Blog

Recently my wife and I started looking for a new house here in Portland, OR.  We bought our house B.C. (i.e., Before Children), when a two bedroom one bath house met our needs perfectly.  Seven years and two daughters later, our notion of the ideal home has changed dramatically.  I, for one, wouldn’t mind living in a house where daddy’s office wasn’t a tent in the backyard.

This time around, I’m much more concerned with the neighborhood schools than with our last home purchase, since my wife and I both believe strongly in public school education for our girls.  Also, I’m cheap, so I thought it might be interesting to use my technical skills to find the best schools in the most affordable neighborhoods, focusing my house-hunting efforts there.

The results of one of my analyses can be found here, in a map visualization I created with Tableau Public.  Tableau Public is a fun and free tool for putting together graphics and data in order to identify pattern and relationships within the data.  In the corporate world, they call this process “business intelligence”.  In my field, they call it “conducting research”.  Whatever you call it, Tableau Public makes the process almost as simple as uploading a You-Tube video. 

I created the visualization by joining together data from two publicly available sources:  Greatschools.org and the National Center for Educational Statistics (NCES) Common Core Database.  Greatschools.org is a website that publishes information about schools that might be of interest to parents, including school descriptions, maps, written reviews, and school ratings (composites based on students’ standardized test scores).  The NCES Common Core Database is a research tool that provides a wealth of information about the schools and districts within all 50 states, including physical addresses, total student enrollment, teacher/pupil ratios, and eligibility rates for free or reduced price lunches.  This database is primarily intended for researchers gathering large amounts of data for many or all schools within a state.  I joined data from these two sources to create a visualization illustrating the relationship between Great Schools Rating and the percentage of students within that school who are eligible for free or reduced price lunch (a common indicator of school poverty) for the neighborhood schools in the Portland Public School District.  Not too surprisingly, the figure shows a downward sloping line, with higher test scores predicted by lower poverty rates.  What did surprise me was how strong a predictor it was:  More than 80% of the variance in school ratings was predicted by the school poverty rate.

My initial reasoning for creating the visualization was simple:  I’d just read Michael Lewis’ book, “Moneyball”, about how the Oakland A’s General manager Billy Beane had employed statistical models to identify undervalued baseball players whom he then recruited to the team at bargain salaries.  I wanted to try something similar with the Portland neighborhood schools – identifying outlier schools in the most affordable neighborhoods with the highest possible school ratings.  While I did find my outliers, I was also struck by just how limited such ratings are, since essentially all they show are which neighborhoods are the most expensive – something I knew already knew from reading real estate listings.

Yet those ratings are important for parents and students with college aspirations:  students from the highest performing public schools are more likely to qualify for merit-based scholarships when they go to college.  In other words, students growing up in the wealthiest neighborhoods qualify for the most aid to go to college, where they acquire degrees and training granting them access to jobs commanding greater salaries.  Eventually, these higher salaries allow them to buy their own houses in the most expensive neighborhoods with the best public schools, continuing the trend in the next generation.

Attending a wealthy neighborhood school doesn’t necessarily cause students to perform at higher levels.  Correlation doesn’t imply causation.  In fact, school poverty rates have very little predictive relationship on how much students improve over time (see our recent High Flyer study).  But families who can afford to live in wealthier neighborhoods generally have the means to provide educational experiences during the summer or outside the classroom that may not be available to students in poorer schools.  For this reason, instead of trying to buy a house in one of the wealthiest Portland neighborhoods, I’m saving my money.  Expensive neighborhoods won’t get my daughters into college.  I will.