3. Methods
The National Longitudinal Study of Children and Youth (NLSCY) is managed jointly by Human Resources Development Canada and Statistics Canada. The data for this study are derived from the first cycle of data collection conducted in 1994. Almost 23,000 randomly selected Canadian families of children newborn to eleven years of age were asked extensive information about their sociodemographic, health, family functioning, and educational characteristics. In each household, the person most knowledgeable (PMK) about the child provided responses on behalf of the child and the family.
The questions analyzed in this study were:
- Has your child ever experienced being hungry because the family had run out of food or money to buy food? If yes, How often?
- How do you cope with feeding your child when this happens? Options include the parent or child skipping meals or eating less; and food procurement strategies such as seeking help from a food bank.
Because approximately 25% of NLSCY respondents did not respond to the question on hunger, we chose to analyze the unweighted sample of 206 families who did respond rather than the 233 weighted samples. This is because we wished to be conservative in our estimates (there were many more significant associations between hungry and not hungry families when the weighted sample was used), and because we did not wish to presume whether non-respondents over- or under-represented hungry families. A lower-than-desired response rate, the small size of the sample available for analysis, and the use of an unweighted sample are clearly limitations of this study.
Statistics were only employed for variables for which all hungry families had responses, again to enhance the robustness of the data analysis. Descriptive statistics mainly involved cross-tabulations of frequencies, t-tests, chi-squares, and one-way analyses of variance. In order to reduce the number of spurious results, p<.005 was chosen as the preferred level of significance for comparisons between those ever experiencing hunger compared with those never experiencing hunger. For the sub-analyses of families who ever reported experiencing hunger, the significance level remained at p<.05.
Notwithstanding these significance levels, for all analyses, coefficient of variation release guidelines on the quality of the estimate were used for cells sizes exceeding 30 (Human Resources Development Canada & Statistics Canada, 1996). Cell sizes less than 30 were not reported. Where estimates have a sample size of 30 or more and low coefficients of variation in the range 0 to 16.5%, no warning is required. For estimates where the cell size is 30 or more and the coefficients of variation range from 16.6% to 33.3%, estimates are flagged with the letter M, indicating that readers should use and interpret the results cautiously. If the letter U appears with an estimate, the reader is advised that the data do not meet Statistics Canada's quality standards and conclusions would be unreliable and likely invalid.
Stepwise logistic regression analyses were conducted using the Statistical Package for the Social Sciences (SPSS) to assess the dependent variables of ever experiencing hunger, hunger frequency, and coping strategies, while controlling for independent sociodemographic, economic, and health status variables. Logistic regression models were truncated within ten consecutive variables or when the variables failed to add value to the model. The final models were then re-run with the remaining variables.
What's New
