mantelhaenzel(A)

Computes the Mantel-Haenszel Odds Ratio Estimate test using data in array A and a significance level of 0.95 (95%). Array A must be an array of contingency tables 2x2.

mantelhaenszel(A, alpha)

As above but the significance level is alpha.


See also breslowday chi2_mantel 


Example


The following data was collected:


Smoker
Heart attack
Total
Yes
No
Alcohol
7
16
23
No alcohol
22
44
66
Total
29
60
89


Non smoker
Heart attack
Total
Yes
No
Alcohol
63
36
99
No alcohol
7
4
11
Total
70
40
110

Entering the first data in a matrix we obtain:


a1=(7,16;22,44)

/  7 16 \

\ 22 44 /


Using function chi2_totable to transform matrix a1 into a proper contingency table we get:


a1=chi2_totable(a1)

/ "" "" "" "" \

| ""  7 16 23 |

| "" 22 44 66 |

\ "" 29 60 89 /


The second table is:


a2=(63, 36; 7, 4)

/ 63 36 \

\  7  4 /


Using function chi2_totable again we get:


a2=chi2_totable(a2)

/ "" "" ""  "" \

| "" 63 36  99 |

| ""  7  4  11 |

\ "" 70 40 110 /


We now put the two 2x2 contingency tables into an array:


A=(a1, a2)

/ / "" "" "" "" \ / "" "" ""  "" \ \

| | ""  7 16 23 | | "" 63 36  99 | |

| | "" 22 44 66 | | ""  7  4  11 | |

\ \ "" 29 60 89 / \ "" 70 40 110 / /


Using function mantelhaenszel we get:


mantelhaenszel(A)


Mantel-Haenszel Odds Ratio Estimate.

Significance level : 0.95 (95.0%)

Results:

OR: 0.9208477791587625

ln(OR) : -0.08246053417306863

Std.Error of ln(OR) : 0.4100965292305143

Asymp.Sig (2-sided): 0.8406392149103488

Confidence interval for OR

Lower bound : 0.4122048180260054

Upper bound : 2.0571342092564486

Confidence interval for ln(OR)

Lower bound : -0.8862349220302721

Upper bound : 0.7213138536841347