All posts in Open Educational Data

Useful Data Points

Posted January 29th, 2011 by Bill

Open Educational Data

I've been thinking about the recently published information about salaries of PPS employees, and about why it struck me as pretty worthless from all perspectives: data journalism, informing the public, exposing something that was kept in the dark, helping forward the public good, etc.

So, in the interest of turning lemons into lemonade, here are some ways that dataset could be contextualized and extended to allow us to see something useful, instead of just putting a spreadsheet online.

The datasets outlined below would provide some different methods to provide detail around the salary data. The display of these datasets should be supported by faceted browsing and map-based visualizations.

Pay Scale and Job Description

This would have provided a reference point about the stated expectations of each role. Without this basic information, it's hard to say if someone is overpaid or not.

This would also separate union from non-union positions, and show how experience and education is (or isn't) valued.

More info available here: http://www.pps.k12.or.us/departments/hr/2763.htm

The choice not to include, or at least link to, this information decreases the value of the data.

School Data

For Oregon, you can grab some useful datasets from http://www.ode.state.or.us/data/reports/toc.aspx

National datasets are available from NCES: http://nces.ed.gov/datatools/index.asp?DataToolSectionID=4

School data, especially overall student population and % of students on the free and reduced lunch plan, should appear alongside teacher data. This would allow people to see number of students per school balanced against number of teachers/teacher salaries. It would also be good to see how some admin and support positions (assistant principals, counselors, etc) are spread across school populations.

Neighborhood Demographic Data

A good portion of this data is available from national census data. Other data points, like crime stats, liquor licenses, and lottery information, should be available from the state or local government. Portland has done some work around open civic data that could be useful here.

Percentage of renters to owners in the neighborhood
Crime stats from the neighborhood
Median income in the neighborhood
Liquor licenses, by type (restaurant, bar, sales)
Permits for lottery machines/gambling

Good Data Supports Good Questions

These data points would allow us to ask some more interesting questions.

How do teacher salaries compare in neighborhood schools where the majority of people rent versus the majority of people own?
Is there equivalency in salaries across school populations? How does this break down when comparing schools with a higher percentage of students on free and reduced lunch programs?
Are there any discernible patterns between salary levels and neighborhood demographics?
Are there any discernible patterns with depth of staffing/support and neighborhood demographics?

Conclusions

The people at Oregon Capitol News just made a spreadsheet searchable. It's a small step up from Excel, but not by much. If we want work with data to actually mean something, we need to have higher standards. Part of having higher standards involves providing additional information, data points, and tools to help interpret the data we are discussing. Mapping and visualization tools seem like pretty obvious additions.

And, we also have an obligation to advocate for the use of data in support of public good. As a point of common human decency, the names of people working in the schools should be anonymized. It's intrusive and unnecessary to publish the salary information of people working within schools. Salary data is one thing, but the argument that publishing personal financial information about people working with kids provides a public good is tenuous at best.

Image Credit: "Life gave us lemons" taken by Steve Mohundro, published under an Attribution-NonCommercial-ShareAlike license.

When Data Does Nothing Useful

Posted January 26th, 2011 by Bill

It looks like the folks over at Oregon Capitol News published all the salaries of every Portland Public School employee under this breathless heading:

Ever wondered how much Portland Public School employees make? Oregon Capitol News now provides a searchable database of employee salaries from Portland Public Schools.

I admire their restraint in not using any exclamation points.

So now, you can find out exactly how much teachers make, and, depending on your political/social leanings, you can say either, "Did you see that Ms Jones makes X a year? Just another example of wasteful government spending," or, "Did you see that Ms Jones makes X a year? Teachers don't get paid anything. Why would anybody go into a profession where you are subjected to long working hours, endless criticism, armchair quarterbacking, and a crappy salary?"

No, the former football coach at Portland's Lincoln High doesn't make $140,000 a year | OregonLive.com

However, one small problem. The folks at Oregon Capitol News got it wrong. They published inaccurate data. And to make matters worse, they don't appear to admit it, or to take any responsibility.

This is an amateur's mistake, and it's one that calls the overall accuracy of what they are attempting to do into question. Anybody working with datasets of any size knows that you need to vet the data you are given for accuracy, reliability, and any one of countless abnormalities that arise from the reality that data takes work to keep in a usable form. Posting this data without any context (like the pay scale? Like job descriptions?) is sloppy. Posting it with inaccuracies, and then denying any accountability for the accuracy of what you publish compounds the problem, and reduces or eliminates the usefulness of the data you release.

And, as there is more of a push toward big data journalism, these types of sloppy mistakes need to be eliminated.

Also of interest: the Oregonian's "reporting" on this story comes complete with a picture with the following caption:

Lincoln High fans are passionate about their Cardinals. But could they pull strings to make a teacher and former winning coach one of the highest paid employees in the Portland district?

The answer to this rhetorical question is, of course, no. That's the point of the article. That's the mistake they are actually reporting on. But, by including incorrect information as a question, they create the impression that maybe, just maybe, those nefarious schools are pulling strings to get Special Advantages (™). I don't know whether this falls under the category of outright dishonesty or a feeble attempt to make a story more sensational by stirring up the fears of a system gone awry. In any case this type of misinformed slop is also not really surprising, because it's published by an education writer in the Oregonian.

Building Toolkits to Access Open Educational Data

Posted June 30th, 2010 by Bill

The push for open data from government has been gaining momentum over the last few years. This has taken concrete shape in the form of data being released in more usable formats, including websites that are especially designed to release general data and educational data at the federal level. Some local governments are following suit.

As more datasets become accessible - and this includes comparable datasets released over multiple years - we're interested in helping these datasets bridge the chasm between data and information. It's one thing to be able to map crime incidents in a specific area between two dates. It's another thing to be able to look at some performance indicators of schools within these same areas. It's still another to be able to look at a low performing school and high performing school, and compare student to teacher ratios, the percentage of students eligible for free and reduced lunch, the number of crime reports in the neighborhood, the median income, the number of liquor licenses, and the percentage of owners to renters. By leveraging open datasets, we will enable more precise questions about how people learn, and the factors that affect learning.

In the upcoming days, weeks, and months, we will begin to document and release the beginnings of a toolkit designed to simplify building and maintaining web sites that use and display open data as it relates to education and learning. Drupal already has much of the functionality in place to support using open data, and the Drupal community can play a leading role in pushing the use of open data.

Our initial three targets include:

Datasets: we will identify existing datasets, and document any needed steps required to prepare them for more widespread use.
Map tiles: much of this data can be represented geographically, at the country level, the state level, or the local level. We will identify existing map data that can be used when developing location-based visualizations, and document the process of creating new map tiles. Any new map tiles we create will be released out under a Creative Commons license.
Search and Visualization techniques: As we develop tools to import and manipulate datasets, we will need to search and filter this information, and make it comprehensible. This will likely result in new code that integrates various pre-existing open source visualization tools with the Views module. Any new code we develop will be released back onto Drupal.org

We hope that the process of looking at information related to learning will help people more effectively discover, understand, and explain the connections between student learning, school performance, the role of the teacher, and the relationship between all of these factors and the socioeconomic milieu in which they occur.

We also want more people to get involved. Toward that end, we have created a working group to discuss strategies and approaches for using open data - if you want to get started working in the space, join the group. Many hands make light work, and much of this work requires research and familiarity with how educational data is structured, as opposed to any hardcore coding or site building expertise.

Within the next few days, I'll have another post out that defines what we see as a starting point for the toolkit, and identify some of the tools that already exist that move us toward that point.