A Little Bit More Evidence of LotteryRegressivity: The Kansas State LotteryPamela Mobilia, Ph.D.Brooklyn College and National Bureau of Economic ResearchThis paper examines the existence of regressivity at the Kansas State Lottery using county level data. The classic "t-test" is used to test if the mean per capita bet in classes, defined as being below or above the median state income, are the same. The results show that the means of the per capita bets are not statistically different. Another test directly tests regressivity using the mean of bets expressed as a percentage of income. The results show that lower income counties bet more as a percentage of income. The second test also defines classes as being below or above the median educational level, labor force participation rate, employment rate, unemployment rate, population density, percentage white and the population size. The results show that Kansas runs a regressive lottery. America's gambling fever has been receiving a great deal of attention as gambling has become more acceptable and readily available in recent years. In fact, some form of gambling is legal and available in every state except for Hawaii and Utah. However, more gambling has been accompanied by more debate. State run lotteries have been championed by legislators as states seek new" sources of revenue. The ultra-high stakes of lotto games have transformed a small change industry into a $16 billion cash cow for financially pressed states (Passell, 1989). Organized religion's opposition (once strong) has been stifled as charitable Monte Carlo Nights and Bingo have become pervasive. Bible belt states have even entered the lottery industry. However, not everyone wants to celebrate or congratulate state lotteries. Gambling has been attacked because of its potential harmful effects on society, the dangers of compulsive wagering, and the erosion of the traditional work ethic, as Americans try to strike it rich with a lucky ticket. Headlines publicizing the gambling-related suspensions of Cincinnati Reds' Pete Rose and New York Yankees' owner George Steinbrenner bring the moral overtones of the activity to public awareness (New York Times, 1989, Vescey, 1990). Recent economic studies on gambling focus on the exploitation of disadvantaged groups by state lotteries, especially on the issue of the regressivity of lotteries (Spiro, 1974; Brinner and Clotfelter, 1975; Suits, 1977; Clotfelter and Cook, 1989). State lotteries extract a fixed percentage of the gross sales for operating expenses and the earmarked purpose of the lottery. This amount determines the price of the lottery ticket and the effective tax rate. This tax rate is called regressive, if when expressed as a percentage of income, it falls as income increases (Clotfelter and Cook, 1989). No matter how you look at the situation, the states' share of lottery receipts amounts to a large tax, and is levied on people who can often ill afford to pay such a tax. Page 5 of the 1988 annual report from the Kansas Lottery claims, "The Kansas Lottery player profile is more than a shade different from the other 28 lottery states. Our typical 'customer' is female--has at least some college or post high school education, and is a member of a $20,000-plus income family!!" This paper examines this statement, and the existence of regressivity at the Kansas State Lottery for the year 1988. This was the first year that the lottery was in operation in Kansas. METHODOLOGYUsing the same techniques as Grossman (1973), the average per capita bet and the bet as a percentage of income are calculated for each county in Kansas. The average of various demographic measures such as income, the median educational level, labor force participation rate, employment rate, unemployment rate, population density, percent of the population that is white, and the size of the population is calculated for each county. The state median for each demographic variable is calculated. The counties are then classified as being in the lower half (below the median) or upper half (above the median) of each demo- graphic variable. The mean value of all the counties below the median and all of the counties above the median is calculated. To test the extremes of the variables, counties are classified as being in the lower three quartiles or in the fourth quartile (upper quartile). Each of the variables is weighted by population. The t test (Newbold, 1984, pp. 365-8) is used to test the hypothesis that the means of each of the classes are the same. Two different values are calculated. One based on the assumption that the variances of the two groups are equal and another on the assumption that variances of the two groups are not equal. An F test is then done to test for the equality of the two variances (Newbold, 1984, pp. 377-8). The pertinent t value is used for the analysis and its significance is determined. For the assumption of unequal variances, the de。