Relationship Between same numerical values

39 Views Asked by At

Let me explain my scenario. I am calculating the percentage of prepositions per document. In beginning, I have taken 20 mobile phone blogs and my mean preposition percentage in 20 blog documents comes out to be 11.61%. Now I have to compare this mean value to the percentage of prepositions in single document lets say it is 10.55%. Now I have to compare these two values i.e 11.61% and 10.55% so as to tell that how much single document deviates from standard data. Can someone suggest me some metric, that would enable me to calculate the difference.

1

There are 1 best solutions below

2
On BEST ANSWER

$11.61\% - 10.55\%=1.06\%$ This is a flip answer, but it is meant to indicate that you need to think about what you are trying to measure. A traditional approach would be to compute the standard deviation of your sample and say that the individual one was so many sigma away from the mean. While that is a well defined calculation, it tends to make people assume the distribution is normal, which is probably not the case. Another approach would be to say the individual one was $35$th percentile or some such. As it relies on counting instead of standard deviation, it is a more robust statistic. Maybe you just found my blog that is just a list of prepositions. Your standard deviation will increase a lot, but the percentile won't change much.