## New Structures:

#### Select school:

#### Select area:

#### New Single-Family Homes:

Two or fewer bedrooms:

Three bedrooms:

Four bedrooms:

Five or more bedrooms:

#### New Townhomes:

Any number of bedrooms:

#### New Duplexes:

Any number of bedrooms:

#### New Garden Condos:

One bedroom:

Two bedrooms:

Three of more bedrooms:

#### New Mid-Rise Condos:

One or two bedrooms:

Three of more bedrooms:

#### New Elevator Condos:

One or two bedrooms:

Three of more bedrooms:

#### New Apartments:

Garden:

Mid-rise:

Elevator:

#### New Committed Affordable Housing Units:

Elevator:

Mid-Rise:

Garden:

Mixed Market/CAF Elevator:

Mixed Market/CAF Garden:

## New Elementary Students:

### Estimated by Arlington Analytics

### Estimated by Arlington Public Schools

The Boring Stuff: APS estimates are derived from formulas published in the Appendix (Student Generation Rates) in the 2019 Enrollment Report. Arlington Analytics estimates are derived from a dataset constructed with the following data:

- 2015-2019 Assessment Data
- Property Data
- Improvement Dwelling
- Housing: Affordable Units
- Realtor.com (for number of units)
- Apartments.com (for number of units)
- Selected property websites
- Spokeo.com (for number of units in very small buildings)
- APS Planning Unit Boundaries
- Planning Unit Statistics

Arlington Analytics estimates the county-wide relationship between residences and elementary school enrollment using a restricted, non-linear least squares estimation approach.
We explicitly do not assume differences in residence effects by school. Larger units (i.e., more bedrooms) are restricted to generate at least as many students on average
as smaller units. In many cases, the number of estimated students generated from larger units is indistinguishable from smaller units. This is why some options (four bedroom single-family houses,
for example) are not explicitly available. Residences can only have a non-negative effect on new student enrollment. Arlington Analytics locates the planning unit for every residence in
Arlington county, adds up the number of residences in each planning unit, and regresses the number of students against the number and types of residences.
This approach allows us to make better predictions across a wider variety of scenarios. For example, our approach allows us to make more sensible predictions of the effects of
much larger new houses relative to older, smaller houses. Additionally, it allows us to make more sensible predictions of the effects of new apartments, condos, and
affordable housing units in wide areas of the county.

Not all students will go to the school to which they are assigned, students may end up attending the county's option schools. Therefore, the effect
on the neighborhood school to which the new construction is designated may be smaller than predicted by either APS or Arlington Analytics. Arlington Analytics estimates
are based a one-year mismatch on the data---the dataset includes 2017-2018 school year enrollment matched with January 2019 property data.