To decide whether or not to purchase a new hand-held laser scanner for use in inventorying stock, tests are conducted on the scanner currently in use and on the new scanner. These data are obtained on the number of 7-inch bar codes that can be scanned per second:

------------------------------------------------------------------

new old

n1 = 61 n2 = 61

x1(bar) = 40 x2(bar) = 29

2 2

s = 24.9 s = 22.7

1 2

------------------------------------------------------------------

2 2

#1. Test H0: sigma = sigma at the alpha = 0.2 level to be sure that pooling is appropriate.

1 2

#2. Does the new laser appear to read more bar codes per second on the average ? Explain.

#3. Since the number of bar codes that can be scanned per second is discrete, we have not satisfied the normality requirement. What theorem justifies the use of the pooled T procedure in this case ?