Re: Final Rating for The Invention of Nature by Andrea Wulff and other matters - lol


Merilee Olson
 

✔️❗️

On Wed, Sep 1, 2021 at 11:40 AM Becky Lindroos <bekah0176@...> wrote:
True -  but I’ll just round down.  I don’t put scores of more than one decimal point on the site:

https://groups.io/g/AllNonfiction/table?id=4030

If that comes up screwy click the top of the ID column (or any column you want to peak at -  the ID column is better than the date column because of something - ???) .

Becky

> On Sep 1, 2021, at 9:40 AM, Merilee Olson <merilee.olson@...> wrote:
>
> If no one has sent in a score LONGER than two decimal points…
>
> On Wed, Sep 1, 2021 at 7:50 AM Becky Lindroos <bekah0176@...> wrote:
> You’re right Merilee.  I think the precise decimal point to use depends either on what you’re measuring or on the data points themselves.   It’s like regular fractions -  you can’t have a bus with 8.75 kids on it even if the data figures out to that.  But you probably shouldn’t have a book rating of 8.4532875 if no one has sent in a score shorter than two decimal places.  Just my observation - and yours apparently - and I think it’s a good thing.  :-)  lol! 
>
> Becky
>
> > On Aug 31, 2021, at 2:51 PM, Merilee Olson <merilee.olson@...> wrote:
> >
> > Aaaaarrrggghhh🙀
> > Fwiw, I do remember learning that you should never state any of your averages to more precision (decimal points) than your least precise data point. However, I think we can cheat a tiny bit and express results to 1 decimal point. Say we had an 8 and a 7, I think we could say 7.5.
> >
> > On Tue, Aug 31, 2021 at 3:28 PM Becky Lindroos <bekah0176@...> wrote:
> > In honor of Kehlman and company I decided to do this both my way and the other way (and possibly only of interest to Merilee?)
> >
> > Easy-peasy method (mine with focus on the mean):
> > Raw ratings:
> > 7 + 8.5 + 10 + 10 + 10 + 10 + 10 + 9.5 + 8
> > Mean: 9.2222222222222222 - 8
> > Median: 10
> > Mode:  10
> >
> > Range   3
> > Minimum         7
> > Maximum. 10
> > Count   9
> > Sum     83
> >
> > Outliers        none
> > ***********
> > **AND***
> > ***********
> > Descriptive Statistics answers (with focus on the median - outliers removed but too bad - there aren’t any):
> >
> > Minimum min =   7
> > Maximum max =   10
> > Range   R =     3
> > Size    n =     9
> > Sum     sum =   83
> > Mean    𝑥⎯⎯⎯
> > x
> > ¯
> >  =      9.22222222
> >
> > Median  𝑥˜
> > x
> > ~
> >  =      10
> >
> > Mode    mode =  10
> >
> > Standard Deviation      s =     1.12113534
> > Variance        s2 =    1.25694444
> > Mid Range       MR =    8.5
> >
> > I don’t remember this:
> > Quartiles               Quartiles:
> > Q1 --> 8.25
> > Q2 --> 10
> > Q3 --> 10
> > Interquartile Range     IQR =   1.75
> >
> > Outliers                none - would be a score of 11.75 or 6.55 (10+1.75 or 8.25-1.75)
> >
> > Sum of Squares  SS =    10.0555556
> >
> > Mean Absolute
> > Deviation       MAD =   0.925925926
> > Root Mean Square        RMS =   9.28260021
> >
> > Std Error of Mean       𝑆𝐸𝑥⎯⎯⎯⎯
> > S
> > E
> > x
> > ¯
> > =       0.373711779
> >
> > I don’ think I ever learned this stuff:
> > Skewness        γ1 =    -1.23197925
> > Kurtosis        β2 =    4.87016357
> > Kurtosis Excess
> > (Kurtosis in Excel and Sheets)  α4 =    0.298734994
> > Coefficient of
> > Variation       CV =    0.121568892
> > Relative Standard
> > Deviation       RSD =   12.1568892%
> >
> > (Extra)
> > Frequency Table. (And we can see there are no outliers)
> > Value   Frequency       Frequency %
> > 7       1       11.11
> > 8       1       11.11
> > 8.5     1       11.11
> > 9.5     1       11.11
> > 10      5       55.56
> >
> >
> > **** So after reading Noise and *thinking about it* median is best for large groups with human noise error involved.  Like in student test scores some might have had a very bad day - some might have had a very good day.  Two students answered (incorrectly) only 1 out of 7 questions. One student is a genius anyway. Outliers will occur on almost every test for which you are trying to establish a class curve (or a hybrid multi-class curve) unless they are very difficult or very easy tests.  As a teacher you want to those kinds of outliers if you can.  They shouldn’t count for or against a whole class. 
> >
> > But with weather statistics nothing is a human-made outlier - all scores are in the range of scientifically measured evidence.  Don’t automatically discount that temperature of 118º just because it’s the highest temperature ever recorded in August: considering climate change, our mean temperatures may be rising.     Otoh, a reading of 178º might be a bad data point, an outlier. (Someone goofed.) .
> >
> > Our ratings are highly influenced by sometimes unpredictable and unmeasurable human judgement and can be very susceptible to outliers. The reader/rater just could not put up with the screechy voice on the Audible tape or the rater was sick and irritable with Covid for 2 weeks.  Whatever. 
> >
> > So I’ll omit the outliers, if there are any, because that’s another point. We haven’t really had any outliers to this point.  If we do I might change the method but …
> >
> > https://www.itl.nist.gov/div898/handbook/prc/section1/prc16.htm
> > **Outliers should be investigated carefully. Often they contain valuable information about the process under investigation or the data gathering and recording process. Before considering the possible elimination of these points from the data, one should try to understand why they appeared and whether it is likely similar values will continue to appear. Of course, outliers are often bad data points.
> >
> > Becky
> >
> >
> >
> >
> >
> >
>
>
>
>
>
>
>






Join AllNonfiction@groups.io to automatically receive all group messages.