the same year, Dr. Wolkoff found that aid per enrolled minority student was $3,615, compared with $3,767 per enrolled nonminority student. In other words, as measured by students enrolled in public schools [187 [187 Misc.2d Misc.2d 107] 107] State-wide, Dr. Wolkoff found that minority students received $152 less than nonminority pupils. The differences between enrolled and attending students reflected a lower attendance rate by minority students compared with their nonminority peers. Dr. Wolkoff showed similar results for the 1996-1997 school year. Dr. Wolkoff found that these differences in average per pupil expenditures State-wide between minority and nonminority students were too small to be statistically significant. Dr. Wolkoff's per capita analysis fails to account for regional cost differences, which, as discussed above, are highest in New York City. He also made no effort to account for the relative advantages that districts outside of New York City enjoy because of STAR tax relief. Additionally, his analysis does not account for the concentration of minority students in New York City and his minority averages appear to have been driven up by a few high-spending districts. In all, the court is not persuaded by either Dr. Berne's or Dr. Wolkoff's per capita analyses of the distribution of State aid. 2. Dr. Berne's Regression Analyses Dr. Berne's regression analyses are probative evidence of disparate impact and provide independent proof of disparate impact. Dr. Berne conducted a statistical regression analysis for the 1995-1996 and 1996-1997 school years to determine whether minority and nonminority students attending similar districts receive the same level of aid. His analysis used five different pupil "units" used by the State to distribute or analyze the State school finance system. He analyzed operating aid, the largest aid category, in isolation, and he analyzed all aid distributed by computerized formulas, which accounts for approximately 95% of all State aid. Based on his review of the State's justifications for its present funding system, and upon a review of the national literature concerning school finance, he selected a number of independent variables to determine their effect on the distribution of State aid. These variables included: (1) property and income as measured by the combined wealth ratio, (2) effective tax rates, (3) levels of student poverty as measured by the provision of free and reduced-price school lunch, (4) the number of ELL students, (5) district attendance rates, (6) district size, (7) percentage of nonminority students, and (8) the number of students with disabilities (this data was available for 1996-1997 only). [187 [187 Misc.2d Misc.2d 108] 108] Dr. Berne conducted 30 regressions for each of the two years studied. Regardless of the variables, weighting, and student counts applied, Dr. Berne determined that the State aid system had a disparate racial impact on minority students across the State in every regression. Fifty-eight of the 60 regressions were statistically significant. Dr. Berne's regressions tend to show that minority students receive less State aid as their over-all concentration increases in a particular district. Dr. Wolkoff leveled several critiques against Dr. Berne's regression analyses and reworked several of the regressions by adding his own variables. Dr. Wolkoff testified that the failure to include nonformula aid (i.e., the 5% of State aid that is not distributed by formulas) may have skewed Dr. Berne's results. Dr. Wolkoff noted that a number of these nonformula aids funnel money by criteria that often sends the aid to the State's minority public school students. For example, categorical reading aid goes only to the "Big 5" school districts, which of course contain the vast majority of the State's minority students. Dr. Wolkoff, though he replicated many aspects of Dr. Berne's regressions, did not attempt to quantify the effect the inclusion of these categorical aids would have on the regressions. Given the relatively small total dollar amount of categorical aids, and the fact that many do not appear on their face to direct a higher proportion of aid to high-minority districts, this criticism does not undermine the validity of Dr. Berne's regressions. Dr. Wolkoff also correctly points out that Dr. Berne did not account for the fact that the operating aid formula contains a "floor," by which all districts, no matter how wealthy, receive at least $400 per pupil in operating aid. Dr. Berne instead assumed an entirely linear relation between wealth and the distribution of operating aid, where the two were inversely related. Dr. Wolkoff attempted to account for this nonlinearity in his regressions. When he did so he found no statistically significant results that tended to show that nonminority students receive more State aid than minority students. The court notes that this result is highly counter-intuitive. Accounting for the $400 floor that is guaranteed even the richest districts—which generally have low percentages of minority students—would appear to exacerbate, not ameliorate, the funding disparities favoring low-minority districts observed by Dr. Berne. In all events, Dr. Berne adequately justified his decision not to model the $400 floor in his regressions. The regression analyses were not meant to exactly replicate the computer aid [187 [187 Misc.2d Misc.2d 109] 109] formulas run by the State. Minority status is obviously not an explicit factor in the State aid formulas. The point of the regression analyses is to see if minority status of students is, sub rosa, a factor that determines the distribution of State aid. This factor cannot be measured if the funding formulas are simply recreated verbatim. For the reasons stated the evidence demonstrates the existence of a disparate adverse impact on minority students caused by the State's school funding system. With respect to the causation issue, the facts in this case demonstrate that disparate funding of school districts in the State has had a real effect on individual students. The court has already found, in ruling on plaintiffs' claim under the Education Article, that lack of sufficient district funding has had a negative effect on student performance. While defendants demonstrated that there is some inefficiency, waste and corruption in the New York City School District, these factors do not break the chain of causation between disparate funding distributed by the State school finance system and poor student performance.