LeeAnn has more up-to-date information, but I think you’re right. There have been some data science courses at private schools, but I think this is the first data science class to be deployed in public schools, particularly at such scale.

As for hurdles, one of the main ones is professional development for teachers. The grant funded us to do a lot of this, but we want the curriculum to live on once the grant is over. PD usually only lasts a week, and it’s hard to get everyone up to speed on a year-long course, particularly with the new technology.

One of the exciting pieces of the course is that students collect “participatory sensing” data with their smartphones. This data isn’t randomly sampled, so the inference you can make is limited. But, it has a lot of interesting properties— the data is geotagged, so students can make maps, there is often text data, etc. We also suggest many data sources in the curriculum, including movie data from the IMDB and the American Time Use Survey.

One of the ways we skirt the inference issue is by using randomization and the bootstrap instead of standard statistical formulas. And there is certainly a component to the course of critical thinking about data. Who collected it? Why? What might be missing?

To add to Amelia’s excellent response: There is a tension between posing authentic, open-ended questions and providing structure. We’ve achieved some success in this direction by emphasizing the role of the “Data Cycle”, which is what the Guideline for Assessment and Instruction in Statistics Education (K-12) call “The statistical investigative process”. By paying attention to this cycle, it becomes clear that students and teachers must be thoughtful about (a) the questions they ask and (b) the data that might be used to answer these questions. This is quite a skill (and we’ve found that many adults struggle here, too!), and our solution has been to provide many opportunities and contexts in which to practice.

Thanks for that question, Irene. I’m the lead PI for the project. There are two components to starting your own IDS class. The first is the curriculum itself. This will appear any day now on our website mobilizingcs.org. The second is the technology suite, which is substantial. We’re currently working on providing a fee-for-service structure to support schools that do not have their own servers or who wish to “test drive” the technology. During this upcoming year we are preparing a “Mobilize in a Box” that will provide instructions for installing and maintaining the software on a server if a district does not want to go the fee-for-service route. In the meantime, this link http://www.mobilizingcs.org/technology provides documentation, source code, and a webpage that lets you do a small-class trial. and play with the technology. http://www.mobilizingcs.org/technology

And, last but not least, this link https://lausd.mobilizingcs.org/#demo
let’s you ‘play’ with the visualization tools that illustrate the mapping and word-clouding abilities that Amelia mentioned up above.

Thank you! I’m looking forward to giving it a try.

Thanks for your question. By year round course, do you mean a course that is to be taken and completed in one school year? The reason I ask is because in the Los Angeles Unified School District (LAUSD), year-round is a schedule to relieve school overcrowding. At LAUSD, the course has been implemented in a variety of different schedules, and there is no reason why it couldn’t be done on a year-round schedule. To date, we have recruited 27 schools to offer the course, 34 teachers have received professional development, and approximately 2,000 students have taken the year-long IDS course. We will be recruiting 20 additional LAUSD teachers to teach IDS in 2016-2017.

Hi Rob! I certainly agree that there is room for both approaches, and that a separate course circumvents some of the obstacles (such as teacher preparation) present in the integration approach.

I do hope that the integration approach does not get neglected, though. In the real world data science is done in context and data habits of mind like “Where’s the data?” are (or should be) important for all members of a work group. I know that IDS promotes this point of view, but I worry that many data science courses won’t.

You’re absolutely right. There must be both. I’m somewhat surprised (and saddened) for instance, that there’s no “statistics” keyword search, even though several projects seem intent on improving data literacy and statistical thinking.

One of the interesting findings - and now that I type this I think it is also a partial reply to Caitlin Martin’s question above- is that it has been very difficult to get classrooms to develop their own projects. We did lots of soul searching after the first pilot to figure out why, and this is still an area of on-going research. One reason seems to be that the participatory sensing tool has certain constraints and these can be, well, constraining. The other, though, seems to be that teachers are initially inexperienced in the statistical investigation cycle, and need practice themselves before they are able to lead a classroom towards posing interesting questions that can be addressed with data. As Caitlin mentions, teachers sometimes neglect to use the data, and we think in part it’s because they simply don’t know what to use it for! We are looking forward to reviewing classroom projects from this second year of IDS to see the extent to which classrooms struck out in new and unexpected directions, or whether they stayed close to the data-collection projects as they are in the curriculum.

Thanks! We do in fact have our own package: mobilizR. Our intent with this package was to simplify the syntax. Our slogan: “No dollar signs”. The idea being that, as often as possible, students use the formula syntax.
That said, R is still counter-intuitive. However, because an objective of the course was to teach computational thinking, we decided it was important to grab the bull by the horns and teach R.

mobilizR, by the way, is based on the awesome package mosaic, which is part of an NSF-funded project led by Randall Pruim that provides everything you need to teach R in an intro stats (college level) course.

Thank you, Elizabeth. The “Civic engagement” happens in several ways. First, the curriculum asks students to collect data about their surroundings (where they dispose of trash, what they eat, their stress levels during the day, how they see water being used), and these give opportunities to discuss what is happening in their surroundings and how this might change. Second, the participatory sensing tool gives classrooms the means to determine what they think will be of interest, and to collect and analyze data as a classroom to address these interests. Thus, there’s lots of potential for the tool to be taken in new directions within the classroom. Finally, when data are collected, they are stamped with a location tag, and, once some privacy gates have been passed, the data can be displayed on a map. We’ve found that displaying a map with information collected by students often leads to questions involving social and civic issues.