为了高手理解我的意思,我具体列出程序,请高手指点!程序如下:
为了研究某种化合物的致癌性,将60只大鼠随机分为4组,分别接受100mg、200mg、300mg和400mg剂量的该化合物 饲料喂养,随访观察1年以后观察其肿瘤发生情况,实验数据见下表,欲考察肿瘤的发生率是否随着剂量水平的变化而呈线性趋势。
60只大鼠肿瘤发生情况
化合物剂量 例数
肿瘤发生情况 未发生 发生 合计
100mg 12 3 15
200mg 10 5 15
300mg 9 6 15
400mg 5 10 15
合计 36 24 60
检验上述数据显性趋势的程序如下
data prg15_1;
do dose=1 to 4;
do response=0 to 1;
input f@@;
output;
end;
end;
cards;
12 3
10 5
9 6
5 10
;
run;
proc freq data=prg15_1;
tables dose*response/trend norow nocol nopercent
scores = table;
weight f;
exact trend;
ods html;
run;
运行上述结果如下,显示结果P(0.0132)<0.05,
Statistics for Table of dose by response
Cochran-Armitage Trend Test
----------------------------
Statistic (Z) -2.5927
Asymptotic Test
One-sided Pr < Z 0.0048
Two-sided Pr > |Z| 0.0095
Exact Test
One-sided Pr <= Z 0.0066
Two-sided Pr >= |Z| 0.0132
而当用普通的卡方检验,如下
data ex7_1;
do dose=1 to 4;
do response=1 to 2;
input f@@;
output;
end;
end;
cards;
12 3
10 5
9 6
5 10
;
run;
proc freq;
weight f;
tables dose*response
/chisq
expected;
run;
结果如下,P(0.06514216730615)>0.05,这是怎么回事,请高手赐教!!
Statistics for Table of dose by response
Statistic DF Value Prob
----------------------------------------------------------------
Chi-Square 3 7.2222 0.06514216730615
Likelihood Ratio Chi-Square 3 7.3681 0.06104542043560
Mantel-Haenszel Chi-Square 1 6.6102 0.01013971166655
Phi Coefficient 0.3469
Contingency Coefficient 0.3278
Cramer's V 0.3469