2

Given a series of values,

3.00,5.00,7.00,4.00,7.00,5.00,22.00,4.00,6.00,7.00,9.00,6.00,4.00

I need to 'rescale' the data so they have new values with a mean of 0 and a standard deviation of 1. Having followed a number of examples for the equation, I get the following,

Mean    6.85
StdDev  4.85

The standard deviation I worked out the long way via taking an original value, subtracting the Mean, squaring the result, summing all the squares, dividing that sum by the count -1 before finally getting the square root of that. 

(3.00 - 6.85) = -3.85    >    -3.85² = 14.79
(5.00 - 6.85) = -1.85    >    -1.85² =  3.41
...
(4.00 - 6.85) = -2.85    >    -2.85² =  8.10

Sum                                   281.69
Variance (using n -1)                  23.47
Standard Deviation                      4.85

I'm sure it's correct as I checked it here http://www.mathsisfun.com/data/standard-deviation-calculator.html

Using the mean and standard deviation I get new values of,

-0.79,-0.38,0.03,-0.59,0.03,-0.38,3.13,-0.59,-0.17,0.03,0.44,-0.17,-0.59

These values were calculated in the following manner 

((3.00 - 6.85)/4.85) = -0.79
((5.00 - 6.85)/4.85) = -0.38
....
((4.00 - 6.85)/4.85) = -0.59

So, to my question.

It's the standard deviation that is the confusing part. I want my new values to have a Mean of 0, and a standard deviation of 1. When I work out the SD for my original values, I get an SD of 4.85. What do I need to do to get an SD of 1 ?

Thank you for taking the time to read the question. John.

Macro
  • 40,561
  • 8
  • 143
  • 148
Johnny
  • 161
  • 1
  • 1
  • 6
  • 2
    Your new values do have a standard deviation of $1$ because after subtracting the mean from each of your original values you divided by $4.85$. Where does the confusion arise? – M. Berk Sep 24 '13 at 11:11
  • @M. Berk. Thank you. The confusion came about because, this is all rather new to me and I thought I would need to do soemthing to 4.85 to get a value of 1. – Johnny Sep 24 '13 at 11:25
  • Take a look at [gung's answer](http://stats.stackexchange.com/a/70555/21054). He nicely explains the different transformations. In your case, you are doing a *standardization*. A standardized variable has a mean of $0$ and a standard deviation of $1$, which is also called [standard score](http://en.wikipedia.org/wiki/Standard_score) or $z$-score. – COOLSerdash Sep 24 '13 at 11:25
  • @COOLSerdash. Brilliant, thanks. Another thing that has confused me are the terms used. Standardization / Normalization I now know are different things, but the terms are used interchangeably on a lot of sites. This will certainly help me. – Johnny Sep 24 '13 at 11:28
  • @JTT. as mentioned ealier, this is new to me and I guess I expected to do something else to the new values to end up with a standard deviation value of 1, rather than 4.85. – Johnny Sep 24 '13 at 11:37
  • Tried to mark the question as 'answered' using @M. Berks comment. However, I'm unable to do this for another 7 hours. – Johnny Sep 24 '13 at 11:55
  • The SD of your new values is $1$, which is what you wanted: what's the problem? – whuber Sep 24 '13 at 20:34

1 Answers1

2

I think it is best to post an answer. It was M Berk's comment that was the revelation for me ...

Your new values do have a standard deviation of 1 because after subtracting the mean from each of your original values you divided by 4.85.

Thank you to COOLSerdash and JTT for the other comments too.

whuber
  • 281,159
  • 54
  • 637
  • 1,101
Johnny
  • 161
  • 1
  • 1
  • 6