# completey randomized ANOVA

• Jun 14th 2006, 11:23 AM
aptiva
completey randomized ANOVA
Forty eight subjects were randomly assigned to one of 4 treatment conditions in a study of different methods of coping with statistics exam anxiety. After the experiment, 12 subjects remained in group 1, 8 in group 2, 10 in group 3, and 10 in group 4. The mean anxiety score of group 1 was found to be 5.5 points higher than the mean of group 2, 3.5 points higher than the mean of group 3, 5.3 points higher than the mean of group 4, and 3.3 points higher than the grand mean. Interestingly, the sum of squares error was exactly 12.0 times larger than its degrees of freedom.

Did the different treatments result in significantly different amounts of anxiety? (á.01)

ANS: F = 5.83, Reject Ho
__________________________________________________ ______________

I am trying to fill in values using an ANOVA table to determine my F-ratio, but how do I find things like degrees of freedom and sum of squares error from what I am given in the problem?
• Jun 17th 2006, 11:18 PM
CaptainBlack
Quote:

Originally Posted by aptiva
Forty eight subjects were randomly assigned to one of 4 treatment conditions in a study of different methods of coping with statistics exam anxiety. After the experiment, 12 subjects remained in group 1, 8 in group 2, 10 in group 3, and 10 in group 4. The mean anxiety score of group 1 was found to be 5.5 points higher than the mean of group 2, 3.5 points higher than the mean of group 3, 5.3 points higher than the mean of group 4, and 3.3 points higher than the grand mean. Interestingly, the sum of squares error was exactly 12.0 times larger than its degrees of freedom.

Did the different treatments result in significantly different amounts of anxiety? (á.01)

ANS: F = 5.83, Reject Ho
__________________________________________________ ______________

I am trying to fill in values using an ANOVA table to determine my F-ratio, but how do I find things like degrees of freedom and sum of squares error from what I am given in the problem?

The number of degrees of freedom of the Total Sum of Square Errors is
the number of participants minus one.

(rough rule of thumb: number of degrees of freedom for a statistic = number of data
values - number of derived statistics used. In this case there is one derived
statistic used to calculate the Total Sum of Square Errors :- the grand mean).

So you have:

$SSE_T=12(40-1)=468$

RonL