ONE-WAY ANOVA

[Link Map]
begin comments text sound
line

To carry out a one-way analysis of variance, select Analysis of Variance > Fixed Effects. Just type the desired formula or create it by clicking the Create Formula button.
Enter the variable of interest (drytime) as the Response and enter the treatment factor (treatmt) as the Main Effect: (+) factor.

You can request specific sections of output before the analysis of variance is carried out by clicking on Results in the dialog box or after the analysis.
To display standard error of means (as well as the default options), select Results from the Analysis of Variance dialog box and check the box for Means. The output is shown below:

The above shows that differences between treatments exist. To get an indication of which treatments have different means, construct confidence intervals for each mean. If any two intervals don't overlap, then the two corresponding means are different.

Confidence Intervals are calculated as follows:

mean - t(s.e)   to   mean + t(s.e)
where t is the t-value associated with the residual degrees of freedom and correspondind to 95% confidence.

If any two confidence intervals overlap slightly, then the treatments will still have significantly different means. To assess whether two individual treatment means are different, you can use a t-test calculated as follows:

t = (mean1 - mean2) / SED
where SED is the standard error of difference associated with any pair of means.

line
Basic statistics in Splus   23.2.99  
                      Page: 21 of 28.