摘要:This paper presents a simple bootstrap procedure to develop multivariate confidence intervals for tourist expenditure profiles and consequent estimates of economic impacts per thousand tourist vis-its. Mean expenditures from replicated visitor expenditure data in-cluded weights to correct for response bias. A covariance matrix for means of 50 expenditure items is estimated through 2,000 bootstrap replications for two separate visitation seasons. Confidence intervals assume multivariate normality of the expenditure means, and focus on endpoints defined by proportionate increases (and decreases) from the original sample data means. An empirical example is pr o-vided from summer and winter visitors to the Florida Keys. Ninety -five percent confidence interval endpoints for spending means were found at 3.87 percent above/below the original sample's point esti-mate for winter visitors and at 6.001 percent for summer visitors.