2
$\begingroup$

This might be already asked and answered. However, all the questions and answers for possible similar questions are too high level for me.

So, I will generate a very simple example and please, do not go crazy in statistics or math. I am pretty sure this can be answered using plain terms and using excel functions.

Lets compare how tall are boys and girls. We go to a school and pick 10 boys and 10 girls. Then we repeat the process in school2 and school3.

So we have three sets of data. m is male, f is female. The numbers is how tall are they in meters.

Then we can use excl functions to calculate mean (=average) and standard desviation: enter image description here

Even we can plot it. The error bars are the standard desviation value. Not sure if I should put only the bar going up or both up+down. enter image description here

Now, lets try to combine the three sets of data: enter image description here

My question is... How I calculate the standard desviation (or should be the standard error)? If you can help to answer (without math) just the concept and how to get it in excel it would be very helpful.

Thanks

Edit: From Peter Flom - Reinstate Monica♦ answer I understood: enter image description here Where the standard desviations of the combined graph are calculated as:

STD = SQRT( (std1^2) + (std2^2) + (std3^2) )

SQRT is EXCEL function for square root. std1, 2, 3 are the standard desviation for school 1 ,2,3 respectively. ^ is the power symbol in excel.

enter image description here

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

A few points:

  1. While standard deviations are not additive, variances are (provided the data are independent, which seems likely here). So, add the variances (just the SDs squared) then take the square root.

  2. The plots you have shown are known as dynamite plots. Their reputation is somewhat dubious. See, e.g. this page.

  3. Excel is not really a good tool for statistical analysis or graphics; if you are going to be doing statistics on anything like a regular basis, you should learn a statistical package such as R, SAS, Stata, Python .... Each has strenths and weaknesses, but all are way better than Excel.

  4. Questions about how to do things in statistical packages are off topic here.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Hi Peter. Thanks for your answers. 1. So, If I understood properly, I should calculate the new standard desviaiton as SQRT( (std1^2) +(sdt2^2) +(std3^2)). Would be this standard desviaiton or standard error? 2. Thanks for the warning. I will not use dynamite plots any more but dots or boxes. 3. I do not use EXCEL, I have used just here to create an example that everyone can understand. I use python. 4. I was not asking or discussing statistical packages here. $\endgroup$
    – daniel_hck
    Commented Nov 25, 2019 at 17:43
  • 1
    $\begingroup$ That would be the SD. $\endgroup$
    – Peter Flom
    Commented Nov 26, 2019 at 0:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.