BA 250 Example 2

Hospital LOS
Pollution

Questions:

1) Can we conclude with 95% confidence that the average hospital Length of stay (LOS) is reduced from 5.0 days following administrative changes ?

data for a relatively large (>30) sample of 100 recent patients is provided in CXA08_04.PRN

2) Can we conclude with 95% confidence that the New engine design is giving mean emission levels less than 20 ppm?

data for a small sample of 10 engines is provided in CXA08_05.PRN

1) Data conversion

Data is provided as ASCII files. import the data into EXCEL sheets.

2) Generate descriptive statistics (with 90% CL)

Why would you use a 90% CL in the descriptive statistics procedure even though you want to test at the 95% level

3) Generate a histogram for the LOS data

Make a histogram of the hospital LOS data. (doing this for the small pollution sample might be a little ridiculous)
Does it look like a normal distribution?
Why is it okay to use statistical tests that assume bell-shaped (t and z distributions) curves?

4) Use a t test to test the hypotheses:

State null and Alternative (that's what the "A" stands for in Ha) Use Alpha=0.05 for all these tests, Hypothesized difference is 0

Is it okay to use t instead of Z for the LOS case even though the sample is large? why or why not?

use excel to generate appropriate P values for the observed T's. What do these mean? What do you conclude?

5) Discussion:

Refer to the descriptive statistics, the histograms, and the P values.

Why do we use standard error instead of standard deviation in calculation of t values?

What do the "critical t values" mean?

What are your conclusions?

What are at least three things you could do to improve ability to detect differences between the hypothesized values and reality?