Table C.2
Relative contributions of predictor variables to goodness of fit of models that predict speed of closure of trajectories, Canada, 1998 to 2001

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Predictor variable1 Degrees of freedom Self-employed versus employees Public sector versus private sector
Wald chi-square2 Pr (ChiSq)3 Wald chi- square Pr (ChiSq)
Sex 1 0.6 0.42 1.0 0.31
Age group in 1996 2 245.7 <.0001 252.0 <.0001
Moderate wealth in 1996 - non homeowner 1 17.6 <.0001 14.4 0.00
High wealth in 1996 - homeowner 1 13.2 0.00 15.8 <.0001
Wealth increase proxy 1 1.0 0.32 1.2 0.27
Pension eligibility in 1996 1 1.3 0.26 3.0 0.09
Cultural background 3 11.8 0.01 19.5 0.00
Standard work history index in 1996 2 24.1 <.0001 17.0 0.00
Marital status in 1996 3 6.1 0.11 6.2 0.10
Marital status change 1 5.6 0.02 7.5 0.01
Care responsibility change index 1 0.0 0.90 0.1 0.81
Other family retirement income reception 1 14.0 0.00 17.5 <.0001
Health status change 1 4.8 0.03 8.9 0.00
Education in 1996 2 12.4 0.00 9.5 0.01
Occupation group in 1996 3 27.6 <.0001 34.6 <.0001
Self-employed in 1996 1 32.3 <.0001    
Class of worker in 1996 to 1997 4 .. .. 80.6 <.0001
1. All names without a date refer to change over time. The reference date for all change measures is the year before closure of the trajectory began. These are all dummy variables, and an unclosed trajectory yields a value of zero. The value is 1 when the change began in the year just before closure started.
2. The rank ordering of Wald chi-square values is a rough indicator of the relative importance of each varaiable in the contributing to the overall goodness of fit of the model.
3. Statistical significance indicated in the column named "Pr (ChiSq)" is only approximate; because the underlying estimates of standard errors are not fully adjusted for the complexity of the survey design. Here the adjustmentcomprises transforming the respondents' weights so that their average is 1.0 . Tests using more appropriate adjustment via bootstrap computations indicate that when the Wald chi-squares are 6.0 or greater, it can be considered that they are statistically significant at the 5% level or better, in the event that bootstrap standard errors were computed. When the Wald chi-squares shown above are between 3.0 and 6.0, they can be considered to be statistically significant at a level between 15% and 5%. Thus, when the Wald chi-square is less than 3 it should be assumed that the parameter estiamte is seriously subject to variability due to eithersample size or to inter-correlation with other predictor variables in the model.
Source: Statistics Canada, Survey of Labour and Income Dynamics, longitudinal file.
Table source: Statistics Canada, 2008, New Frontiers of Research on Retirement: Technical Annex, catalogue number 75-515-XWE.