Determining the percentage of points away from the Mean

[nemesis24]

I am doing a summer project where i have got lab data of fluxuating flow patterns in a flume, the data i have plots discharge against time and my aim is to find whether a majority of measurements are close to the mean.

I thought the best way to do this would be with standard deviation where i could find out coeffient of variance in one SD (SD/mean*100), but my question is related to finding the CoV in second and third SD - is this even possible?

 

[mathman]

I am unfamiliar with the terms you are using. However variance is the average of the square of the differences from the mean while standard deviation is square root of variance.

 

[nemesis24]

Yeah it seems like i should expand on what I am trying to do to best describe the work i want to do.

I have a lot of data where i have to show mathimatically there is a deviation from the mean of the sample data, i have decided i should do a paired t-test to show the variance. However, i am not entirely clear how to conduct a paired t-test for my given data - the sample i have in question has measured the velocity of flow in a flume at 50hz (so measured velocity once every 0.02sec), what i plan to do is use the data to show the data trend over a 1hz (or every 1 sec) then use a paired t-test to analyse the total sample data and the 1hz one.

I am just unsure how i can do that because for the total data i have over 2000 points but for the 1hz sample i have a little over 100, so working out the difference is a little awkward - and the SD's... I hope this makes a little more sence

 

[mathman]

I am not particularly familiar with statistic analysis, so it is hard for me to help. I don't know what a paired t-test is.

 

[statdad]

If I'm understanding your "explanation" then I have these comments:

You mention doing a paired t test, but your sample sizes are different? Paired t tests require equal sample sizes (the "classical" setting is is comparing results on pre- and post-test data)

This: " thought the best way to do this would be with standard deviation where i could find out coeffient of variance in one SD (SD/mean*100), but my question is related to finding the CoV in second and third SD - is this even possible?" - do you mean 2 and 3 standard deviations from the mean, or are you somehow referring to higher moments of the data?

Finally, "what i plan to do is use the data to show the data trend over a 1hz (or every 1 sec) then use a paired t-test to analyse the total sample data and the 1hz one."
Does this mean: for the data collected every .02 sec, begin at the first measurement and combine every 50 to represent 1 second, and use this to compare to the other data? if so, perhaps some more explanation - but even so, I doubt that would work.

 

[nemesis24]

Yeah maybe i should start again from the beginning (PS thanks for the reply and sry for the late post but i lost internet for a few days so only just saw the update), also i should point out my background is not in stats at all so i may be throwing some terms about that i picked up from the internet.

I have a lot of data and i have split this data into smaller segments so i have a graph that looks like the attachment.

What i want to do is compare the blue variance to the green line which is fairly steady, i was thinking i would use a paired/unpaired t-test to do this but again i am have no bases in stats and don't know if that is the correct way to go about the situation.

 

[Bacle]

I don't know if I am too far off here, but, AFAIK, paired t-tests are used to
t-test whether m-means of different populations are e-equal (Sorry for the
bad joke). Is that what you are trying to do.?

 

[nemesis24]

Im just trying to show that there is a different bettween the two sets of data that would reject a null hypothesis, i thought a paired (or to be exact a unpaired) test would show this