For the maths/stats people (Mean and STDEV)

20x is the derivative of x^20. Its integrate is x^20+c


Lol i have no idea what that actually means

DOH, im trying to ploy Standard deviation on a graph,

not sure how to do it if the deviation is not spread evenly around the mean :S

:(

ahh, my graphs gona be huge and unreadable. :~|

and thats still assuming the STEV to be evenly spread around the mean :S

infact,

my mean is 84, my stdev is 22.

its obviously not evenly spread around the mean or i would have a value >100% if i plot the stdev like that :S

JohnKerr2

No.

If you have a mean of 20 and a Standard Deviation of 5, you're essentially saying that most of your sample fall within 5 points of the mean (so between 15 and 25).

It varies if you have a non-standard distribution.

how does that work if the mean is say 95% with a STDEV greater than 5? say 10

Cant be 85-105% can it?

JohnKerr2

If your data is skewed (that is, the plot doesn't fall in a symmetrical curve) then means and standard deviations aren't particularly useful.

if the data is skewed the median is a better description of the data as opposed to using the mean.
 
The inter quartile range i think is then more useful than the SD also
 
 
 

The standard deviation is the common measure of statistical dispersion. A low standard deviation means most of the values will be close to the mean.

just t

DOH,


back to the drawing board

thanks johnkerr

your welcome also....considering he said nothing i hadnt already said.....
 
 
;(

john_cappa

JohnKerr2

If your data is skewed (that is, the plot doesn't fall in a symmetrical curve) then means and standard deviations aren't particularly useful.

if the data is skewed the median is a better description of the data as opposed to using the mean.
 
The inter quartile range i think is then more useful than the SD also
 
 
 

Can really use the median.


I have 5 sets of results comparing 3 different treatments to a control and im trying to display the mean differences from the control from each of the treatments.



E.g. Measuring  enzyme inactivation in cabbage after cooking compared with a raw sample

                        1           2          3          4          5        Avg       STDEV
Steamed          44        97        88        93        97        84          22
Boiled               98        99        96       98        100       98          1
Microwaved      97        98        97        99        100     98           1

I have plotted the Avg (mean) on a graph, wanted the error bars to display the stdev to show any potential overlap between the steamed and other treatment.

I have done a statistical analysis to show the significance, but it doesnt show a significant difference between the steamed and the other treatments because they are being compared with the raw which is MUCH higher and i cant really rerun the stats without the raw to show the sig dif because i got told that is not the correct protocol, so i was trying to display it graphically

So you have samples of just five for each condition?