Using a paired t-test instead

2 We could analyse the data using a two-sample t-test 

We could analyse this data using the two-sample /-test. Comparing the weights at the end of the two treatment periods, we would obtain the following result  

The confidence interval includes zero, so the result is non-significant. The P value (0.155) confirms non-significance. 

The difference in variability reflects the fact that some individuals may be much bigger than others, but those who start largest generally end largest and the smallest 

CH12 THE PAIRED t-TEST - COMPARING TWO RELATED SETS OF MEASUREMENTS 

---