RobertNiles.com
About Robert
Reporting Help
Finding Data on the Internet
Questions for Candidates
Stats Lessons
Mean
Median
Percent
Per capita
Standard Deviation
Margin of Error
Data Analysis
Sample Sizes
Stats Tests
Help Board
Bookstore


The Cartoon Guide to Statistics

This isn't some dumb-downed whitewash. It's a must-read for a beginner taking his or her first steps toward understanding stats.
More information
-->

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

pooled standard error
Message posted by Lisa Appeddu (via 164.58.150.32) on October 5, 2001 at 11:19 AM (ET)

I have three treatments with uneven experimental units. How do I calculate a pooled standard error of the mean?


READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: pooled standard error
Message posted by Tomi (via 154.32.142.220) on October 5, 2001 at 4:27 PM (ET)

Let's refer to your three sets as x_i, y_i and z_i, with differing numbers of measurements being n_x, n_y and n_z.

Then you can calculate the mean values by: x_bar = sum {from i=1 to i=n_x} (x_i) etc.

For each set calculate the sum of the square deviations: ssd_x = sum ((x_i - x_bar)^2) etc.

The pooled standard deviation will be: psd= (ssd_x + ssd_y + ssd_z)/(n_x + n_y + n_z - 3).

The standard error of the mean can then be calculated by: t*psd/(square root (n_x + n_y + n_z)), where t is taken from t tables and depends on the confidence level.

Hope this helps.


Re: pooled standard error
Message posted by Jack Tomsky (via 208.249.113.130) on October 5, 2001 at 6:24 PM (ET)

In Tomi's notation, the standard error of the mean is sqrt(psd).


Re: pooled standard error
Message posted by Tomi (via 154.32.143.67) on October 6, 2001 at 2:32 AM (ET)

As Jack pointed out, there was an error in what I wrote. I should have said psd = sqrt( (ssd_x + ssd_y + ssd_z)/(n_x + n_y + n_z - 3)).

I still think that this is the pooled standard deviation, not the pooled standard error of the mean, and that the next step is still necessary. Am I wrong, Jack?


Re: pooled standard error
Message posted by Jack Tomsky (via 208.249.113.130) on October 8, 2001 at 2:03 PM (ET)

If xbar is an unbiased estimate of the population mean, then the standard error and standard deviation are equivalent. As a separate issue, the confidence interval on the population mean is

xbar +/- t*psd/sqrt(n_x + n_y + n_z)



Your $5 contribution helps cover part the $500 annual cost of keeping this site online.

Niles Online Sites:RobertNiles.comTheme Park InsiderViolinist.com

RobertNiles.com™, the site, content and services 咀opyright 1996-2002, Robert Niles.
All rights reserved. Questions? Comments? Read my Privacy Policy, or E-mail me!