Analysis of Categorical Data II

The two-dimensional contingency tables, which were the subject of the previous chapter, will have been familiar to most readers. Cross-classifications of more than two categorical variables, however, will not usually have been encountered in an introductory statistics course. It is such tables and their analysis that are the subject of this chapter. Two examples of multidimensional contingency tables, one resulting from cross-classifying three categorical variables and one from cross-classifying five such variables, appear in Tables 10.1 and 10.2. Both tables will be examined in more detail later in this chapter. To begin our account of how to deal with this type of data, we shall look at three-dimensional tables. Table 10.1 Cross-Classification of Method of Suicide by Age and Sex

		Method
Age (Years)	Sex	1	2	3	4	5	6
10–40	Male	398	121	455	155	55	124
41–70	Male	399	82	797	168	51	82
>70	Male	93	6	316	33	26	14
10–40	Female	259	15	95	14	40	38
41–70	Female	450	13	450	26	71	60
>70	Female	154	5	185	7	38	10

Note. Methods are 1, solid or liquid matter; 2, gas; 3, hanging, suffocating, or drowning; 4, guns, knives, or explosives; 5, jumping; 6, other. Table 10.2 Danish Do-It-Yourself

			Accommodation Type
			Apartment			House
Work	Tenure	Answer	Age <30	31–45	46+	<30	31–45	46+
Skilled	Rent	Yes	18	15	6	34	10	2
		No	15	13	9	28	4	6
	Own	Yes	5	3	1	56	56	35
		No	1	1	1	12	21	8
Unskilled	Rent	Yes	17	10	15	29	3	7
		No	34	17	19	44	13	16
	Own	Yes	2	0	3	23	52	49
		No	3	2	0	9	31	51
Office	Rent	Yes	30	23	21	22	13	11
		No	25	19	40	25	16	12
	Own	Yes	8	5	1	54	191	102
		No	4	2	2	19	76	61

Note. These data arise from asking employed people whether, in the preceding year, they had carried out work on their home that they would previously have employed a craftsman to do; answers, yes or no, are cross-classified against the other four variables, which are age, accommodation type, tenure, and type of work of respondent.