Less than 1% of the sample had negative income and were
subjected to this recoding. Since households with negative
incomes may be more likely to have credit card debt, they may
experience more benefits from shopping. Thus, we elected to
retain these households in our sample