Metal wheelsets... a statistical analysis

Three popular brands, 200 each
As compared to 100 plastic sets.

I'm not sure what conclusion you'd draw from this, other than the obvious advantage of metal wheelsets.

I did a stastical analysis on seven groups of 100 wheelsets:
1.) 100 InterMountain
2.) 100 InterMountain
3.) 100 Tangent
4.) 100 Tangent
5.) 100 Kadee
6.) 100 Kadee
7.) 100 Accurail

The tests consisted of 20 groups of 5 measurements from each group.
I then constructed bell curves from each.

Measurement #1 was overall axle length from point to point.
[Note]: The test was conducted for consistency only. I've never seen a spec for axle length, and I'm not sure there is one.
However, I assigned arbitrary numbers as highs & lows as min/max specs, for purposes of the test, and to confine the bell curves into readable parameters.

Good bell curves are normally fairly wide distributions, even within the range of specifications, but the three metal-wheel curves were somewhat narrow... indicating control and consistency.

RESULTS:
InterMountain were all toward the higher end of the overall distribution... 1.011"/1.022.
(Range = .011", Avg = 1.015").

Tangents were all toward the lower end of the distribution... 1.002"/1.010".
(Range = .008", Avg = 1.006").

Kadees were more widely distributed... 1.002"/1.016".
(R = .014", Avg = 1.010").

The plastic Accurail wheelsets however, were a different story.
Not only were they widely distributed, but the bell curve showed two peaks, which is symptomatic of different batches.
(R = .031", Avg = 1.017").

Measurement #2 was wheel-gauge over flanges.
RESULTS:
No metal wheelsets were out-of-gauge = 0%.
Four Accurail wheelsets were out-of-gauge narrow. Two were OOG wide = 6%.

The only defect I found in any of the metal wheelsets was 6 out of 200 Tangent were discolored, e.g., 'half moon' brownish wheels.