Iteration 6 - Analysis: Economic Profile and Poverty

Analysis

ResultSets

For the analysis queries in this section, we’ll often be reporting multiple results in answer to a given query.

To clarify this process, let’s use a dedicates ResultSet object that aggregates and provides an interface into the given results.

The ResultSet will represent the aggregate collection of results, and each data element will be respresented as a ResultEntry within that collection.

The ResultSet should have the following methods:

• #matching_districts - returns an array of ResultEntry objects
• #statewide_average - returns a single ResultEntry object representing the average across the state

The ResultEntry should be a simple wrapper around a hash that exposes certain keys of the hash as methods.

The accepted keys will be the following:

• :free_and_reduced_price_lunch_rate
• :children_in_poverty_rate
• :high_school_graduation_rate
• :median_household_income

Each of these keys should be accessible by invoking a method of the same name.

Example:

r1 = ResultEntry.new({free_and_reduced_price_lunch_rate: 0.5,
	children_in_poverty_rate: 0.25,
    high_school_graduation_rate: 0.75})
r2 = ResultEntry.new({free_and_reduced_price_lunch_rate: 0.3,
	children_in_poverty_rate: 0.2,
    high_school_graduation_rate: 0.6})

rs = ResultSet.new(matching_districts: [r1], statewide_average: r2)

rs.matching_districts.first.free_and_reduced_price_lunch_rate # => 0.5
rs.matching_districts.first.children_in_poverty_rate # => 0.25
rs.matching_districts.first.high_school_graduation_rate # => 0.75

rs.statewide_average.free_and_reduced_price_lunch_rate # => 0.3
rs.statewide_average.children_in_poverty_rate # => 0.2
rs.statewide_average.high_school_graduation_rate # => 0.6

Note That every key might not be present for every ResultEntry – the methods whose keys are missing should return nil in these cases.

Which districts have both high poverty and a high school graduation rate?

Which districts match all of these criteria:

• Above the statewide average in number of students qualifying for free and reduced price lunch
• Above the statewide average percentage of school-aged children in poverty
• Above the statewide average high school graduation rate

rs = ha.high_poverty_and_high_school_graduation
# => <ResultSet>
rs.matching_districts.count
# => 2
rs.matching_districts
# => [<ResultEntry>, <ResultEntry>]
rs.matching_districts.first.name
# => "District 1"
rs.matching_districts.first.free_and_reduced_price_lunch_rate
# => 0.021
rs.matching_districts.first.children_in_poverty_rate
# => 0.023
rs.matching_districts.first.high_school_graduation_rate
# => 0.67
rs.statewide_average
# => <ResultEntry>
rs.statewide_average.free_and_reduced_price_lunch_rate
# => 0.019
rs.statewide_average.children_in_poverty_rate
# => 0.023
rs.statewide_average.high_school_graduation_rate
# => 0.67

Where do we see significant income disparity?

There are obviously going to be some wealthy and some poorer school districts. But which districts show a high disparity in income by both having a high median income and a high rate of child poverty?

Which districts satisfy all of the following:

• Above the statewide average in median household income
• Above the statewide average percentage of school-aged children in poverty

rs = ha.high_income_disparity
# => <ResultSet>
rs.matching_districts.count
# => 2
rs.matching_districts
# => [<ResultEntry>, <ResultEntry>]
rs.matching_districts.first.name
# => "District 1"
rs.matching_districts.first.median_household_income
# => 52000
rs.matching_districts.first.children_in_poverty_rate
# => 0.023
rs.statewide_average
# => <ResultEntry>
rs.statewide_average.median_household_income
# => 48000
rs.statewide_average.children_in_poverty_rate
# => 0.017

How does kindergarten participation variation compare to the median household income variation?

Does a higher median income mean more kids enroll in Kindergarten? For a single district:

ha.kindergarten_participation_against_household_income('ACADEMY 20') # => 1.234

To determine this value, we’ll take the district’s kindergarten_variation and divide it by the district’s median_income_variation, where:

• kindergarten variation is defined as the district’s average kindergarten participation compared against the state’s average as described in iteration 1.
• median income variation defined as the district’s average median income divided by the state’s average median income as defined in iteration 3

Then dividing the kindergarten variation by the median income variation results in 1.234 in the sample.

Statewide does the kindergarten participation correlate with the median household income?

If this result is close to 1, then we’d infer that the kindergarten variation and the median income variation are closely related.

Let’s consider the kindergarten_participation_against_household_income and set a correlation window between 0.6 and 1.5. If the result is in that range then we’ll say that these percentages are correlated. For a single district:

ha.kindergarten_participation_correlates_with_household_income(for: 'ACADEMY 20') # => true
ha.kindergarten_participation_correlates_with_household_income(for: 'COLORADO') # => true

Then let’s look statewide. If more than 70% percentage of districts across the state show a correlation, then we’ll answer true.

ha.kindergarten_participation_correlates_with_household_income(for: 'STATEWIDE') # => true

And let’s add the ability to just consider a subset of districts:

ha.kindergarten_participation_correlates_with_household_income(:across => ['ACADEMY 20', 'YUMA SCHOOL DISTRICT 1', 'WILEY RE-13 JT', 'SPRINGFIELD RE-4']) # => false