1. The Use of Census Data and Spatial Statistical Tools in GIS to Identify Economically Distressed Areas
Presented by:
Barbara Gibson & Ty Simmons
SCAUG User Group Meeting, Broken Arrow
March 2nd, 2010

2. Introduction What we were asked to do
Creating the Economic Conditions Index
Using Spatial Statistical Tools
Mapping the Results

3. What we were asked to do
Approached by transportation planning staff to create an economic conditions index (similar to a Florida study) as a part of a TIGER grant application
The index categorizes block groups based on their level of distress as measured by 3 factors
Unemployment
Families in poverty
Substandard housing
The index is based solely on 2000 census data by block group
TIGER (Transportation Investment Generating Economic Recovery) which we were awarded!! The only one in the state! The project was a multi-model bridge on I-244 over the Arkansas River. Bridge would be the first of it’s kind in Tulsa built to accommodate highway, high-speed intercity and commuter rail, and pedestrian and bicycle traffic. The high speed passenger rail component gives the project national significance………….
The index score for each census block group was based on comparing its indicator scores to the average scores for the county in which the block group was located in our case Tulsa County

4. Unemployment – Tulsa County
P43. SEX BY EMPLOYMENT STATUS FOR THE
POPULATION 16 YEARS AND OVER [15]
Universe: Population 16 years and over
Total: P ,088
Male: P ,309
In labor force: P ,810
In Armed Forces P
Civilian: P ,516
Employed P ,173
Unemployed P ,343
Not in labor force P ,999
Female: P ,779
In labor force: P ,228
In Armed Forces P
Civilian: P ,181
Employed P ,683
Unemployed P ,498
Not in labor force P ,551
13,841 / 432,088 = 3.20%
Percent
Unemployment
Start with unemployment
P43 – table reference number for the data

5. Unemployment – Tract 7 BG 2
Census Tract 7
Block Group 2
P43. SEX BY EMPLOYMENT STATUS FOR THE
POPULATION 16 YEARS AND OVER [15]
Universe: Population 16 years and over
Total: P
Male: P
In labor force: P
In Armed Forces P
Civilian: P
Employed P
Unemployed P
Not in labor force P
Female: P
In labor force: P
In Armed Forces P
Civilian: P
Employed P
Unemployed P
Not in labor force P
Unemployment
42 / 691 = 6.08%
Percent
Unemployment

6. Unemployment Index Value
Unemployment Index Value for Census Tract 7 Block Group 2
6.08
1.9
3.20

7. Families in Poverty – Tulsa County
P90. POVERTY STATUS IN 1999 OF FAMILIES BY
FAMILY TYPE BY PRESENCE OF RELATED
CHILDREN UNDER 18 YEARS BY AGE OF
RELATED CHILDREN [41]
Universe: Families
Total: P ,189
Income in 1999
below poverty level: P ,962
12,962 / 148,189 = 8.75%
Percent
Families in
Poverty
P90 – table reference number for census

8. Families in Poverty – Tract 7 BG 2
Census Tract 7
Block Group 2
P90. POVERTY STATUS IN 1999 OF FAMILIES BY
FAMILY TYPE BY PRESENCE OF RELATED
CHILDREN UNDER 18 YEARS BY AGE OF
RELATED CHILDREN [41]
Universe: Families
Total: P
Income in 1999
below poverty level: P
37 / 207 = 18.14%
Percent
Families in
Poverty

9. Family Poverty Index Value
Family Poverty Index Value for Census Tract 7 Block Group 2
18.14
2.07
8.75

10. Substandard Housing Index
Housing Index is based on 3 variables
Housing Units lacking complete plumbing facilities
Home value for all owner occupied housing units
Year Structure Built for all housing units
To create the substandard housing index by block group we found an index for each variable and then took the average of the 3 overall to a have a composite index score.

11. Plumbing Facilities – Tulsa County
H47. PLUMBING FACILITIES [3]
Universe: Housing units
Total: H ,953
Complete plumbing facilities H ,404
Lacking complete plumbing facilities H ,549
1,549 / 243,953 = 0.63%
Percent
Housing Units
Lacking Complete
Plumbing

12. Plumbing Facilities – Tract 7 BG 2
Census Tract 7
Block Group 2
H47. PLUMBING FACILITIES [3]
Universe: Housing units
Total: H
Complete plumbing facilities H
Lacking complete plumbing facilities H
22 / 367 = 5.99%
Percent
Housing Units
Lacking Complete
Plumbing

13. Lacking Complete Plumbing Index Value
Lacking Complete Plumbing Index Value for Census Tract 7 Block Group 2
5.99
9.51
0.63

14. Home Value – Tulsa County
H84. VALUE FOR ALL OWNER-OCCUPIED HOUSING
UNITS [25]
Universe: Owner-occupied housing units
Total: H ,131
Less than $10,000 H ,264
$10,000 to $14,999 H ,279
$15,000 to $19,999 H ,345
$20,000 to $24,999 H ,065
$25,000 to $29,999 H ,751
$30,000 to $34,999 H ,659
$35,000 to $39,999 H ,605
$40,000 to $49,999 H ,057
$50,000 to $59,999 H ,084
$60,000 to $69,999 H ,140
$70,000 to $79,999 H ,268
$80,000 to $89,999 H ,183
$90,000 to $99,999 H ,290
$100,000 to $124,999 H ,583
$125,000 to $149,999 H ,885
$150,000 to $174,999 H ,700
$175,000 to $199,999 H ,562
$200,000 to $249,999 H ,378
$250,000 to $299,999 H ,837
$300,000 to $399,999 H ,598
$400,000 to $499,999 H ,142
$500,000 to $749,999 H
$750,000 to $999,999 H
$1,000,000 or more H
Total number of owner-
occupied housing units
with a value < $90,000
76,700
76,700 / 140,131 = 55%
Percent of owner-
Occupied housing units
With a value < $90,000
For Home value and year structure built have to first look at the median for the county. In our case Tulsa Counties medain home value was $85,000. Since we are using grouped values you have to make a choice and we went with all owner-occupied housing units with a value less than $90,000.
Tulsa County Median Value for all Owner-occupied housing units = $85,000

15. Occupied housing units
Home Value – Tract 7 BG 2
Census Tract 7
Block Group 2
H84. VALUE FOR ALL OWNER-OCCUPIED HOUSING
UNITS [25]
Universe: Owner-occupied housing units
Total: H
Less than $10,000 H
$10,000 to $14,999 H
$15,000 to $19,999 H
$20,000 to $24,999 H
$25,000 to $29,999 H
$30,000 to $34,999 H
$35,000 to $39,999 H
$40,000 to $49,999 H
$50,000 to $59,999 H
$60,000 to $69,999 H
$70,000 to $79,999 H
$80,000 to $89,999 H
$90,000 to $99,999 H
$100,000 to $124,999 H
$125,000 to $149,999 H
$150,000 to $174,999 H
$175,000 to $199,999 H
$200,000 to $249,999 H
$250,000 to $299,999 H
$300,000 to $399,999 H
$400,000 to $499,999 H
$500,000 to $749,999 H
$750,000 to $999,999 H
$1,000,000 or more H
Total number of
owner-occupied
housing units
with a
value < $90,000
172
172 / 192 = 89.58%
Percent of owner-
Occupied housing units
With a value < $90,000

16. Home Value Index
Home Value Index for Census Tract 7 Block Group 2
89.58
1.63 55

17. Year Built – Tulsa County
H34. YEAR STRUCTURE BUILT [10]
Universe: Housing units
Total: H ,953
Built 1999 to March H ,196
Built 1995 to H ,270
Built 1990 to H ,202
Built 1980 to H ,570
Built 1970 to H ,908
Built 1960 to H ,062
Built 1950 to H ,160
Built 1940 to H ,598
Built 1939 or earlier H ,987
Total number of
housing units
Built before 1970
111,807
Tulsa County Median Year Structure
Built for all housing units = 1972
111,807 / 243,953 = 46%
Percent of housing
Units Built
before 1970
Year built we follow the same principle. For Tulsa county the median year built was 1972 so we went with all housing units built before 1970.

18. Percent of housing units
Year Built – Tract 7 BG 2
Census Tract 7
Block Group 2
H34. YEAR STRUCTURE BUILT [10]
Universe: Housing units
Total: H
Built 1999 to March H
Built 1995 to H
Built 1990 to H
Built 1980 to H
Built 1970 to H
Built 1960 to H
Built 1950 to H
Built 1940 to H
Built 1939 or earlier H
Total number of
housing units
Built before 1970 =
302
302 / 367 = 82.29%
Percent of housing units
Built before 1970

19. Year Built Index
Year Structure Built Index for Census Tract 7 Block Group 2
82.29
1.79 46

20. Substandard Housing Index
Indicator scores for each block group were summed and
averaged to provide an overall substandard housing
Index score
Lacking complete plumbing index score = 9.51
Home Value index score = 1.63
Year Structure Built index score = 1.79
Substandard Housing Index Score for
Census Tract 7 Block Group 2 =
( ) / 3 = 4.31
Now that we have the indicator scores for the 3 variables for housing we can develop our composite index. Scores are for our sample block group – census tract 7 block group 2

21. Economic Conditions Index
Indicator scores for each census block are then summed
and averaged to provide an overall economic conditions
index score
Unemployment Index = 1.9
Family Poverty Index = 2.07
Substandard Housing Index = 4.31
Index score for Census tract 7 block group 2 =
( ) / 3 = 2.76
Now that we have our composite index score for substandard housing we will use the same formula for our overall economic conditions score.
An index score was created for each of the 410 block groups within Tulsa County

22. Index Scores Mapped
The results were mapped based on the natural breaks method using five classes.
We knew north Tulsa was an economically distressed area from the beginning and our analysis illustrates that.
We wanted to somehow emphasize the clustering of the high index score values around the I-244 bridge relative to the rest of Tulsa County, so we began exploring the spatial statistical tools in ArcGIS.

23. Using Spatial Statistical Tools
Are found in ArcToolbox
Toolsets Include:
Analyzing Patterns
Mapping Clusters
Measuring Geographic Distributions
Modeling Spatial Relationships
Cluster and Outlier Analysis Tool
Spatial Statistical Tools come standard with ArcGIS, is not an extension. Modeling Spatial Relationships toolset is not available with ArcView.

24. Using Spatial Statistical Tools
Spatial Statistics:
The love child of Geography and Statistics
Were developed specifically for use with geographic data
Incorporates space, such as proximity, area, and connectivity into the statistical process
Allows you to analyze spatial
Distributions
Patterns
Processes
Relationships
Differs from traditional statistics in that you are not making inferences about the data, rather you typically are dealing with all the available data in your study area
Traditional statistics typically works with a random sample with you trying to determine if your sample data is a good representation of the population at large.
For example, what are the chances that the results from my exit poll will reflect the final election results.
On the other hand, when you compute a statistic for the entire population with spatial statistics, you do not have an estimate, but rather a fact, since you are dealing with all the possible data.

25. Using Spatial Statistical Tools
The Statistics behind it all
The randomization null hypothesis – is used by many of the tools in the spatial statistics toolbox for statistical significance testing. It postulates that there is no spatial pattern among the features, or among the values associated with those features, in the study area.
Most statistical tests begin by identifying a null hypothesis, which is a statement of no effect or no difference.
The cluster and outlier analysis tool, like many of the spatial statistical tools, uses the randomization null hypothesis, which postulates that there is no spatial pattern among the features, or among the values associated with those features, in the study area.
If I were to pick up the index score values and throw them into the block groups I would have one possible spatial arrangement. The randomization null hypothesis states that if I do this an infinite number of times most of the time the pattern will be different than the observed pattern (what our actual index score map looks like). The randomization null hypothesis states that your data is one of many, many possible versions of complete spatial randomness. The data values are fixed; only their spatial arrangement could vary.

26. Using Spatial Statistical Tools
The Statistics behind it all
Z score – test of statistical significance that helps you decide whether or not to reject the null hypothesis. They tell us how many standard deviations our index scores are from the mean and in what direction
P-value – the probability that you have falsely rejected the null hypothesis. The smaller the p-value is, the stronger the evidence is against the null hypothesis
In order to determine whether or not to reject the null hypothesis, you have to derive a Z score and a p-value.
The Z score and the p-value help us determine whether or not the clustering we see on our map is actually statistically significant.

27. Using Spatial Statistical Tools
The Statistics behind it all
Both the z score and p-value are associated with the standard normal distribution, which relates standard deviations with probabilities and allows significance and confidence to be attached to the Z scores and p-values
Very high or low (negative) Z scores with very small p-values are found in the tails of the normal distribution
Using a 95% confidence level, the Z scores would be and and the p-value would be 0.05, which means you can reject the null hypothesis
95%
2.5%
When you perform a feature pattern analysis, such as cluster/outlier analysis, and it yields small p-values and either a very high or very low (negative) Z score, this indicates it is very unlikely that the observed pattern is some version of the theoretical spatial random pattern represented by your null hypothesis, thus you can reject the null hypothesis.

28. Using Spatial Statistical Tools
The Statistics behind it all
A Z score between and means the p-value will be larger than 0.05, thus the null hypothesis cannot be rejected
95%
2.5%
On the other hand, if your Z scores are between and and your p-value is larger than 0.05, you cannot reject the null hypothesis.

29. Using Spatial Statistical Tools
Cluster and Outlier Analysis
Analysis
Identifies clusters of features with similar magnitudes, as well as spatial outliers
It does this by calculating
Local Moran’s I Value
Z score
P-value
COType field
Interpretation
Positive I value indicates a cluster
Negative I value indicates an outlier
COType field distinguishes between statistically significant
Cluster of high values (HH)
Cluster of low values (LL)
Outlier with a high value surrounded primarily by low values (HL)
Outlier with a low value surrounded primarily by high values (LH)
It is important to note that the Cluster and Outlier Analysis tool requires projected data to accurately measure distance.
The Local Moran's index evaluates whether the pattern expressed is clustered, dispersed, or random. It can only be interpreted within the context of the computed Z score or p-value.
COType field gives you an alpha code for statistically significant features of HH, LL, HL, or LH

30. Using Spatial Statistical Tools
Cluster and Outlier Analysis

31. Mapping the Results

32. Mapping the Results Census Tract 7, Block Group 2
Local Moran’s I Value =
Z score =
P-value = 0
COType field = HH
A positive Local Moran’s I Value means that this block group is part of a cluster
The large Z score means that the block group is statistically significant.
The very small (non existent) p-value means we can safely reject our null hypothsis
And the Cluster type field reveals that this is part of a cluster of high economic distress values.

33. Other Applications Brownfield identification funding
Kendall-Whittier Tulsa Community Foundation
Anticipate using for environmental justice maps for 2035 regional transportation plan

34. Questions?
Contact Information:
Ty Simmons –
Barbara Gibson –
Phone: