Assessing the Extent of Randomization Bias in the Canadian Self-Sufficiency Experiment