An Assessment of Cuyahoga’s
Fair Housing Act: 1980-2000
Dr. William Bowen
Mukesh Kumar
William Olson
Levin College of Urban Affairs at
Cleveland State University
UAA 2003
Fair Housing Activities
• Prohibits discrimination on basic of “race or
color, national origin, religion, sex, family
status, and/or handicap
• Realtors, landlords, private homeowners
prohibited from screening, steering,
redlining, blockbusting, or otherwise
impeding free choices of individuals in
search of residence
UAA 2003
Outcome Interpretation
• Do various indicators of segregation suggest that
there has been change in the degree to which
different racial groups live separately from one
another in different parts of the urban environment
• Indicators: evenness (dissimilarity index, exposure
(isolation index), concentration (density index),
centralization (distance from CBD)
• Do not consider that under a regime of completely
free and unconstrained choice, many individuals
would prefer to live by other individuals in their
same social category
UAA 2003
Process Interpretation
• Are individuals, as a consequence of Fair Housing
activities, less constrained in the process of
deliberating alternative residential locations?
• Any two individuals with identical budgets, albeit
from different social categories, should face an
identical choice set of alternative feasible
residences.
• The interest is as much in what does not happen as
what does happen.
UAA 2003
The Residential Choice Process
Present Deliberations Alternative Future
of Home Seeker Residences Outcomes
|---A1 Yk|A1
|---A2 Yj|A2
|--- ………… ……
X ? |--- ………… ……
|--- ………… ……
|---An Yl|Aa
UAA 2003
Newcomb’s Problem of Choice
• Story concerning a being who has shown himself
capable on a million occasions in the past, of
predicting human choices with perfect accuracy.
• When clearly specified, it leads to a contradiction,
and so by reductio ad absorbum, the story is
impossible.
• The distinguishing characteristic of free choice is
its inherent unpredictability. If one can predict it,
then it is not free.
UAA 2003
Can We Find Evidence of
Constraints?
• Problem is one of estimating relations
between two sets of components of
residential segregation.
A. Determined (constrained) component
B. Free (unconstrained) component
Changes over time in the ratio of magnitude of
determined component to free component
indicates change in level of free choice.
UAA 2003
Linear Models
• Total sum of squares = Explained sum of
squares (deterministic) + Residual sum of
squares (free)
• Residual variation = f (sample &
measurement errors, model
misspecification, and inherently
unpredictable events)
UAA 2003
Summary of Outcome Measures
• Data from 1980, 1990, 2000 Census (tract level, Cuyahoga
County)
– Collected variables in:
• Social Area Analysis
• Factorial Ecology literature
Table 1.
Summary of Segregation Indices
Year
Index 1980 1990 2000
Gini (evenness) 0.9638 0.9499 0.9127
Entropy (evenness) 0.7391 0.7079 0.6154
Interaction (exposure) 0.1699 0.1752 0.2097
Isolation (exposure) 0.8213 0.8117 0.7557
Delta (concentration) 0.8384 0.7959 0.7353
RCE (centralization) 0.3762 0.4147 0.4388
Dissimilarity (evenness) 0.8677 0.8396 0.7825
UAA 2003
Percentage Blac k
0 - 20
1980 20 - 40
40 - 60
60 - 80
80 - 100
N
W E
S
Resid ential C hoice a nd Fa ir Ho using
UAA Co nference
Percentage Blac k
0 - 20
1990 20 - 40
40 - 60
60 - 80
80 - 100
N
W E
S
Resid ential C hoice a nd Fa ir Ho using
UAA Co nference
Percentage Blac k
0 - 20
2000 20 - 40
40 - 60
60 - 80
80 - 100
N
W E
S
Resid ential C hoice a nd Fa ir Ho using
UAA Co nference
Summary of Process Measures
Table 3.
Ratio of Explained Sum of Squares to Residual Sum of Squares
Models Equation 1980 1990 2000
All Variables 10 6.36 5.79 4.86
All Factors 11 2.81 2.75 2.36
Social Area Analysis 12 2.44 2.38 2.61
Economic Variables 13 2.22 2.01 1.83
Family Status Variables 14 0.9 0.95 1.08
Housing Variables 15 0.15 0.25 0.34
UAA 2003
Table 2
Summary of Regressions (Extent of Free Choice)
Adjusted R-squared
Model Regressions 1980 1990 2000
(10) All Variables 0.862 0.856 0.821
(11) All Factors 0.753 0.697 0.681
(12) Social Area Analysis 0.731 0.698 0.699
(13) Economic Factors 0.72 0.679 0.621
(14) Demographic Factors 0.385 0.436 0.45
(15) Housing Factors 0.152 0.221 0.283
Table 3
One-Sample Kolmogorov-Smirnov Test
Kolmogorov-Smirnov Z Asymp. Sig. (2-tailed)
Model Regressions 1980 1990 2000 1980 1990 2000
[10] All Variables 0.865 0.845 0.994 0.443 0.473 0.277
[11] All Factors 1.368 0.915 0.905 0.047 0.372 0.386
[12] Social Area Analysis 1.148 0.903 0.959 0.143 0.388 0.317
[13] Economic Factors 1.112 1.278 1.891 0.168 0.076 0.002
[14] Demographic Factors 2.063 2.491 1.629 0.000 0.000 0.010
[15] Housing Factors 3.475 2.813 2.118 0.000 0.000 0.000
Table 4
One-Sample Kolmogorov-Smirnov Test
1980 1990 2000
Housing
Model 16 Adj. R-squared* 0.855 0.848 0.732
Model 17 Adj. R-squared** 0.028 0.016 0.051
K-S Significance 0.678 0.620 0.974
Family Status
Model 18 Adj. R-squared* 0.784 0.743 0.750
Model 19 Adj. R-squared** 0.180 0.174 0.093
K-S Significance 0.634 0.862 0.870
Economic
Model 20 Adj. R-squared* 0.744 0.741 0.705
Model 21 Adj. R-squared** 0.274 0.281 0.306
K-S Significance 0.410 0.462 0.596
* Dep var. is % African American. Indep. Var.s include all but the category
** Dep var. is Residual from above model. Ind var.s include the category
Histogram Normal P-P Plot of Regression Standardiz
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Resi
50 1.00
40
.75
30
Expected Cum Prob
.50
20
10 Std. Dev = .98 .25
Mean = 0.00
0 N = 337.00
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
3.
-.5
.5
.5
.0
.5
.0
00
00
50
00
50
00
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Housing Observed Cum Prob
1980,1990
Histogram Normal P-P Plot of Regression Standard
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Res
40 1.00
30 .75
20
Expected Cum Prob
.50
10
Std. Dev = .98 .25
Mean = 0.00
0 N = 337.00
-3
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
-.5
.5
.0
.5
.0
.5
.0
00
00
50
00
50
0
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Observed Cum Prob
Histogram Normal P-P Plot of Regression Standardize
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Resid
50 1.00
40
.75
30
Expected Cum Prob
.50
20
10 Std. Dev = .98 .25
Mean = 0.00
0 N = 337.00
-3
-3
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
-.5
.5
.5
.0
.5
.0
.5
.0
00
00
50
00
50
0
0
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Demography Observed Cum Prob
1980,2000
Histogram Normal P-P Plot of Regression Standardiz
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Resi
50 1.00
40
.75
30
Expected Cum Prob
.50
20
10 Std. Dev = .98 .25
Mean = 0.00
0 N = 337.00
-3
-3
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
-.5
.5
.5
.0
.5
.0
.5
.0
00
00
50
00
50
0
0
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Observed Cum Prob
Histogram Normal P-P Plot of Regression Standardize
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Resid
50 1.00
40
.75
30
Expected Cum Prob
.50
20
10 Std. Dev = .97 .25
Mean = 0.00
0 N = 337.00
-3
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
3.
-.5
.5
.0
.5
.0
.5
.0
00
00
50
00
50
00
0
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Economic Observed Cum Prob
1990,2000
Histogram Normal P-P Plot of Regression Standardize
Dependent Variable: Unstandardized Residual Dependent Variable: Unstandardized Residu
50 1.00
40
.75
30
Expected Cum Prob
.50
20
10 Std. Dev = .97 .25
Mean = 0.00
0 N = 337.00
-3
-2
-2
-1
-1
0.00
0.
1.
1.
2.
2.
3.
3.
-.5 0
.5
.0
.5
.0
.5
.0
00
00
50
00
50
00
50
0
0
0
0
0
0
0.00 .25 .50 .75 1.00
Regression Standardized Residual Observed Cum Prob
Conclusions
• Based on Outcome Measures, Fair Housing
activities have had some (slight) effect
• Based on Process Measures, the results are
mixed
A) African-Americans have less economic free
choice
B) The social/family and housing structure
constraints are less severe…
UAA 2003