MGMT524HandoutR. L. ANDREWS

Chapter 10 procedures for two groups, 6/14/06

Parameters / Sample type / Section / Excel / Tools / Data Analysis
Means / Independent / I.
II.
III. / z-test Two Sample for Means
t-test Two-Sample Assuming Equal Variances
t-test Two-Sample Assuming Unequal Variances
Means / Paired / IV. / t-test Paired Two Sample for Means
Proportions / Independent / V. / (No Excel component for this)
Proportions / Paired / not covered
Variances / Independent / VI. / F-test Two Sample for Variances
Variances / Paired / not covered

I.Section 10.1, pg 312. Two phenomena with unknown means, 1 and 2, and the standard deviations,1 and 2 have known values. Two independent samplesn1and n2have been selected from the respective phenomena for inference about the difference in the unknown means.

100(1-)% Confidence Interval for 1 - 2:

H0: 1 - 2 = D0 versus Ha: 1 - 2 ( < oror >) D0

Test Statistic for the test is .

Use the standard normal distribution for p-values or critical values.

The table valueZ1-/2 can be found withNORMSINV(1-/2).

Excel Data Analysis procedure: z-test Two Sample for Means.

II.Section 10.1, pg 313. Two phenomena with unknown means, 1 and 2, and unknown standard deviations, 1 and 2, but it is assumed that 1 = 2= . Two independent samplesn1and n2have been selected from the respective phenomena for inference about the difference in the unknown means. From the sample data the estimator of the common variance 2is

and it is used to find

100(1-)% Confidence Interval for 1 - 2:

The table value t1-/2withdf = n1+n2-2can be found with TINV(,n1+n2-2).

H0: 1 - 2= D0 versus Ha: 1 - 2 ( < oror >) D0

Test Statistic for the test is .

Use the t distribution with n1+n2-2 degrees of freedom for p-values or critical values (an acceptable practice is to use the standard normal for p-values if the degrees of freedom is greater than 30).

Note: this test is robust to the assumption of equal variances if n1n2.

Excel Data Analysis procedure: t-test Two-Sample Assuming Equal Variances.

III.Section 10.1, pg 318. Two phenomena with unknown means, 1 and 2, and unknown standard deviations, 1 and 2, 2 independent samplesn1and n2, inference about the difference in the unknown means. For this section the degrees of freedom are defined by a complicated formula and can be calculated by Excel or a statistical package. Use the t distribution with package calculated df.

The 100(1-)% Confidence Interval for 1 - 2 is

df = Round, where .

H0: 1 - 2= D0 versus Ha: 1 - 2 ( < oror >) D0 The Test Statistic is

Excel procedure: t-test Two-Sample Assuming Unequal Variances.

IV.Section 10.2, pg 322.

Two phenomena with unknown means, 1 and 2, data are gathered bypaired sampling.

A difference is calculated for each of the n pairs of observations. The mean of the sample differences is denoted and the sample standard deviation of the differences is .

100(1-)% Confidence Interval for 1 - 2:

H0: 1 - 2= D0 versus Ha: 1 - 2 ( < oror >) D0 The Test Statistic is

Use the t distribution with n- 1 degrees of freedom for p-values or critical values. (An acceptable practice is to use the standard normal for p-values if the degrees of freedom is greater than 30.)

Excel Data Analysis procedure: t-test Paired Two Sample for Means.

V.Section 10.3, pg 332. Two phenomena with unknown proportions, 1 and 2. Two independent samplesn1and n2have been selected from the respective phenomena and the sample proportions p1and p2are to be used for inference about the difference in the unknown proportions. Both n1and n2should be 50 or more and preferably n1*p15 & n1*(1-p1)5 & n2*p25 & n2*(1-p2)5.

100(1-)% Confidence Interval for 1 - 2:

H0: 1 - 2= 0 versus Ha: 1 - 2 ( < oror >) 0.

First use all the sample data to find a pooled proportion, p, where

is the Test Statistic for the test.

Use the standard normal distribution for p-values or critical values.

There is no Excel procedure.

VI.Section 10.4, pg 338. Two normally distributed phenomena with variances, and . Two independent samplesn1and n2have been selected from the respective phenomena for inference about the ratio of the unknown variances. 100(1-)% Confidence Interval for :

Lower Limit = Upper Limit = .

H0: = 1 versus Ha: ( < oror >) 1

Test Statistic for the test is . Use the f distribution with n1-1 numerator degrees of freedom and n2-1 denominator degrees of freedom for p-values or critical values.

Excel Data Analysis procedure: F-test Two Sample for Variances.