Healthy Immigrant Children: A Demographic and Geographic Analysis - October 1998

3. Methodological Issues

Previous

Contents

Next

Identifying immigrant and non-immigrant children for analytical purposes in contrast to legal purposes is not a straightforward process. Conceptually, there are 3 groups of children to distinguish: those children born outside of Canada of non-Canadian parents who can be unambiguously identified as immigrant children; those children born in Canada where at least one parent is an immigrant whom we identify as non-immigrant children of immigrant parents; and a third group of children born in Canada whose parents are non-immigrants and these we identify as non-immigrant children of non-immigrant parents . Our study focuses on the first two groups, identified in the American sociological literature as the "1.25" and "1.5" generations.

Various forms of parental organization add to the complexity of distinguishing immigrant children from non-immigrant children. We make the following distinctions:

1. Immigrant Children
   1. Two parents, both immigrants
   2. Lone parent, immigrant
2. Non-Immigrant Children of Immigrant Parents
   1. Two parents, both immigrants
   2. Lone parent, immigrant
   3. Two parents, one an immigrant or immigrant parent, one other adult
3. Non-Immigrant Children of Non-Immigrant Parents
   1. Two parents or two adults, non-immigrant
   2. Lone parent, non-immigrant

This report uses two methods of analysis, cross-tabulations and logistic regression. The results of the cross-tabulations are presented in tabular form where the above categories are used to distinguish immigrant children, non-immigrant children of immigrant parents and non-immigrant children of non-immigrant parents. In the logistic regression models, we differentiate immigrant children born outside of Canada from those born in Canada. In other words, the contrast is between Category A - Immigrant Children in one group and Category B - Non-Immigrant Children of Immigrant Parents and Category C - Non-Immigrant Children of Non-Immigrant Parents combined into a second group.

For both the cross-tabulations and the logistic regression models, the weighting procedures recommended by Statistics Canada are used, but this does not obviate the fact that the total sample size for the NLSCY Cycle 1 is 22,831 and that the number of children who fall into Categories A and B is very small (see below) relative to the total sample size. In the case of the tabular data, the Statistics Canada procedures for identifying large coefficients of variation are applied, identifying estimates which need to be treated with great caution. Where the number of observations in a cell falls below 10, the results are suppressed completely.

Table 1: The Distribution of Category A Immigrant Children in NLSCY by Age, Sex and Years Since Immigration

Age	Years Since Immigration
	0 to 4	5 to 9	10+
Females
0	3	0	0
1	5	0	0
2	9	0	0
3	10	0	0
4	9	0	0
5	8	1	0
6	9	4	0
7	11	12	0
8	11	18	0
9	14	14	0
10	7	12	1
11	5	8	4
Total	101	69	5
Males
0	2	0	0
1	3	9	0
2	7	0	0
3	11	0	0
4	12	1	0
5	10	4	0
6	13	6	0
7	8	7	0
8	4	10	0
9	8	12	1
10	12	18	3
11	9	19	3
Total	99	86	7
Source: NLSCY 1994

Table 1 summarizes the distribution of Category A - Immigrant Children by age, sex and years since immigration. Most of the children have arrived in Canada in the past nine years. Only 367 children fall into Category A and they make up only 1.6 percent of all children in the NLSCY sample. Given that approximately 17 percent of the total Canadian population was born outside Canada, immigrant children are clearly under-represented in the NLSCY sample. Table 2 provides a similar summary of the distribution of Category B -Immigrant Children with the additional category of family organization provided to capture the complexity of this group. The total number of children who fall into Category B is 2,735, 12.0 percent of the NLSCY. This leaves 19,729, or 86.4 percent of children in the NLSCY sample, who fall into Category C - Non-Immigrant Children of Non-Immigrant Parents.

Table 2: The Distribution of Category B Non-Immigrant Children of Immigrant Parents in NLSCY by Age, Sex, Years Since Immigration and Family Organization

Age	Single Parent Years since Immigration			Two Parents/Both Immigrants Years since Immigration			Two Parent/One Immigrant or Immigrant Parent/One Other Adult Years since Immigration
	0 to 4	5 to 9	10+	0 to 4	5 to 9	10+	0 to 4	5 to 9	10+
Females
0	0	3	3	31	12	13	10	14	58
1	4	1	8	23	16	15	10	9	64
2	1	2	3	7	11	16	5	14	47
3	1	2	4	4	14	9	4	9	62
4	0	1	6	1	19	21	0	7	55
5	0	5	12	3	4	20	0	9	58
6	1	1	8	2	4	25	1	7	64
7	0	2	4	0	6	23	1	5	55
8	0	0	9	1	3	29	2	3	50
9	1	0	7	0	1	22	0	1	59
10	0	0	11	0	0	31	0	1	56
11	0	1	4	0	0	24	0	0	62
Total	8	18	79	72	90	248	33	79	690
Males
0	1	2	6	26	17	13	14	8	62
1	1	0	8	28	31	16	7	22	73
2	2	1	5	15	15	10	4	12	57
3	0	2	12	4	24	12	2	13	66
4	0	2	8	1	17	14	2	15	61
5	0	1	10	1	18	20	0	3	53
6	0	2	8	0	13	19	2	7	60
7	0	1	8	3	3	24	0	3	59
8	1	1	5	2	3	26	0	3	61
9	0	0	6	0	0	31	0	0	66
10	0	0	9	1	0	27	1	1	49
11	0	0	4	0	2	24	1	0	65
Total	5	12	89	81	143	236	33	87	732
Source: NLSCY 1994

Three logistic regressions models are presented for each question analyzed:

1. The Basic Model contains the basic characteristics of the child and the person most knowledgeable (PMK) about the child or the household as independent variables;
2. The Ethnicity Model uses the ancestor question for the child, added to the list of independent variables;
3. The Community Model drops the ancestor question for the child and adds a set of independent variables to measure various socio-economic characteristics of the adult population of the places where children live.

Table 3 summarizes the independent variables used in the three models. In the case of categorical variables, the reference category is defined in the right-hand column. The remaining variables are all treated as continuous in the logistic regression models.

All of the logistic regression models have overall fits which are statistically significant at p < 0.0001. The focus of the discussion, therefore, is on the interpretation of the parameter estimates and their corresponding odds ratios. For ease of presentation, the odds ratios are discussed in terms of their percent effect on the dependent variable. They are rounded off to the nearest percent.

For the Basic and Ethnicity Models, the odds ratios represent how much the likelihood of the dependent variable changes based on a unit change in the coefficient of the independent variable. For the Community Models, we re-calculate the odds ratio based on a 10 percent change in the coefficient of the independent variable, applying the formula exp^r*10, where exp is the exponent and r is the parameter estimate.

Table 3: Independent Variables Used in Logistic Regression Models

Independent Variables	Definition	Reference Category
AGE	Age of child
FEMALE	Sex of child	Male
EUROPE	Region of birth of PMK	Canada
ASIA	Region of birth of PMK	Canada
OTHER	Region of birth of PMK	Canada
NOENGFR	PMK speaks neither official language	PMK speaks at least one official language
IMMGRAN	Immigrant child as identified under Category A	All other children
SPAR	Only PMK present and children	All other parental arrangements
METRO	Urban places with populations > 500,000	Urban places with populations < 100,000
MEDIUM	Urban places with populations between 100,000 and 500,000	Urban places with populations < 100,000
RURAL	Rural places	Urban places with populations < 100,000
FRENCH	Ancestor question for child	All other categories
GERMAN	Ancestor question for child	All other categories
ITALIAN	Ancestor question for child	All other categories
CHINESE	Ancestor question for child	All other categories
POLISH	Ancestor question for child	All other categories
PORTUG	Ancestor question for child	All other categories
SASIAN	Ancestor question for child	All other categories
BLACK	Ancestor question for child	All other categories
NAMIND	Ancestor question for child	All other categories
PCT65	Percent of the population aged 65 and over
TOTIMM	Percent of the adult population who are immigrants
RECIM	Percent of the adult population who immigrated between 1988 and 1991
LOWED	Percent of the population aged 15 and over with less than grade 9 education
UNI	Percent of the population aged 15 and over with a university education
UNEMP25	Unemployment rate for the population age 25 and over
GOVTRAN	Percent of neighbourhood income from government transfers
MEDINCF	Median income for census families
NEIPROB	Derived neighbourhood problem index
Source: NLSCY 1994 PMK = Person most knowledgeable

Previous

Contents

Next