OT - Math (variance) question

[Alf]

I am doing a stats site for Hattrick and plan on using variance/standard deviation to calculate a series strength.

My data will always be in sets of 8 values and I need to calculate how well these values distribute => so I suppose that variance was the "value" I am looking for.


How do you calculate this variance value ?

The following site here (http://davidmlane.com/hyperstat/A16252.html) has 3 different formulaes (the first two look alike to me) whereas the last one has (N-1) instead of N.

Actually, standard deviation might be what I am looking for after reading a bit more (my maths lessons are far away in my mind).

95% of the population is between Mean-2*StandardDev and Mean+2*StandardDev. Am I correct here ?

Thanks for the future replies

[Fonzie]

Hi Alf-

You'll want to use the last formula with (n-1) in the denominator. And yes, +/- 2 SDs encompasses 95% of the normal distribution. +/- 3 SDs covers 99%.

[Alf]

Thanks Fonzie. Care to explain why it is better ?

Why should I use the N-1 formula ? To me the N formula seems more accurate ? Don't be afraid of speaking in mathematical terms, I should understand.

Off to bed. Will read you tomorrow hopefully !

[SirFozzie]

the answer is 2/3.

[JeeberD]

Why did I even bother clicking on a math thread? Ick...

[Fonzie]

Alf -

Sorry, I should've explained my reasoning a bit better. The decision of which formula to use is somewhat subjective, as it depends on whether you believe you are calculating the SD of an entire population (the "N" version of the equation; all data points are available) or just a sampling of the population (the "N-1" version of the equation, which is more common). The subjective part comes with defining the term "population" in your particular case. Is the population going to be defined as all games within this series for this season? If so, then you'll likely have the entire population at your disposal for SD calculation. If the population in this case is, however, something more along the lines of all games in all of Hattrick, then the data you'll have available will simply be a sample of the greater population.

Of course, if you're going to be doing this work in an Excel spreadsheet you can simply use their STDEV function and not worry about any of this stuff.

Hope that helped!

[Alf]

I am not using excel, I am doing this with my own little hands in PHP to show on a web page automatically. But I can do that calculation, no pb.

Each week, in each series of hattrick, there are 4 games => 8 team ratings. So I know exactly that the population has "8" entries (no more, no less). I'll do the test using both (n-1) and n denominators and come back later.

Thanks for your time.