Statistical Methods for Psychology

those who reported one category of childhood abuse have an odds ratio of 0.163/0.106 5

1.54. Thus the odds of being abused as an adult for someone from the Category 1 group are

1.54 times the odds for someone from the Category 0 group. For the other two groups the

odds ratios relative to the Category 0 group are 2.40 and 6.29. The effect of childhood sexual

abuse becomes even clearer when we plot these results in Figure 6.2. The odds of being

abused increase very noticeably with a more serious history of childhood sexual abuse.

Odds Ratios in 2 32 3 kTables

Just as we can compute an odds ratio for a 2 3 2 table, so also can we compute an odds

ratio when that same study is replicated over several strata such as departments. We will

define the odds ratio for all strata together as

For the Berkeley data we have

OR=

©(n 11 kn 22 k>n..k)

©(n 12 kn 21 k>n..k)

Section 6.11 Effect Sizes 163

Figure 6.2 Odds ratios relative to the non-abused category

Odds Ratios of Adult Abuse

6 5 4 3 2 1

012

Sexual Abuse Category

Odds Ratios Relative to Category = 0

3

Department Data n 11 kn 22 k/n..k n 12 kn 21 k/n..k

B 353 207

17 8

4.827 6.015

C 120 205

202 391

57.712 50.935

D

E

138 279

131 244

53 138

94 299

42.515

27.135

46.148

22.212

(continues)