses in Liquid Sky—8
Weathered Sulfur Vents—8
Trojan Nuclear Plant—7
Buena Vista Park—7
Moon Bow and Blue Moon—5
Jazz Dinner—4
Christmas Oranges—3
Blue Forest—1

The two black-and-white photographs did better than I expected, seeing that's not what I'm known for. (The fact that I think they're really great photographs is, of course, irrelevant.)

Not counting the dye transfer, more than half the selections were for older photographs. I'm considering "Weathered Sulfur Vents" a "new" photograph, because it's part of the last decade, and all the other old stuff is really old. If one restricts "new" to the past year or so, barely more than a third of the selections were for new photos. I didn't particularly expect that, but seeing as the "new" work is having to compete with that culled from nearly 40 years of photography, I suppose I shouldn't be surprised. Were it to play out differently, it would suggest that my past taste wasn't anywhere close to as good as I thought it was, or that I had become incredibly better in just a few recent years. Anything's possible; not everything is likely. On the other hand, I'm disappointed the bottom three didn't do better. I really like them! I know, something has to be last, but still..

Make of it all what you will. As I said, for amusement value only. I hope you're appropriately amused.

Ctein

Ctein's regular weekly column appears every Thursday morning.

Featured Comment by - et -: "Just to throw a little more confusion into the discussion, I'll reply to the question by Marcus, in the comments, about the 'scientific bent' of contributors. I am an aerospace engineer who worked on the Apollo Program for the first eight years of my professional career. My print choices were 'Jazz Dinner' and 'Buena Vista Park.' However—if 'Apollo 17 at Dawn' had been one of the choices, that print would have been my preferred choice by far. Strictly a personal matter due to memories and my association of the image with a special part of my life—not directly due to either my scientific bent or the artistic value of the print. (Although I do love that image from an artistic viewpoint also.) How do we account for factors like that in looking at the statistics? I suggest that we get many, many more people to contribute to Ctein, so the sample size will become much larger and the results more statistically valid. :-)"

Comments

[Marcus]

Interesting that the top pictures are shuttle pics. Do you have any idea how may of the subscribers have a scientific bent either professionally or personally? I choose two from the bottom, both classed new and opted for the dye transfer print

[Christine Bogan]

I'm one of the contributors and I found me in a conflict with the choice. For me it was "Historical event beats art." That's why I have chosen the columbia photos. I thought such photos I'll buy once in a lifetime. My next choices would have been grasses in liquid sky and Jazz dinner.
Ctein, it also depends on the meanings of the pictures for the contributors which are in choice. It's not only about your art.
But next year is a new choice!I hope so.
Christine

[Tahir Hashmi]

If I were to guess the reasons why the last 3 turned up last, I'd say:

* They're common scenes -- most people would be able to find them on their own and shoot them fairly well.

* There probably is some context that you carry for the photos, having been there when they were shot, but the viewers don't have it.

[David Zivic]

Interesting, I'll tell you what I find interesting. I have seen a photo of Senor Ctein, and understand he is an expert at a photographic printing process that is fading away. Yet, his 5 most popular images are of a subject matter that is at the forefront of mans technical achievements.....I'm just saying !

[Tom]

With a line like this in the description ...

""Apollo-Soyuz in Floods" is another Contributer request. I made this photo in 1975; it took me 3 years to figure out how to print it."

... I had to see a print.

[Bahi]

There may well be a significant subset who like Sealing Wax Palm (dye transfer print) and Christmas Oranges—the colours, tones and composition might appeal in similar ways. We went for the Sealing Wax Palm (the chance to see any dye transfer of yours seemed too good to miss, partly because we thought the fact that you chose a particular picture for this process said something about its tones and its colours that a web page doesn't) but the Christmas oranges would have been the second choice, along with one of the (spectacular looking) Shuttle launch pictures. Could well have been different if we'd visited an exhibition, though--very frustrating to see so much work online by someone who deals with tone, texture and detail so well.

It sounds like Mike is, to some extent, dealing with the opposite sort of problem.

BTW, haven't received the print yet but that's my fault. It's at the local PO. Going to call them now!

[Cyril]

For what it's worth, using simple approximation one would expect that if 100 people chose from 13 pictures at random (clearly not the case here), most pictures would get picked between 3 and 13 times.

So it's not clear that the actual numbers are that significant (as you write, "something has to be last"). Maybe the subject ordering (3 "space pictures" in the top 4) is more significant though.

[Ken Sky]

An old joke: My Jewish mother wanted to make me feel guilty yet again. For my birthday, she gave 2 shirts. When I went to my room to try them on, and returned wearing one of them to show her how well it looked, her response was, "So what was wrong with the other shirt?" So what's not to like about Blue Forest? ;)

[Dennis]

I took a look at the page alongside the stats, and my personal tastes would run similar. Building on a couple other comments, I'd suggest that the more popular photos are better suited to being displayed by themselves while the less popular ones (at least in some cases) might be better suited to being displayed in a portfolio. The Christmas decorations photos in particular strike me like that - I would be far more interested in seeing a series ofthose than one or two. If you wanted to have more fun with numbers, you might try to make sense of picks by individuals - did some people choose both the Columbia and the Apollo prints ? Or did people who chose the more popular prints also choose less popular prints ? Did the person who chose blue forest also choose 'oranges' ?(i.e. were people trying to put togehter their own series or going for diverse standalone images ?)

[Ray Hudson]

Trivia warning!

I think Trojan was in OR. (We had a big fight about it being here.) You were standing in WA when you took that picture maybe. I think (emp. on 'think') the protocol is to name where the thing is, ie. 'Mt. St. Helens, WA', not 'Mt. St. Helens, OR' 'cause I happened to be in downtown Portland when I took it. I'm just saying.

Ray Hudson

[Nathan deGargoyle]

Which one you choose as a print probably depends to some extent on what you are going to do with it. How does it fit in with your decor, your family's taste, how you want to present yourself to visitors (technophile ot technophobe?).

To take this to an extreme, if I was looking for a desktop for my laptop I'd choose Columbia by Searchlight to look at whenever I booted up (if I didn't have this rather nice pic of my granddaughter there ;) )

But for the wall in my study at 12x18 I'll take Alcoa Building, Pittsburgh ~ 1972 every time.

It's horses for courses!

[ctein]

Dear Cyril,

Quite so. While the outliers are rather overpopulated for a Poisson distribution, it's not so much so that the majority of the preferences couldn't be chance. My instinct is not, but the way to find out will be to look at the picks by the next 20 Contributors and see if they follow the same overall pattern.

My scientific instincts are saying this'll turn out to be real, but there were some more subtle correlations that, while curious and intriguing, haven't convinced me of their reality. So i didn't write about them

(Note to non-math types-- we're geeking here on some subtleties of doing statistical analysis. Contrary to what they suggested to you in high school, getting meaningful statistics is very, very tricky. Aphorisms aside, it's not all that easy to lie with statistics; words are so much better suited to the task. But it is very, very easy to be mistaken.)

pax / Ctein

[ctein]

Dear Ray,

Hah! You're right!

I had it misremembered as being one of Washington State Power's ventures.

I'll sign it differently in the future. The seven extant prints shall become highly collectable and excessively valued as a result, no doubt.

Perhaps I should 'accidentally' sign a few upside-down. The rare "inverted Ctein" variant. Hmmmm.

Laurie Edison and I did a considerably more malevolent take on the plant in '91 or '92, shortly before it shut down:

http://ctein.com/collab14.htm

pax / Ctein

[- et -]

Just to throw a little more confusion into the discussion, I'll reply to the question by Marcus about the "scientific bent" of contributors.

I am an aerospace engineer who worked on the Apollo Program for the first eight years of my professional career. My print choices were Jazz Dinner and Buena Vista Park. However - if "Apollo 17 at Dawn" had been one of the choices, that print would have been my preferred choice by far. Strictly a personal matter due to memories and my association of the image with a special part of my life - not directly due to either my scientific bent or the artistic value of the print. (Although I do love that image from an artistic viewpoint also.)

How do we account for factors like that in looking at the statistics? I suggest that we get many, many more people to contribute to Ctein, so the sample size will become much larger and the results more statistically valid. :-)

[Philip Partridge]

I'm not at all surprised by the choices at either end of the distribution. Most people in this increasingly urban, artificial world seem to like edifices and light/sound/action to the relative disregard of nature or subjects they feel they could have shot themselves. There is a reason for all the flashes going off at the Olympics etc. Subtlety is not a big selling point in the mass marketplace...