S¼ 20 ; 911 15 ; 922
¼ 4989
sS¼ð 175 Þð 303 Þ
ð 3 Þð 478 Þð 477 Þð 4783 313 473 1093 1783 1133 Þ1 = 2
¼ 2794 : 02
leading to
z¼4989
27 : 94 : 02
¼ 1 : 79
which shows a significant association between the mother’s age and preterm
delivery (one-sidedpvalue¼0.0367); the younger the mother, the more likely
the preterm delivery.
6.8 NOTES ON COMPUTATIONS
Samples of SAS program instructions were provided at the end of Examples 6.8
and 6.14 for complicated procedures. Other computations can be implemented
easily using a calculatorprovidedthat data have been summarized in the form
of a two-way table. Read Section 1.4 on how to use Excel’sPivotTableproce-
dure to form a 22 table from a raw data file. After the value of a statistic
has been obtain, you can use NORMDIST (Section 3.6) to obtain an exactp
value associated with azscore and CHIDIST to obtain an exactpvalue asso-
ciated with a chi-square statistic; the CHIDIST procedure can be used simi-
lar to the case of a one-tailed TDIST (Section 3.6). For the method of Section
6.5, and instead of writing a SAS program, one can calculate expected fre-
quencies (usingformulaanddrag and fill, Section 1.4), then input them into the
CHITEST procedure of Excel.
EXERCISES
6.1 Consider a sample ofn¼110 women drawn randomly from the mem-
bership list of the National Organization for Women (N.O.W.),x¼ 25
of whom were found to smoke. Use the result of this sample to test
whether the rate found is significantlydi¤erentfrom the U.S. proportion
of 0.30 for women.
6.2 In a case–control study, 317 patients su¤ering from endometrial carci-
noma were matched individually with 317 other cancer patients in a
hospital and the use of estrogen in the six months prior to diagnosis was234 COMPARISON OF POPULATION PROPORTIONS