Statistical study of the collecting habits of classic gamers

I posted the following to Usenet, some time ago. Figured it might be interesting here, too.
The original data was gathered from Digital Press' Collector's Guide, third edition,
and the data was verified by a list kept by Doug M. of Intellivision collection fame.

    Apparent rank order of systems, by popularity among collectors:
  - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    (a)  Atari 2600         Most cart collector's first-choice system;
                            87% of those polled collected games for it.
                            That's nearly nine-out-of-every-ten people!
    (b)  Atari 7800         This system is fully compatible with the 2600
                            which may account for it begin in second
                            place. 66% of those polled collected for it.
    (c)  ColecoVision       Just under the Atari 7800 in popularity, at 
                            62% ... but a drop-off of 25% from the 2600!
    (d)  Intellivision      Nearly tied with the ColecoVision, at 56%. 
    (e)  Atari 5200         Just under the Intellivision, at 54%.
    (f)  Odyssey 2          Taking another 20% jump downward, at 34%.
    (g)  Vectrex            Staying consistent, at 32% of those polled.
    (h)  Fairchild          Taking another large jump down; now at 18%.
    (i)  Bally Astrocade    Tied with the Fairchild Channel F at 18%.
    (j)  Emerson 2001       And dropping down to 11% to hold last place.
    (k)  Any other systems? Hard to tell ... usually not listed.

What is interesting, is if you go *really* nerdy on this (as I obviously have), and you actually put this data into a chart format, you get two "clumps" or groups of collectors... excuse the lame ascii art, but here's what I mean:

    Number of       Number of persons who 
    game systems    collect for that many systems
     1 system only  5   XXXXX    
     2 systems      7   XXXXXXX
     3 systems      9   XXXXXXXXX
     4 systems      9   XXXXXXXXX
     5 systems      9   XXXXXXXXX
     6 systems      7   XXXXXXX
     7 systems      3   XXX

     8 systems      6   XXXXXX
     9 systems      8   XXXXXXXX
    10 or more     21   XXXXXXXXXXXXXXXXXXXXX

Look at the X's sideways; left-to-right becoming bottom-to-top. Note that from "1 system" to "7 systems" there is a nice, even distribution of data. A "bell-curve" forms. (Or as close to one as you get with only 84 people in your study, and using crude ascii graphics.)

This is interesting -- at least to me -- in itself. The fact that a bell-curve formed at all, is usually a way to see if your methods and conclusions are valid ... it formed, but it didn't include all the data. The results are apparently valid, but the conclusions then are a bit unexpected.

Anyway, then you have an entirely different, second set of collectors starting with "8 systems" and going to "10 or more". Their data shows a sudden, upward spike, shooting far higher than the top of the bell ever got. Very interesting; again, at least to me. It could just mean that the way the data was collected is flawed (not enough systems listed by name for the second group? In other words, they are collecting for machines the others do not even consider collecting for?)

See how the middle of the "bell curve" is from 3 to 5 systems? That means the average collector probably collects for that many systems. According to the survey, it is outside the norm to collect more than that, or less than that. Note that only 5 persons collected for 1 system only; that type of focus is very rare among the collectors in the poll.

Comparing the numbers of people in both groups is interesting, too. There are roughly 60% of all those polled, in the first (bell curve) group. The second (spike) group has approximately 40% of all those polled. In other words, its not a perfect 50/50 split, but each group has nearly as many persons in it, as the others.

What does it all mean? Who knows. But the statistics show SOME picture of what's going on, even if you have to interpret the data closely, to see a picture develop clearly.

Just for what its worth, folks! Real paper magazines do statistical studies of their readership from time to time, so why shouldn't we, eh?

(Note: After posting this all to Usenet, others saw it and commented on it. One person -- sorry I can't recall whom -- said that after collecting a certain number of systems, some collectors say "oh what the hell" and start collecting for everything. I agree; that seems to fit the pattern I've seen, in myself and in others. That, and at some point, collecting starts to get addicting! I wrote up a seperate text on that theme; see my "Are you a collect-aholic?" text to see if you're addicted.)

Ward Shrake

Click here for main menu