Avan:) Posted January 1, 2012 Report Share Posted January 1, 2012 What is the highest standard deviation that is acceptable in IB lab reports? For example, when you do tests like the mann-whitney test (in psychology) there is a certain value and when you get a value above it, you know it's significant (or insignificant). The problem with the standard deviation is that the "best" or least value you can get is 0, but then there is no maximum for standard deviation. I got 20.83 for some results, and is that ok? How do I know whether that is too large anyway, for any experiment? thanks Reply Link to post Share on other sites More sharing options...
Sandwich Posted January 1, 2012 Report Share Posted January 1, 2012 The thing you're talking about from Psychology would be the "p" value (>0.05, >0.1 etc.).You're correct in saying that you can't have a maximum or minimum standard deviation. I'm not entirely sure that you've understood the concept of SD though - which is fine, all stats is confusing. For me, haha.20.83 will be the value of ONE standard deviation for your experiment. This means that, assuming your data has a normal distribution, about 68% of your population has a value of the population mean +/- 20.83. So you can actually tell a lot just from this.http://en.wikipedia....ion_diagram.svg^^ that's a normal distribution. You can see that 68% (34.1+34.1) lies either side of the mean - i.e 1 standard deviation!If your mean is 400 and your range is from 100 to 800, then +/-20.83 obviously means you've got a lot of data sitting around the mean. If your mean is 50, the range is 10 to 80 and you have +/-20.83 either side of that, obviously your data really isn't sitting around the mean. In fact, 68% basically includes almost all the whole range of data in this case. So you CAN use the standard deviation to interpret your data - just add 20.83 on either side of your mean value and you should be able to see for yourself whether it looks significant or not!Hopefully that makes sense. There are things you can do to get a p value from an SD (I think via conversion to a Z-value... got to love stats...) and you should also look at finding the 95% confidence interval. Neither of these are necessary features of a lab report though.But hopefully you now understand your SD a bit better and can write about it! 1 Reply Link to post Share on other sites More sharing options...
Avan:) Posted January 1, 2012 Author Report Share Posted January 1, 2012 woah, thanks! You "said" it in such a nice, patronizing way ... I feel dumb right now yup, gotta love stats! Reply Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.