Sei sulla pagina 1di 1

BN2102 Tutorial Week 5.

ANOVA

Analysis of data of 4 training protocols


We are trying to evaluate the efficacy of 4 different training protocols for the Da Vinci robot.
The protocols are identified by G1, G2, G3 and SUT. Forty medical students were divided in 4
groups, each group undergoing one of the four different training protocols on the simulator. The
file GEARS_scores.dat contains data on the GEARS scores of an in vivo robotic surgery on a
porcine model conducted after the training sessions. The file has 4 columns, each representing
the scores of a different groups of students: the first column contains the score of the students
who underwent protocol G1, the second contains the scores of students who underwent protocol
G2, the third contains the scores of students who underwent protocol G3 and the fourth the
scores of students who underwent protocol named SUT.
Use ANOVA to determine whether there is any statistically significant difference in the mean
scores of the different protocols. To do so, the following step-by-step procedure is recommended:

ˆ Import the data file in Matlab and extract 4 arrays from the data matrix, each corre-
sponding to one training protocol.

ˆ Compute mean and variance for each of the 4 groups.

ˆ Compute the key quantities s2with and s2bet and obtain the F ratio.

ˆ Using a 95% confidence level, verify that F > Fcrit .

ˆ Using the Bonferroni correction, perform 6 pairwise Student t tests among the groups
and determine where the difference in the means originated from.

Further analysis on 3 new samples


The first analysis was unable to confirm any difference among the G1, G2 and G3 protocols.
After some time, a new version of protocol G3 is proposed. You decide to conduct another
study to check whether you can detect any difference among the 3 protocols. In this study,
you recruit 135 medical students and divide them in 3 groups of 45 each. The GEARS scores
are collected for each student, after the training protocol, when performing an in vivo robotic
surgery on a mouse. The number of subjects in this study is quite high. When all of them
are done, it is late in the evening and you have the file GEARS_scores_2.dat containing the
GEARS scores to analyse. Check the ANOVA assumptions before performing the analysis.

Potrebbero piacerti anche