No Relative Percentages In This Case
Hello DMM,
In this example, calculating the relative percentage of a difference between percentages makes no sense simply because the difference of 1% (or 40%rel) between 2% in Set I and 3% in Set II is not statistically significant but due to randomly distributed variations in measurement procedures.
Student's t-test shows that the calculated t-value of 0.113 for the difference of 0.20% (or 4.1%rel) between means of 5.0% for Set I and 4.80% for Set II is below the tabulated value of t0.05;4=2.776 at 5% probability. Hence, the observed difference is statistically identical to zero, and not a measure for a systematic error or bias. By implication, no significance can be attached to either the magnitude or the sign of the difference in percent or relative present between any of these pairs.
An example of Student's t-test can be found in "Sampling in Mineral Processing", a .pdf copy of which is posted on my website. Let me have you email address if you want me to transmit an Excel template with you own data. Be aware that borrowing or buying a statistical textbook may cause addiction!
Kind regards,
Jan W Merks ■
Can I use a percent of a percent?
I hope I do not bore you with this trivial question but I need some advice
Take two data sets I and II
I II
a 2 3
b 4 7
c 12 5
d 3 3
e 4 6
I want to compare the diferences in the percentage totals for item a
% a in I = 2/25 = 8%
% a in II = 3/24 = 12.5%
Q- What are the negitives if I take the percent difference of these percentages? Is there any reason I would not want to use this as a basis of comparision to the respective quantities in each group?
(.08 - .125)/.08 = -56.25%
THANKS!! ■