Definitely a right-skewed distribution and, from the frequency distribution tabulated below, it appears that we may have a bimodal distribution. This may be indicative of there being two discrete populations that have been lumped together. In this circumstance, those two populations just might be:
1). ordinary collectors/afficianados and 2). certifiable lunatics.
Thinking that I'm awfully close to fitting into the second group...
Charlie "Statistics means never having to say you're certain" Young Honey Brook, PA
p.s. Harv:
Shoot me the Excel file and I'll do a bit more stats work on it. Distribution goodness-of-fit testing via Shapiro-Wilks and Kolmogorov-Smirnoff, maybe some upper confidence limits of the mean via the H-statistic if it is lognormal, etc.
> Hi Kids,
>
> The final bike count report I came up with was:
>
> 3120 bikes, 169 people reporting = my simple average of 18.46 bikes per
person.
>
> But a more formal assessment using raw numbers (no names of course), which
I provided to our esteemed co-lister Harvey Sachs, yielded the following
much more interesting statistical information:
>
> >167 unique and useful responses.* 10 is the median number of bikes
reported in a correspondent's collection.** That is, half the respondents
had 10 bikes or fewer, and half had 10 or more. I have a hunch this will
give "cover" to a fair number of people who differ from their spouses on the
optimum size of a collection. :-)
>
> 2 reported 100 or more bikes (105, 300)
> 11 reported 50 - 99 bikes.
> 7 reported 40 - 49
> 7 reported 30 - 39
> 18 reported 20 - 29
> 15 reported 15 - 19
> 29 reported 10 - 19
> 46 reported 5 - 9, making that group the most common collection size.
> 33 reported 1 - 4 bikes
>
> >I'd be happy to provide the data in Excel form to anyone who asks, and
have attached it for you, too.
>
> >Harvey Sachs
> >McLean VA
>
> >*There were a few other entries marked "add on" that did not refer to
specific entries, so I couldn't use them. The one entry of 22 marked "add
to 50" was included. The other six "add on" probably make little difference
for our purposes.
>
> >** For the Wonks, the "mean" was almost twice as high as the median,
indicating a strongly skewed (non-Gaussian) distribution. This means, among
other things, that the arithmetic "standard deviation" is meaningless, and
not reported. Wow, does that sound official, or what? :-) Your mileage
may vary.
>
> End Quote. If you want the Excel spreadsheet, Harvey's email address is:
sachs@erols.com
>
> Very, very best holiday wishes to you all,
> Paulie Davis
> Los Angeles, California