|Canadian Council on Social Development|
441 MacLaren Street
Ottawa, ON K2P 2H3
|Conseil canadien de
Tel: (613) 236-8977 Fax: (613) 236-2750
Income and Child Well-being:
A new perspective on the poverty debate
by David P. Ross and Paul Roberts
Almost all the data used in our 27 charts are drawn from the 1994 cycles of the NLSCY and the NPHS. When this was not the case, it was noted. Responses to these two surveys were in one of three forms. In a few instances, responses were simply a 'Yes' or 'No.' In others, respondents were asked to choose from among several categories; for example, "How often does your child exhibit certain feelings? Never, Seldom, Often, or Always?" In the remainder of cases, responses were presented in the form of scales which combined separate answers to a series of related questions into single scores. For any of the three reporting categories, the responses were then assigned to a child or family, which allowed an association to be made with a family income level.
It is also necessary to clarify what we chose as summary measures to represent the 27 variables that are linked with income. Regardless of the response category, we concentrated on the proportion of children with the "poorest scores" - that is, the scores considered to be the least desirable from the standpoint of good child development - for each outcome or living condition. We then assigned them to one of seven broad income groups. For example, behavioural scores most often follow a broad distribution from top to bottom; however, we have concentrated only on those children (and their families) whose scores fell into the bottom part of the distribution.
How did we select the "poorest" scores? In situations where there was only a yes or no answer, we selected those children whose parents provided a negative response. Where multiple choice or scaled responses created a distribution of scores, we selected a cut-off. Two points about these cut-offs are worth noting. First, they do not necessarily have diagnostic or clinical significance - they simply represent the poorest scores that the sample allows us to focus on. For example, the children we selected as "highly hyperactive" would not necessarily all be diagnosed as hyperactive, but they did have the poorest scores on the hyperactivity scale compared to other children. For the statistical purposes of our report, that is all we need to know, since we are not trying to identify these children for treatment.
Second, the cut-offs vary as a result of the limits imposed by the particular distribution and sample size for each outcome and living condition. We could have selected a constant, and arbitrary, cut-off - such as the bottom third - but we opted instead to get the "poorest" scores possible for each factor. In most cases, scores are drawn from the bottom 10 or 20 per cent of the distribution of all responses. By choosing the poorest outcomes possible, we increase the likelihood that the outcomes or conditions are restrictive and special, and in the case of behavioural outcomes, we increase the likelihood that they more closely approximate a clinical diagnosis.
It is important to note that our report is restricted to an examination of two-parent families. (The authors have provided an extensive examination of child outcomes in lone-parent families elsewhere. ) Lone-parent families were excluded for one simple reason: the vast majority of lone-parent families have low incomes, a fact that would have skewed the results presented here. The graphic analysis would thus have been open to misinterpretation, since movement along the bottom axis would have represented not only income changes, but changes in family type as well, and results at the low end of the income scales would represent not only low-income families, but would disproportionately represent lone-parent families. The reverse would hold at higher income levels. Since the graphs are meant to visually depict the relationship between income (not family type) and child outcomes, it was necessary to "standardize" as much as possible the family types under study. Since over 80 per cent of children live in two-parent families, we chose this family type for our analysis.
It should also be noted that the definition of income used in the charts and accompanying discussion is gross income. Gross income includes earnings, investment income, miscellaneous income (such as child support payments, scholarships, private disability benefits, and so on) and government transfers, before any taxes are deducted. Due to the restricted level of information available on the public release micro-database for the NLSCY, we have had to use income ranges, rather than individual family incomes in our charts. This means that in the charts, all incomes in the $20,000 to $29,999 range, for example, are "centred" at one point. Unfortunately, this gives the impression that the point represents a distinct income level, when in fact it represents that income range. It also means that we frequently obtain rather sharp "breaks" in the lines at these centres, rather than the smooth line that would result from using continuous individual incomes. For simplicity of presentation in the textual discussion, we refer to income ranges by the uppermost dollar figure. Thus, for example, the range $20,000 to $29,999 is variously referred to as income "up to," 'below" or "exceeding" the $30,000 level.
As an interpretive aid to reading the charts, four reference points are provided below: i) median family income; ii) Statistics Canada's LICOs; iii) the results of polling by the Gallup organization; and iv) the Fraser Institute's poverty line.
i) The median income in 1994 for families with children aged 4 to 11 years falls between $50,000 and $60,000. This reference point means that 50 per cent of the families are above this income level, and 50 per cent are below it.
ii) Statistics Canada's LICOs vary for different family sizes and by the size of the community. For a family of four, the 1994 LICOs range from $21,500 for families of four living in rural areas, to $31,000 for families in the largest urban areas where the population exceeds 500,000. If a population-weighted LICO were selected for a family of four (in effect, a national average), it would represent a family living in a large urban community of 100,000 to 499,000 people. This amount would equal $26,600.
iii) Since 1967, the Gallup public opinion polling organization has occasionally asked the following question: "Generally speaking, what do you think is the least amount of money a family of four needs each week to get along in this community?"
In each of the 18 years since this question was first asked, the median response has very closely followed the corresponding LICO (see Chart 28). The question was not asked in 1994, but interpolating between the two nearest years, we estimate a value for 1994 equal to an annual income of $25,200.
iv) The Fraser Institute lines are more idiosyncratic, as they vary geographically according to the local cost of essentials. However, for 1994 they have estimated a national poverty line for a family of four equal to $15,900.
For each of the above poverty measures, as well as the median family income figure, we have selected the value that represents a family of four, since this is the actual median figure for family size represented in the charts, consisting of two adults and two children.
In assessing the importance of this information, the totality of the result is of overriding concern. Each of the 27 factors is interesting and informative in its own right. What is most important, however, is the confirmation of a persistent and pervasive association between family income and child development outcomes and living conditions. All of these results cannot be a statistical accident, nor can the level of family income be considered to be an unimportant factor influencing the opportunities for healthy child development.
The Canadian Council on Social Development (CCSD) is an independent, national, non-profit organization focussing on issues of social and economic security.
[ CCSD Home Page ] [ CCSD Publications ] [ CCSD E-mail ]