What is considered a high standard deviation?


by 0TheSwerve0
Tags: considered, deviation, standard
0TheSwerve0
0TheSwerve0 is offline
#1
May14-06, 03:03 AM
0TheSwerve0's Avatar
P: 405
I am working on a primatology project where I need to find a way to measure and plot the (somewhat arbitary) level of variation in behavior (ie from one individual to the next) within a single species and then compare it to another species level of variation. I have entered the data for the select behaviors that will go into this calculation, and I have calculated the mean and standard deviation for one species. But what is considered a high standard of deviation? My conclusion depends on such calculations, so any help is appreciated, even suggestions for other ways to mathematically calculate values of variation and represent them in charts/graphs. I only have data for 4-5 individuals per species, so I don't think all charts/graphs will be appropriate.

p.s. I haven't taken any real math since high school, so a simple explanation is best. So far, I've read that if you divide the stdev by the mean and get 10% or over, it's considered a high stdev.
Phys.Org News Partner Science news on Phys.org
NASA's space station Robonaut finally getting legs
Free the seed: OSSI nurtures growing plants without patent barriers
Going nuts? Turkey looks to pistachios to heat new eco-city
Tide
Tide is offline
#2
May14-06, 03:18 AM
Sci Advisor
HW Helper
P: 3,149
That depends on what you are trying to do. The standard deviation is generally less important than the Z-Score - i.e. how many standard deviations a particular measurement is from the mean. If you are attempting to model some process then a "large" standard deviation may indicate some problem with your particular model or data collection methodology.

Standard deviation has meaning only in a particular context. In physics, for example, a large standard deviation in the speed of gas particles means the gas is hot!

Also, if you are attempting to model something, you may be more interested in regression or correlation analysis.
Gokul43201
Gokul43201 is offline
#3
May14-06, 05:57 PM
Emeritus
Sci Advisor
PF Gold
Gokul43201's Avatar
P: 11,154
What you can do is compare standard deviations for the behavioral traits of the two species and say that one species shows a larger deviation than the other. By itself, the standard deviation can't be high or low, without a context. The context is usually provided either by theory (which for something like what you are studying, is hugely inadequate) or by more statistics. If most apecies show a deviation of about d in some parameter, and a particular species shows a much larger deviation (3d, say), you might be able to say something about that species (eg : they like to screw around ). Without something else to compare to, there's not much you can say.

p.s. I haven't taken any real math since high school, so a simple explanation is best. So far, I've read that if you divide the stdev by the mean and get 10% or over, it's considered a high stdev.
That itself must have been written in some specific context. In general, that statement makes little sense. In the standard normal distribution, for instance, the mean is zero. So, by that argument, any standard normal distribution will have a standard deviation that is too high.

0TheSwerve0
0TheSwerve0 is offline
#4
May15-06, 10:11 PM
0TheSwerve0's Avatar
P: 405

What is considered a high standard deviation?


Cool, that's what I did. As for the 10% thing, I just googled a quick search and found a forum message about it. We know how reliable that can be


Register to reply

Related Discussions
Mean Absolute Deviation/Standard Deviation Ratio Set Theory, Logic, Probability, Statistics 6
Standard Deviation Precalculus Mathematics Homework 16
Changes to Standard Deviation?? Set Theory, Logic, Probability, Statistics 8
Standard Deviation? Classical Physics 3
standard deviation Set Theory, Logic, Probability, Statistics 4