{\rtf1\ansi \deflang1033\deff0{\fonttbl
{\f0\fnil \fcharset0 \fprq2 Courier 10cpi;}{\f1\fnil \fcharset0 \fprq2 Courier New;}{\f2\fnil \fcharset0 \fprq2 Courier 12cpi;}
{\f3\fnil \fcharset0 \fprq2 Times New Roman;}}{\colortbl
\red0\green0\blue0;}
{\stylesheet{\fs20 \snext0 Normal;}
{\*\cs1 \additive footnote ref;}
{\*\cs2 \additive\f1 NORMAL INDEN;}
{\*\cs3 \additive\f1 HEADING 1;}
{\*\cs4 \additive\f1 HEADING 2;}
{\*\cs5 \additive\f1 HEADING 3;}
{\*\cs6 \additive\f1 HEADING 4;}
{\*\cs7 \additive\f1 HEADING 5;}
{\*\cs8 \additive\f1 HEADING 6;}
{\*\cs9 \additive\f1 HEADING 7;}
{\*\cs10 \additive\f1 HEADING 8;}
{\*\cs11 \additive\f1 HEADING 9;}
{\*\cs12 \additive\f1 FOOTNOTE TEX;}
{\*\cs13 \additive\f1 FOOTNOTE REF;}
{\*\cs14 \additive\f1 HEADER;}
{\*\cs15 \additive\f1 FOOTER;}
{\*\cs16 \additive\f3 1;}
{\*\cs17 \additive _Equation Ca;}
{\*\cs18 \additive caption;}
{\*\cs19 \additive toa heading;}
{\*\cs20 \additive index 2;}
{\*\cs21 \additive index 1;}
{\*\cs22 \additive toc 9;}
{\*\cs23 \additive toc 8;}
{\*\cs24 \additive toc 7;}
{\*\cs25 \additive toc 6;}
{\*\cs26 \additive toc 5;}
{\*\cs27 \additive toc 4;}
{\*\cs28 \additive toc 3;}
{\*\cs29 \additive toc 2;}
{\*\cs30 \additive toc 1;}
{\*\cs31 \additive footnote tex;}
{\*\cs32 \additive endnote refe;}
{\*\cs33 \additive endnote text;}
{\*\cs34 \additive Default Para;}
}\margl1440\margr1296\enddoc\ftnrestart\aftnnar \sectd \sbknone\endnhere 
{\*\pnseclvl1\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl2\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl3\pnlcrm\pnstart1{\pntxta .}}
{\*\pnseclvl4\pndec\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl5\pnlcltr\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl6\pnlcrm\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl7\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl8\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl9\pnlcrm\pnstart1}

\pard \qc\sl0 
{\plain \f1 REPRESENTATIVE BUREAUCRACY AND DISTRIBUTIONAL EQUITY:\par
}\pard \sl0 
{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 ADDRESSING THE HARD QUESTION\par
}\pard \sl0 
{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 Kenneth J. Meier, Texas A&M University\par
}\pard \sl0 
{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 Robert D. Wrinkle, University of Texas Pan-American\par
}\pard \sl0 
{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 J. L. Polinard, University of Texas-Pan American\par
}\pard \sl0 
{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 All data and documentation to replicate this analysis are 
available from the authors.  Address correspondence to: K.J. 
Meier, Dept. of Political Science, Texas A&M University, College 
Station, TX 77843, email kmeier@polisci.tamu.edu  \par
}{\plain \f1 \par
}{\plain \f1 }\pard\page 
\pard \qc\sl0 
{\plain \f1 Representative Bureaucracy and Distributional Equity:\par
}\pard \sl0 
{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 Addressing the Hard Question\par
}\pard \sl0 
{\plain \f1 \par
}{\plain \f1 \par
}\pard \qc\sl0 
{\plain \f1 Abstract\par
}\pard \sl0 
{\plain \f1 \par
}{\plain \f1 \par
}\pard \sl480 
{\plain \f1 \tab Research on representative bureaucracy has failed to deal 
with whether or not representative bureaucracies produce minority 
gains at the expense of nonminorities.  Using a pooled time series 
analysis of 350 school districts over six years, this study 
examines the relationship between representative bureaucracy and 
organizational outputs for minorities and nonminorities.  Far from 
finding that representative bureaucracy produces minority gains at 
the expense of nonminorities, this study finds both minority and 
nonminority students perform better in the presence of a 
representative bureaucracy.  This finding suggests an alternative 
hypothesis to guide research, that representative bureaucracies 
are more effective than their nonrepresentative counterparts.\par
}{\plain \f1 \par
}{\plain \f1 }\pard\page 
\pard \qc\sl480 
{\plain \f1 }{\plain \b\f1 Representative Bureaucracy and Distributional Equity:\par
}\pard \qc\sl480 
{\plain \b\f1 Addressing the Hard Question}{\plain \f1 \par
}\sect \sectd \sbknone\pgnrestart\pgndec\pgnx720\pgny720\endnhere 
{\*\pnseclvl1\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl2\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl3\pnlcrm\pnstart1{\pntxta .}}
{\*\pnseclvl4\pndec\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl5\pnlcltr\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl6\pnlcrm\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl7\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl8\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl9\pnlcrm\pnstart1}

\pard \sl480 
{\plain \f1 \par
}{\plain \f1 \tab Despite initial skepticism that a bureaucracy widely 
representative of the people would make policy decision more 
responsive to the public (Meier and Nigro 1976), recent studies 
for both blacks and Latinos have found that under certain 
conditions passive racial and ethnic representation can produce 
active representation, that is policies that benefit minorities 
(Meier and Stewart 1991; Hindera 1993a; 1993b; Meier 1993; Seldon, 
Brudney and Kellough 1998).  Such findings in widely differing 
bureaucracies from urban school districts to EEOC regional offices 
to Farmer's Home Administration loan programs generate some 
optimism simply because public bureaucracies are slowly becoming 
more representative in terms of race and ethnicity.  At the same 
time, policy outcomes are often perceived as a zero-sum game, the 
gains of one group must be compensated for with losses from 
another.{\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \super \chftn }{}{\plain .  Politics is, after all, the determination of who gets what, 
when, and how (Lasswell 1936).  Politics thus implies conflict and 
redistribution.  Despite the economics notion that there are 
Pareto optimal policy changes, the day-to-day conflict in the 
policy process suggests otherwise.   }}}
}{\plain \super\f1 }{\plain \super\f1 }{\plain \f1   The fervor of the current affirmative action debate, 
for example, reflects the sincere beliefs by some whites that such 
policies take opportunities away from individuals simply because 
they are white (Eastland 1996).  This study seeks to move the 
literature on representative bureaucracy to a direct consideration 
of such redistributional consequences. Given the inter-racial and 
inter-ethnic distributional concerns, empirical work on this 
question is desperately needed.\par
}\sect \sectd \sbknone\pgndec\pgnx6156\pgny15120 
{\*\pnseclvl1\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl2\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl3\pnlcrm\pnstart1{\pntxta .}}
{\*\pnseclvl4\pndec\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl5\pnlcltr\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl6\pnlcrm\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl7\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl8\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl9\pnlcrm\pnstart1}

\pard \sl480 
{\plain \f1 \tab First, a brief review of representative bureaucracy theory is 
presented.  The theory has evolved from a rather simple view of 
bureaucratic motivation to a fairly involved description of how 
individuals' values affect decisions in complex organizations.  
Second, we present a model based on this review and operationalize 
it using data from school districts in Texas.  Because our concern 
is redistributional consequences, we estimate the impact of 
representative bureaucracy on both minority and Anglo students.  
These equity concerns are best illustrated through traditional 
regression analysis and a relatively recent analytical technique, 
substantively weighted least squares (SWLS).  Third, using the 
results of this model, we suggest an alternative specification of 
how representative bureaucracies might influence public policy and 
a new hypothesis about representative bureaucracies.\par
}\pard \qc\sl480 
{\plain \f1 }{\plain \b\f1 The Theory of Representative Bureaucracy}{\plain \f1 \par
}\pard \sl480 
{\plain \f1 \tab Until ten years ago, the theory of representative bureaucracy 
was relatively simple.{\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \super \chftn }{}{\plain .  Several individuals have sketched out this theory, perhaps the 
first being Norton Long (1952).  A recent presentation is Seldon 
et al. (1998).  }}}
}{\plain \super\f1 }{\plain \super\f1 }{\plain \f1   In every organization individuals who 
make decisions exercise discretion because organization rules 
cannot cover every contingency and because organizational 
socialization is rarely total (Downs 1967; Thompson 1967).  If 
individuals are assumed to be utility maximizers, then individual 
bureaucrats with discretion are likely to use that discretion to 
make decisions that reflect their own values.  One source of these 
values is the socialization process, and one of the most enduring 
relationships is the impact on race and ethnicity on values 
(Carmines and Stimson 1989).  Representative bureaucracy, thus, 
suggests that if a bureaucracy is broadly representative of the 
public it serves, then it is more likely to make decisions that 
benefit that public (Thieleman and Stewart 1996).  \par
}{\plain \f1 \tab The initial work examining a variety of value sources has 
been replaced by work focusing on values related to race.  Not all 
policy decisions are likely candidates for influence by a 
representative bureaucracy.  Bureaucrats must have discretion over 
decisions that are directly linked to race or ethnicity.  Meier 
(1993) contends that representation is also enhanced by political 
support and that, in some cases, a critical mass is necessary.  
Seldon et al. (1998) introduce the concept of a representative 
role that mediates between one's background and one's policy 
decisions.  \par
}{\plain \f1 \tab Quite clearly the work of Seldon et al. (1998), Hindera 
(1993a;1993b), Meier (1993) and Meier and Stewart (1991) 
demonstrate that minority access to positions in the bureaucracy 
influences policies in such a way as to benefit minority 
clientele.  What they do not address are the distributional 
consequences.  Are the results of representative bureaucracy 
achieved by taking benefits from some other group of clientele?  
Or might it be the case that representative bureaucracies are also 
more effective and thus minority gains are not at the expense of 
nonminority clientele? \par
}\pard \qc\sl480 
{\plain \f1 }{\plain \b\f1 An Operational Model of Representative Bureaucracy}{\plain \f1 \par
}\pard \sl480 
{\plain \f1 \tab Educational institutions are ideal organizations to examine 
questions of representative bureaucracy.  School systems are the 
nation's largest public employer; the modal bureaucrat is more 
likely to grade papers in Des Moines than shuffle them in 
Washington.  Schools employ numerous professionals who are only 
loosely supervised; and, thus, discretion permeates the 
organization.  School systems also generate a great deal of data 
so that the redistributional results of a representative 
bureaucracy can be assessed.\par
}{\plain \f1 \tab Our model of representative bureaucracy sets up what is 
called an education production function whereby student 
performance is a function of inputs to the organization and 
various policies.  To this production function, we add measures of 
representative bureaucracy to determine if it matters given 
controls for other factors that affect performance.  Because our 
concern is redistributional consequences, we run these models for 
both minority students and Anglo students.\par
}{\plain \f1 \tab Our units of analysis are 350 Texas school districts with at 
least 1000 students.  To make sure that the districts are 
performing relatively similar functions, we further limit the 
analysis to districts with more than 10 percent but no more than 
90 percent Anglo students, that is, multiracial districts.  Data 
are available for six years from 1991 through 1996 so we pool 
these data resulting in a total of 2097 usable cases with no 
missing data.  Pooled models are frequently affected by problems 
of serial correlation and heteroscedasticity.  To control for 
serial correlation, a set of five dummy variables representing 
individual years is included in the model.{\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \super \chftn }{}{\plain .  Serial correlation was a significant problem.  After a large 
drop in scores for these districts in 1992, they generally 
followed an upward trend.  The use of dummy variables to correct 
for such problems is the traditional strategy for shallow pools 
such as this one (Stimson 1985).  }}}
}{\plain \super\f1 }{\plain \super\f1 }{\plain \f1   Heteroscedasticity 
was assessed using the White test; corrections for the modest 
heteroscedasticity found did not alter the findings so the 
ordinary least squares results will be presented.\par
}{\plain \b\f1 Dependent Variable: Student Performance}{\plain \f1 \par
}{\plain \f1 \tab The state of Texas requires students in grades 3, 5, 8, and 
10 to take standardized tests every year.  The percentage of 
students who pass these tests in each school district is the 
dependent variable.  Our measures include pass rates for all 
students, pass rates for minority students (blacks and Latinos), 
and pass rates for Anglos.\par
}{\plain \b\f1 Independent Variables: Controls}{\plain \f1 \par
}{\plain \f1 \tab To ensure that any relationships found between representation 
and student test scores are not spurious, a variety of other 
factors that influence education performance are included in the 
model.  Because our interest is in testing aspects of the theory 
of representative bureaucracy, our discussion of these variables 
is brief.  All are culled from the education literature and are 
frequently used in education production functions. Production 
functions typically include measures of environmental constraints, 
resources applied to the process, and district policies designed 
to improve performance.{\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \super \chftn }{}{\plain .  This literature is far too large to cite comprehensively.  See 
the extended bibliography in Burtless (1996).  }}}
}{\plain \super\f1 }{\plain \super\f1 }{\plain \f1   \par
}{\plain \f1 \tab In the context of educational policy, poverty is a serious 
constraint on student performance.  Poverty not only means 
students lack access to learning tools in the home (computers, 
educational toys, etc.) but is also correlated with a less stable 
and less supportive home environment (e.g., single parent 
households, high rates of teen pregnancy, and low educational 
expectations; Necochea and Cune 1996; Fuller et. al. 1996).  Our 
measure of poverty is the percent of students who qualify for free 
or reduced-price meals in the school lunch program.\pard \sl480 
}{\plain \*\cs1 }{\plain \*\cs1\super {\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \*\cs1\super \chftn }{}{\plain \*\cs1 .  Some of the other measures might also tap dimensions of socio-economic status.  Expenditures, because they are in part 
determined by the local tax base, are an example of an indirect 
measure of economic well being.  }}}
}{\plain \*\cs1\f1 }{\plain \*\cs1\f1   The 
relationship to performance should be negative.\par
}{\plain \*\cs1\f1 \tab The relationship between expenditures and educational 
outcomes is one of the most contested relationships in educational 
policy.  Examining a wealth of studies, Hanushek (1986; 1989; 
1996) contends that there is no consistent relationship between 
money and student outcomes.  Although this finding has been 
challenged by others (Hedges and Greenwald 1996), it remains the 
conventional wisdom.  In recent longitudinal studies, however, 
Murray (1995), Evans, Murray and Schwab (1997), and Murray, Evans 
and Schwab (1995) found that districts that increased expenditures 
had improved performance afterward.\par
}{\plain \*\cs1\f1 \tab Three "expenditure" variables are included in the analysis--per pupil expenditures for instruction, average teacher salary, 
and percentage of money from state funds.  Per pupil expenditures 
for instruction are used in preference to total per pupil spending 
because many Texas districts spend lavishly on extracurricular 
activities.  Our concern is academic performance, so the spending 
measure should be based on classroom instruction.  Education is 
personnel intensive, and most spending pays salaries of teachers 
and other staff.  Higher salaries are perceived in economic theory 
as a way to attract better qualified persons to a profession 
(Hanushek and Pace 1995).  Finally, state aid can be used to 
compensate for inequities in local tax bases.  Although Texas is 
not known for redistributive educational policies and has a long 
history in court on this issue (}{\plain \*\cs1\i\f1 San Antonio Independent School 
District v. Rodriquez}{\plain \*\cs1\f1 , 1973; }{\plain \*\cs1\i\f1 Edgewood Independent School District 
v. Kirby}{\plain \*\cs1\f1 , 1987; See also Texas Research League 1986; Accountable 
Cost Advisory Committee 1986; Weiher 1988), greater funds from 
state governments can compensate for a meager local tax base.  All 
relationships should be positive.\par
}{\plain \*\cs1\f1 \tab Education policies are adopted to influence student 
performance.  Two such policies deal with the learning 
environment--class size and gifted classes.  }{\plain \f1 }{\plain \super\f1 }{\plain \f1 Although many studies 
indicate that only major changes in class size are effective, 
schools with smaller class sizes should have an advantage at the 
margin (see Pate-Bain et al. 1992; Nye et al. 1992; Hedges and 
Greenwald 1996; Hanushek 1996, 54).  The measure is the number of 
students per teacher in the district.  Gifted classes are 
generally conceded to be the best education that a school system 
offers (See DeHaan 1963).  The number of students enrolled in 
gifted classes varies greatly across these districts (0 to 31%), 
and greater access should result in better performance.  Gifted 
classes should be positively related to performance, and class 
size should be negatively related. \par
}{\plain \f1 \tab Teachers are a crucial element in a student's educational 
environment.  As a profession based on life-long learning, there 
should be some advantage to teachers with adequate experience, 
especially in multiracial districts.  Our measure is the average 
years of teacher experience which should be positively related to 
student performance.  \par
}{\plain \f1 \tab One factor we cannot control for is the innate abilities of 
the students, especially those abilities not correlated with 
poverty.  A common strategy in such circumstances is to use exam 
scores by the same cohort of students for earlier grades (see 
Smith and Meier 1995).  Unfortunately, the state of Texas only 
reported pass rates for individual grades for three of these years 
and never reported grade level data by race.  While this suggests 
the models are somewhat underspecified, the levels of explained 
variation compare favorably to other models that do use a cohort 
control (Chubb and Moe 1990; Smith and Meier 1995).\par
}{\plain \b\f1 Independent Variables: Representation}{\plain \f1 \par
}{\plain \f1 \tab Bureaucratic representation is measured at the street level 
(Lipsky 1980)--the percentage of black and Latino teachers.  
Although intra-minority redistributional consequences are 
important in this area (Meier and Stewart 1991), in the interests 
of parsimony we limit our analysis to minority versus majority 
comparisons.  This emphasis makes the combined percentage of both 
black and Latino teachers the appropriate measure.  Representative 
bureaucracy hypothesizes that presence of minority teachers will 
improve the performance of minority students. The education 
literature suggests three ways minority teachers can do this: (1) 
minority teachers are more effective at teaching minority students 
(Moore and Johnson 1983, 472; Aaron and Powell 1982, 55); (2) 
minority teachers serve as role models for minority students (Cole 
1986, 332); and (3) minority teachers mitigate the negative 
consequences of grouping, tracking, and discipline (Meier and 
Stewart 1991).\par
}{\plain \f1 \tab If minority teachers improve the education performance of 
minority students, then their impact on majority students should 
also be assessed.  If minority teachers spend more time with 
minority students or teach in a different manner, then Anglo 
students could be worse off.  The redistributional concerns of 
representative bureaucracy can be assessed directly by looking at 
the impact of minority teachers on Anglo students.\par
}\pard \qc\sl480 
{\plain \f1 }{\plain \b\f1 Findings}{\plain \f1 \par
}\pard \sl480 
{\plain \b\f1 Minority Representation and Distributional Equity }{\plain \f1 \par
}{\plain \f1 \tab Table 1 presents the production function model for all 
students, minority students, and Anglo students.  Although the 
control variables are not our concern here, that they all are 
generally in the predicted direction and significant is reassuring 
in terms of model specification.  The only anomaly is for teacher 
experience with positive coefficients for all students and Anglo 
students and a negative correlation for minority students.\par
}\pard \qc\sl480 
{\plain \f1 [Table 1 About Here]\par
}\pard \sl480 
{\plain \f1 \tab The relationship of concern is the slope for minority 
teachers.  In the all students regression, a one percentage point 
increase in minority teachers is associated with a drop of .1113 
points in the all student pass rate.  That relationship along with 
the positive relationship for minority students--a one percentage 
point increase in minority teachers is associated with a .0614 
percentage point increase in minority student pass rates--implies 
that there are redistributional tradeoffs, that minority students' 
gains come at the expense of nonminority students.  Examining the 
Anglo regression, however, challenges this conclusion.  A one 
percentage point increase in minority teachers is associated with 
a .0730 percentage point }{\plain \ul\f1 increase}{\plain \f1  in the Anglo pass rate.  This 
finding is strong support for the lack of redistributional 
consequences.  In fact, a representative bureaucracy appears to 
benefit Anglos slightly more than it benefits minorities.\par
}{\plain \f1 \tab The puzzling finding for all students suggests that greater 
attention needs to be focused on the composition of the all 
student pass rates.  With 98.6 percent of students falling into 
the categories of black, Latino or Anglo, the "other" students 
cannot account for this finding.  Somewhat surprisingly, if one 
regresses Anglo and minority pass rates on the overall pass rate, 
one can account for }{\plain \ul\f1 only}{\plain \f1  77 percent of the variance (regression 
not shown).  \par
}{\plain \f1 \tab The residuals from this regression, however, clarify the 
situation.  Table 2 compares the districts with positive residuals 
to those with negative residuals.  Clearly the difference between 
these two sets of districts cannot be found in the superior 
performance of any subset of students.  Although the positive 
residual districts have a pass rate of 61 for all students 
compared to 49.3 for the negative residuals, the pass rates for 
Anglos and the pass rates for minorities are virtually identical 
in both sets of districts.  What is different about the two sets 
of districts is their racial composition.  The positive residuals 
have 73.4 percent Anglo students compared to 33.4 for the negative 
residual districts.  So while each group of students performs the 
same, differences in the district's overall racial composition 
distinguish between the two sets of districts.  The relationship 
for all students, therefore, is a statistical artifact reflecting 
the unequal racial distribution of students across these 
districts.\par
}\pard \qc\sl480 
{\plain \f1 [Table 2 About Here]\par
}\pard \sl480 
{\plain \f1 }{\plain \b\f1 Representative Bureaucracy as Nonlinear}{\plain \f1 \par
}{\plain \f1 \tab The residual pattern along with the fact that the positive 
residual districts have far fewer minority teachers (6.3% versus 
29.0%) suggests that the impact of representative bureaucracy 
might be nonlinear.  Testing a hypothesis presented by Thompson 
(1982), for example, Meier (1993) found a nonlinear relationship 
between school administrators and student performance suggesting a 
critical mass was needed to have an impact.\par
}{\plain \f1 \tab Table 3 presents the results of a nonlinear estimation of 
representative bureaucracy (this table omits the coefficients for 
the other variables).  These results support Thompson's argument.  
For all students, both the linear and squared minority teachers' 
terms are statistically significant.  At low levels of minority 
representation, the impact on students is negative; but at higher 
levels it becomes positive.  Taking the first derivative of these 
terms and setting them equal to zero predicts that the impact of 
minority teachers on all students will become positive at 
approximately 32.3% minority teachers, a somewhat higher critical 
mass than Meier (1993) found in his nonlinear analysis. \par
}\pard \qc\sl480 
{\plain \f1 [Table 3 About Here]\par
}\pard \sl480 
{\plain \f1 \tab The equation in table 3 for minority students shows the same 
nonlinear pattern, negative at low levels and positive as a 
critical mass is reached.  The threshold for a positive impact on 
minority students is much lower than that for all students, 
approximately 21.8% minority teachers.  For Anglo students the 
relationship is linear; including the squared term adds no 
additional explanation to the equation.  These findings, along 
with the lower levels of minority teachers in the "all students" 
dominated set of districts, strongly implies that there are no 
redistributional consequences of representative bureaucracy in 
these organizations.  Anglo students actually do better in 
bureaucracies that are more representative of minorities; so 
minority student benefits do not come at the expense of 
nonminority students.\par
}{\plain \f1 \tab The size of impacts for minority teachers are not large, but 
then we should not expect them to be.  Environmental factors have 
a dominant influence on school systems and student performance.  
Not every minority student is exposed to a minority teacher.  Not 
all districts will exploit fully the talents of their teachers, 
including minority teachers.  The difference in results for the 
all student rates and the minority and Anglo pass rates suggests 
that additional efforts to discriminate among these different 
bureaucracies are needed.\par
}{\plain \b\f1 A Substantively Weighted Analysis}{\plain \f1 \par
}{\plain \f1 \tab  In a recent methodological development, Meier and Keiser 
(1996) argue that relationships between variables can vary across 
agencies.  This position implies that ordinary least squares (OLS) 
analysis will not provide all the vital information in a policy-relevant situation.  Because OLS seeks to generalize to the 
average case, it can miss those agencies that do better than 
average or any other policy relevant sets of agencies.   \par
}{\plain \f1 \tab The basic process of substantively weighted analytical 
techniques (SWAT) is that cases of interest are weighted more 
heavily than cases of less interest.  One form of SWAT, 
substantively weighted least squares (SWLS), starts with an OLS 
regression and then designates a set of cases that are selectively 
downweighted in a series of regressions until the cases of 
interest are weighted as 1.0 and those not of interest are 
weighted .05 (Meier and Keiser 1996).\par
}{\plain \f1 \tab Because our concern is tradeoffs, organizations that produce 
more equitable results are those of interest.  To measure equity, 
we divide the minority pass rate by that for Anglo students.  This 
measure has a mean of .588 and a standard deviation of .117 (in 
the average district the odds of a minority student passing the 
exam are .588 the odds of an Anglo student).  Our interest is in 
those districts that produce more equitable results so we select 
the top ten percent of districts on this scale (those above .734).  
We then run a series of regressions weighting the equity districts 
at 1.0 and decreasing the weight of the other districts in 
increments of .05 until a final regression with weights of 1.0 and 
.05.\par
}{\plain \f1 \tab Another way to address the equity question is to simply use a 
dummy variable interaction between equity districts and the 
variable for minority teachers.  SWLS has two advantages over this 
approach.  First, it does not induce the large amounts of 
collinearity that an interaction does.  Second, SWLS permits all 
slopes to vary, thus making the variable of interest compete 
against all other variables for explanatory power.  The dummy 
variable approach produced similar conclusions to SWLS for the 
linear specification, but suffered from too much collinearity to 
provide good nonlinear estimates.{\super \chftn {\footnote\ftnalt \pard \sa240\sl480 
{\plain \super \chftn }{}{\plain .  The tolerances for minority teachers, minority teachers 
squared and the two interaction terms were all below .15.  The 
tolerance for minority teachers fell below .1.  }}}
}{\plain \f1 \par
}{\plain \f1 \tab The final SWLS and dummy variable regression results appear 
in table 4.  The first column presents SWLS results for a linear 
specification.  Comparing these coefficients with the findings in 
table 1 provides information on what these equity districts do 
different from what the nonequity districts do.  For the all 
student pass rate the relationship remains negative but }{\plain \ul\f1 it ceases 
to be statistically significant}{\plain \f1 .  Equally important changes occur 
in the representation coefficients for the minority and Anglo pass 
rates.  The minority coefficient increases from .0614 to .0968, 
approximately a 58 percent increase.  The Anglo coefficient jumps 
dramatically from .073 to .187, an increase of 156 percent.  For 
those districts more concerned with equity, these relationships 
suggest that both minority and Anglo students benefit more than 
their cohorts in districts not concerned with equity.  In such 
districts, even at low levels of representation there are no 
inter-racial redistributional consequences.  The comparable dummy 
variable findings show similar results for minorities and Anglos 
(columns 3 and 4).\par
}\pard \qc\sl480 
{\plain \f1 [Table 4 About Here]\par
}\pard \sl480 
{\plain \f1 \tab The second column of table 4 reports the SWLS nonlinear 
specifications for the equity equation.  Again the relationship 
for Anglos remains linear.  For the minority pass rate the 
relationships are approximately the same as they are for OLS [See 
Table 3] suggesting that if the linkage is nonlinear it is the 
same in the equity districts as the other districts.  For the all 
student pass rates, the coefficients are about half the size for 
the Anglo districts.  While the nonlinear specification has slight 
gains, they are rather modest so the conservative conclusion is 
the relationship between minority representation and student 
performance is linear for those districts highly concerned with 
equity and that this relationship is much stronger than the 
relationship for all districts. \par
}{\plain \f1 \tab A visual depiction of the SWLS analysis for representative 
bureaucracy--that is, minority teachers--is found in Figure 1.  
This figure shows how the slopes change at different weights, as 
nonequity districts are downweighted.  The impact of minority 
teachers on student pass rates changes dramatically for Anglo 
students, rising approximately 2.5 times in the final weighted 
regression.  The impact on minority student pass rates also 
increases, although not as sharply, to almost 1.5 times the 
beginning slope.  The change in the impact of minority teachers 
for the all students pass rate falls to zero as the nonequity 
districts are downweighted.  These findings, again, suggest that, 
}{\plain \ul\f1 in districts with greater equity, minority teachers are associated 
with increased student performance for both Anglo and minority 
students}{\plain \f1 .\par
}\pard \qc\sl480 
{\plain \f1 [Figure 1 about here]\par
}\pard \sl480 
{\plain \f1 \tab }{\plain \b\f1 Conclusion}{\plain \f1 \par
}\pard \sl480 
{\plain \f1 \tab Representative bureaucracy is offered as a partial solution 
for reconciling the conflict between democracy and administration.  
Given inter-racial and inter-ethnic political tensions, empirical 
work on the redistributional consequences of representative 
bureaucracy is badly needed.  This study of educational outcomes 
provides traditional OLS analyses along with the use of SWLS in a 
partial attempt to meet that need.  SWLS allows an investigation 
into the differences between classes of cases based on performance 
(Meier and Keiser 1996). In this study the performance measure 
involves street-level bureaucrats (minority teachers) in equitable 
and non-equitable districts. The findings of both the SWLS and the 
OLS suggest that moving toward a more representative bureaucracy 
will not have outcomes detrimental to the established majority. 
Instead of minority students' gains coming at the expense of Anglo 
students, both groups benefit from higher levels of minority 
representation of street level bureaucrats--teachers--in the 
education system.  There appear to be no redistributional 
consequences.\par
}{\plain \f1 \tab This paper is the first study to explore redistributional 
consequences of representative bureaucracy using SWLS.  It should 
not be the last.  A virtual wealth of appropriate data sets are 
now available for investigation using this technique.  The 
opportunity to identify the impact of changes in the bureaucracy 
in several settings and over time would determine whether 
distributional changes come at a cost to established interests.\par
}{\plain \f1 \tab School districts, similar to all organizations, vary widely 
in how they take resources [inputs] and translate them into 
educated students [outputs].  On a substantive level, this study 
has established that equity districts translate inputs, 
particularly minority teachers, into outputs different from how 
nonequity districts do. The result is two specific findings.  
First, these representative bureaucracies do }{\plain \ul\f1 not}{\plain \f1  benefit one group 
of students at the expense of another.  Second, organizations vary 
so that even if future work does find redistributional 
consequences in standard OLS analysis, that analysis must be 
extended in a SWAT framework to determine if }{\plain \ul\f1 all}{\plain \f1  bureaucracies 
produce the same patterns.\par
}{\plain \f1 \tab The lack of redistributional consequences which results 
because all groups of students are better off suggests a 
provocative new hypothesis for organizational studies: 
}{\plain \ul\f1 Representative bureaucracies are more effective at meeting their 
goals than nonrepresentative bureaucracies in similar 
circumstances}{\plain \f1 .  The underlying logic of this hypothesis rests on 
the notion that discriminatory personnel policies result in less 
able employees which, in turn, detrimentally affect agency 
performance.  Representative bureaucracies do not erect such 
artificial barriers to organizational performance and thus will 
perform at a higher level than those with discriminatory hiring 
practices.  The provocative nature of this hypothesis suggests 
that representative bureaucracies are to be preferred on both 
normative and empirical grounds.  While this is clearly only a 
single study addressing one policy area, the findings and this 
logic suggest that the linkage between representative bureaucracy 
and performance has sufficient support to make it a credible 
hypothesis in future research.\par
}{\plain \f1 }\pard\page 
\pard \qc\sl480 
{\plain \f1 }{\plain \b\f1 References}{\plain \f1 \par
}\pard \fi-720\li720\sl480 
{\plain \f1 Aaron, Robert and Glen Powell.  1982.  "Feedback Practices as a 
Function of Teacher and Pupil Race during Reading Group 
Instruction." }{\plain \i\f1 Journal of Negro Education}{\plain \f1  51 (Winter): 50-59.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Accountable Cost Advisory Committee. 1986. "Accountable Cost Study 
and Recommendations of the Accountable Cost Advisory 
Committee to the State Board of Education." Austin, TX: Texas 
Education Agency.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Burtless, Gary.  1996.  }{\plain \i\f1 Does Money Matter? The Effect of School 
Resources on Student Achievement and Adult Success.}{\plain \f1   
Washington: Brookings.  \par
}\pard \fi-720\li720\sl480 
{\plain \f1 Carmines, Edward G. and James Stimson. 1989. }{\plain \i\f1 Issue Evolution: Race 
and the Transformation of American Politics}{\plain \f1 .  Princeton: 
Princeton University Press.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Chubb, John and Terry Moe.  1990.  }{\plain \i\f1 Politics, Markets and America's 
Schools}{\plain \f1 .  Washington: Brookings.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Cole, Beverly P.  1986.  "The Black Educator: An Endangered 
Species." }{\plain \i\f1 Journal of Negro Education}{\plain \f1  55 (Summer): 326-334.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 DeHaan, Robert F. 1963. }{\plain \i\f1 Accelerated Learning Programs}{\plain \f1 . Washington: 
Center for Applied Research in Education, Inc.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Downs, Anthony.  1967.  }{\plain \i\f1 Inside Bureaucracy}{\plain \f1 .  Boston: Little, 
Brown.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Eastland, Terry.  1996.  }{\plain \i\f1 Ending Affirmative Action: The Case for 
Colorblind Justice}{\plain \f1 .  New York: Basic Books.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 }{\plain \i\f1 Edgewood Independent School District v. Kirby}{\plain \f1 . Texas SupCt, No. C-8353, (1989).\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Evans, William N., Sheila E. Murray, and Robert M. Schwab.  1997.  
"Schoolhouses, Courthouses, and Statehouses After }{\plain \i\f1 Serrano}{\plain \f1 ." 
}{\plain \i\f1 Journal of Policy Analysis and Management}{\plain \f1  16 (Winter), 10-31.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Fuller, Bruce, Costanza Eggers-Pierola, Susan D. Holloway, Xiaoyam 
Liang and Marylee F. Rambaud. 1996. "Rich Culture, Poor 
Markets: Why do Latino Parents Forego Preschooling?" }{\plain \i\f1 Teachers 
College Record}{\plain \f1  97 (Spring):400-418.}{\plain \f1 \par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hanushek, Eric A. and Richard R. Pace.  1995.  "Who Chooses to 
Teach (and Why)?"  }{\plain \i\f1 Economics of Education Review}{\plain \f1  14 
(June):107-117.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hanushek, Eric A.  1986.  "The Economics of Schooling: Production 
and Efficiency in Public Schools." }{\plain \i\f1 Journal of Economic 
Literature}{\plain \f1  24 (September):1141-1177.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hanushek, Eric A.  1996.  "School Resources and Student 
Performance." In }{\plain \i\f1 Does Money Matter? The Effect of School 
Resources on Student Achievement and Adult Success,}{\plain \f1  ed. Gary 
Burtless. Washington: Brookings.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hanushek, Eric A. 1989.  "Expenditures, Efficiency, and Equity in  
Education: The Federal Government's Role."  }{\plain \i\f1 American  
Economic Review}{\plain \f1  79 (May):46-51.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hindera, John. 1993a.  "Representative Bureaucracy:  Imprimis 
Evidence of Active Representation in the EEOC District 
Offices."  }{\plain \i\f1 Social Science Quarterly}{\plain \f1  74 (March): 95-108.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hindera, John.  1993b.  "Representative Bureaucracy: Further 
Evidence of Active Representation in the EEOC District 
Offices." }{\plain \i\f1 Journal of Public Administration Research and 
Theory}{\plain \f1  3 (October), 415-429.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Hedges, Larry V. and Rob Greenwald.  1996.  "Have Times Changed?  
The Relation between School Resources and Student 
Performance." In }{\plain \i\f1 Does Money Matter? The Effect of School 
Resources on Student Achievement and Adult Success,}{\plain \f1  ed. Gary 
Burtless. Washington: Brookings.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Lasswell, Harold.  1936.  }{\plain \i\f1 Politics: Who Gets What, When, How?}{\plain \f1   New 
York: McGraw Hill.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Lipsky, Michael. 1980.  }{\plain \i\f1 Street Level Bureaucracy}{\plain \f1 .  New York:  
Russell Sage Foundation.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Long, Norton.  1952.  "Bureaucracy and Constitutionalism." 
}{\plain \i\f1 American Political Science Review}{\plain \f1  46 (September), 808-818.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Meier, Kenneth J. 1993.  "Latinos and Representative Bureaucracy: 
Testing the Thompson and Henderson Hypotheses." }{\plain \i\f1 Journal of 
Public Administration Research and Theory}{\plain \f1  3 (October): 393-415.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Meier, Kenneth J. and Lael R. Keiser.  1996.  "Public 
Administration as a Science of the Artificial: A Methodology 
for Prescription." }{\plain \i\f1 Public Administration Review}{\plain \f1  56 
(September/October): 459-466.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Meier, Kenneth J. and Lloyd G. Nigro.  1976.  "Representative 
Bureaucracy and Policy Preferences." }{\plain \i\f1 Public Administration 
Review}{\plain \f1  36 (July/August), 458-469.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Meier, Kenneth J. and Joseph Stewart, Jr. 1991.  }{\plain \i\f1 The Politics of 
Hispanic Education}{\plain \f1 .  Albany: SUNY Press.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Moore, Helen A. and David R. Johnson.  1983.  "A Reexamination of 
Elementary School Teachers' Expectations: Evidence of Sex and 
Ethnic Segmentation." }{\plain \i\f1 Social Science Quarterly}{\plain \f1  64 
(September): 460-475.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Murray, Sheila E. 1995. "Two Essays on the Distribution of 
Education Resources and Outcomes." PhD. diss. Department of 
Economics, University of Maryland.\par
}\pard \fi-1440\li720\sl480\tx720 
{\plain \b\f1 \tab }{\plain \f1 Murray, Sheila E., William N. Evans and Robert M. Schwab. 1995. 
"Money Matters After All: Evidence From Panel Data on the 
Effects of School Resources." University of Kentucky and 
University of Maryland working paper: The Martin School.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 National Education Goals Panel. 1996. }{\plain \i\f1 The National Education Goals 
Report: Building A Nation of Learners}{\plain \f1 . Washington: U.S. 
Government Printing Office.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Necochea, Juan and Zullmara Cune.  1996.  "A Case Study of Within  
District School Funding Inequities."  }{\plain \i\f1 Equity & Excellence in 
Education}{\plain \f1  29 (September): 69-77.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Nye, Barbara A., Jayne Boyd-Zacharias, B. Dewayne Fulton, and Mark 
P. Wallenhorst. 1992. "Smaller Classes Really are Better." 
}{\plain \i\f1 American School Board Journal }{\plain \f1 179 (May): 31-33.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Pate-Bain, Helen, C.M. Achilles, Jayne Boyd-Zacharias, and Bernard 
McKenna. 1992. "Class Size Does Make a Difference." }{\plain \i\f1 Phi Delta 
Kappan}{\plain \f1  74 (November): 253-56.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 }{\plain \i\f1 San Antonio Independent School District v. Rodriquez}{\plain \f1 . 411 U.S. 1 
(1973).\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Seldon, Sally Coleman, Jeffrey L. Brudney, and J. Edward Kellough.  
1998.  "Bureaucracy as a Representative Institution: Toward A 
Reconciliation of Bureaucratic Government and Democratic 
Theory."  }{\plain \i\f1 American Journal of Political Science}{\plain \f1  42 (July): 
716-.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Smith, Kevin B. and Kenneth J. Meier.  1995.  }{\plain \i\f1 The Case Against 
School Choice}{\plain \f1 .  Armonk, NY: M.E. Sharpe.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Stimson, James.  1985.  "Regression in Time and Space:  A 
Statistical Essay."  }{\plain \i\f1 American Journal of Political Science}{\plain \f1  29 
(November): 914-947.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Texas Research League. 1986. "Bench Marks for 1986-87 School 
District Budgets in Texas." Austin, TX: Texas Research 
League.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Thielemann, Gregory S. and Joseph Stewart, Jr.  1996.  "A Demand 
Side Perspective on the Importance of Representative 
Bureaucracy." }{\plain \i\f1 Public Administration Review}{\plain \f1  56 (March/April): 
168-73.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Thompson, James D.  1967.  }{\plain \i\f1 Organizations in Action}{\plain \f1 .  New York: 
McGraw Hill.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Thompson, Frank J. 1976. "Minority Groups in Public 
Bureaucracies." }{\plain \i\f1 Administration and Society}{\plain \f1  8 (August) :201-226.\par
}\pard \fi-720\li720\sl480 
{\plain \f1 Weiher, Gregory R. 1988. "Why Redistribution Doesn't Work: State 
Educational Reform Policy and Governmental Decentralization 
in Texas." }{\plain \i\f1 American Politics Quarterly}{\plain \f1  16 (April): 193-210.\par
}\pard \sl480 
{\plain \f1 \par
}\pard\page 
\sect \sectd \sbknone\margrsxn360 
{\*\pnseclvl1\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl2\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl3\pnlcrm\pnstart1{\pntxta .}}
{\*\pnseclvl4\pndec\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl5\pnlcltr\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl6\pnlcrm\pnstart1{\pntxtb (}{\pntxta )}}
{\*\pnseclvl7\pndec\pnstart1{\pntxta .}}
{\*\pnseclvl8\pnlcltr\pnstart1{\pntxta .}}
{\*\pnseclvl9\pnlcrm\pnstart1}

\pard \qc 
{\plain \f1 Table 1. Representative Bureaucracy: Its Distributional Aspects\par
}\pard 
{\plain \f1 ______________________________________________________________________________\par
}{\plain \f1 ______________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \ul\f1 Independent Variable      All Students   Minority Students    Anglo 
Students}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 Minority Teachers           -.1113            .0614            .0730\par
}{\plain \f1                             (.0158)          (.0191)          (.0172)\par
}{\plain \f1 \par
}{\plain \f1 Expenditures\par
}{\plain \f1 \par
}{\plain \f1   Instruction Funds K        .3564ns         3.2466           1.8146\par
}{\plain \f1                             (.7933)          (.9568)          (.8656)\par
}{\plain \f1 \par
}{\plain \f1   Teacher Salaries K         .7638           1.4008            .7008\par
}{\plain \f1                             (.1288)          (.1553)          (.1405)\par
}{\plain \f1 \par
}{\plain \f1   State Aid                  .0230            .0639           -.0032ns\par
}{\plain \f1                             (.0091)          (.0109)          (.0099)\par
}{\plain \f1 \par
}{\plain \f1 Policy\par
}{\plain \f1 \par
}{\plain \f1   Gifted Classes             .2827            .1778            .2642\par
}{\plain \f1                             (.0473)          (.0570)          (.0516)\par
}{\plain \f1 \par
}{\plain \f1   Class Size                -.7563           -.4024           -.4868\par
}{\plain \f1                             (.1684)          (.2031)          (.1837)\par
}{\plain \f1 \par
}{\plain \f1 Poverty                    \par
}{\plain \f1 \par
}{\plain \f1   Percent Low Income        -.4090           -.2584           -.1648\par
}{\plain \f1                             (.0140)          (.0169)          (.0153)\par
}{\plain \f1 \par
}{\plain \f1 Teacher Experience           .3693           -.7589            .3633\par
}{\plain \f1                             (.1056)          (.1273)          (.1152)\par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 R-Square                     .71              .54             .54\par
}{\plain \f1 \par
}{\plain \f1 Standard Error              6.47             7.80            7.06\par
}{\plain \f1 \par
}{\plain \f1 F                         383.78           190.13          190.57  \par
}{\plain \f1 \par
}{\plain \f1 N                        2097             2097            2097\par
}{\plain \f1 _________________________________________________________________________\par
}{\plain \f1 All relationships significant at .05 except those marked ns.\par
}{\plain \f1 Numbers in parentheses are standard errors.\par
}{\plain \f1 Dummy variables for individual years are not reported.\par
}{\plain \f1 }\pard\page 
\pard \qc 
{\plain \f1 Table 2. Comparing the Positive Residual Districts with the Negative 
Ones\par
}\pard 
{\plain \f1 ___________________________________________________________________________\par
}{\plain \f1 ___________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \ul\f1 Variable                Negative Residuals      Positive Residuals       
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 Pass Rates\par
}{\plain \f1 \par
}{\plain \f1   All Students                 49.3                    61.0\par
}{\plain \f1 \par
}{\plain \f1   Minority Students            39.5                    40.3\par
}{\plain \f1 \par
}{\plain \f1   Anglo Students               65.8                    67.3\par
}{\plain \f1 \par
}{\plain \f1 Percent of Students\par
}{\plain \f1 \par
}{\plain \f1   Black                        13.6                    11.6\par
}{\plain \f1 \par
}{\plain \f1   Latino                       52.4                    13.6\par
}{\plain \f1 \par
}{\plain \f1   Anglo                        33.4                    73.4\par
}{\plain \f1 \par
}{\plain \f1 _________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \f1 Residuals from Regression of All Student Pass Rate on Minority and Anglo\par
}{\plain \f1 Pass Rates\par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 }\pard\page 
\pard \qc 
{\plain \f1 Table 3. The Nonlinear Impact of Representative Bureaucracy:\par
}\pard 
{\plain \f1 \par
}\pard \qc 
{\plain \f1 Distributional Consequences\par
}\pard 
{\plain \f1 _____________________________________________________________________________\par
}{\plain \f1 _____________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \ul\f1 Independent Variable      All Students   Minority Students    Anglo 
Students}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 Minority Teachers           -.5254           -.3578           -.0816*\par
}{\plain \f1                             (.0372)          (.0453)          (.0419)\par
}{\plain \f1 \par
}{\plain \f1 Teachers Squared             .0081            .0082            .0030\par
}{\plain \f1                             (.0007)          (.0008)          (.0007)\par
}{\plain \f1 \par
}{\plain \f1 Expenditures\par
}{\plain \f1 \par
}{\plain \f1   Instruction Funds K       1.1647ns         4.0646           2.1164\par
}{\plain \f1                             (.7694)          (.9377)          (.8656)\par
}{\plain \f1 \par
}{\plain \f1   Teacher Salaries K         .6993           1.3355            .6767\par
}{\plain \f1                             (.1245)          (.1518)          (.1401)\par
}{\plain \f1 \par
}{\plain \f1   State Aid                  .0163ns          .0572           -.0057ns\par
}{\plain \f1                             (.0088)          (.0107)          (.0099)\par
}{\plain \f1 \par
}{\plain \f1 Policy\par
}{\plain \f1 \par
}{\plain \f1   Gifted Classes             .2516            .1464            .2526\par
}{\plain \f1                             (.0458)          (.0558)          (.0515)\par
}{\plain \f1 \par
}{\plain \f1   Class Size                -.6071           -.2513ns         -.4310\par
}{\plain \f1                             (.1631)          (.1988)          (.1836)\par
}{\plain \f1 \par
}{\plain \f1 Poverty                    \par
}{\plain \f1 \par
}{\plain \f1   Percent Low Income        -.3836           -.2327           -.1554\par
}{\plain \f1                             (.0137)          (.0167)          (.0154)\par
}{\plain \f1 \par
}{\plain \f1 Teacher Experience           .4087           -.7191            .3779\par
}{\plain \f1                             (.1021)          (.1244)          (.1148)\par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 R-Square                     .73              .56             .55\par
}{\plain \f1 \par
}{\plain \f1 Standard Error              6.25             7.62            7.03\par
}{\plain \f1 \par
}{\plain \f1 F                         392.34           192.52          179.44  \par
}{\plain \f1 \par
}{\plain \f1 N                        2097             2097            2097\par
}{\plain \f1 ___________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \f1 * p = .051\par
}{\plain \f1 All relationships significant at .05 except those marked ns.\par
}{\plain \f1 Numbers in parentheses are standard errors.\par
}{\plain \f1 Dummy variables for individual years are not reported.\par
}{\plain \f1 }\pard\page 
\pard \qc 
{\plain \f1 Table 4. Alternative Representation Coefficients\par
}\pard 
{\plain \f1 ____________________________________________________________________________\par
}{\plain \f1 ____________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \f1                           SWLS                  Dummy Variable 
Interaction \par
}{\plain \ul\f1                      Linear    Non Linear       Equity        Nonequity   
 }{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 All Students       -.0114 ns     -.2416         .2291           -.1400 \par
}{\plain \f1                    (.0181)       (.0480)       (.0156)          (.0362)\par
}{\plain \f1 \par
}{\plain \f1 All Students2        ----         .0044\par
}{\plain \f1                                  (.0009)\par
}{\plain \f1 \par
}{\plain \f1 Minority Students   .0968        -.3435         .6189            .0150 
ns\par
}{\plain \f1                    (.0223)       (.0586)       (.0428)          (.0185)\par
}{\plain \f1 \par
}{\plain \f1 Minority Students2   ----         .0085\par
}{\plain \f1                                  (.0011)\par
}{\plain \f1 \par
}{\plain \f1 Anglo Students      .1870         .2028         .1265            .0686\par
}{\plain \f1                    (.0183)       (.0489)       (.0405)          (.0175)\par
}{\plain \f1 \par
}{\plain \f1 Anglo Students2      ----        -.0003 ns\par
}{\plain \f1                                  (.0009)\par
}{\plain \f1 ____________________________________________________________________________\par
}{\plain \f1 \par
}{\plain \f1 SWLS Final Regression Estimates (Weights 1, .05)\par
}{\plain \f1 All coefficients significant except those marked ns.\par
}{\plain \f1 Standard errors in parentheses.\par
}{\plain \f1 \par
}{\plain \f1 \par
}{\plain \f1 }\pard\page 
\pard \qc 
{\plain \f1 }{\plain \b\f1 Notes\par
}\pard 
{\plain \f2 \pard 
}}