Path to Personalization: Better Models and Better Tests
Gene Wilhoit recently said, "We've given up on schooling; we need to shift to systems that enable high-quality learning and educational opportunity for all." Gene describes three critical attributes of a system of personalized learning:
- Clear, high expectations: for the knowledge, skills, and dispositions necessary for success in college and careers;
- Competency-based education: demonstrating learning through meaningful assessment; accountability for the learner and those who support the learner.
- Customized pathways: anytime, anywhere learning opportunities with comprehensive supports.
Wilhoit's vision is focused on the student learning experience: what is learned, how it's learned, where learning happens, how learning is demonstrated, and how learning is credentialed. The former head of the Council of School State Chief Officers said teachers deserve professional development and resources equal to the challenge.
Wilhoit leads the National Center for Innovation in Education at the University of Kentucky.
With his colleagues Linda Pittenger and Carmen Coleman, Wilhoit led a discussion on Monday about the ways that technology is improving formative assessment and the opportunity that it creates to rethink big year-end tests.
The trio embraced a broader set of outcomes, then tested subjects. Gene said, "It is how one applies knowledge that makes the difference of success and failure." (See a discussion of What High School Graduates Should Know And Be Able To Do for a summary of Coleman's work in Danville, Kentucky).
With Linda-Darling Hammond, Wilhoit's vision for a new system is laid out in their August paper, Accountability for College and Career Readiness: Developing a New Paradigm. Gene outlined the next-gen system vision at the iNACOL Symposium.
Wilhoit thinks there are about 15 states ready to adopt a forward leaning accountability system. Anti-testing sentiment and red state frustration with federal intervention could make the number of states ready to move even higher, especially with a one or two lead states that show the way.
Student-centered hunch. As recently noted, it's noteworthy that student-centered learning is the new sector hypothesis. Twenty years ago aligned instruction was the big bet, and NCLB doubled down on that bet with year end tests for age cohorts. As noted last month, it's easy to say that managed instruction is old school and digital learning is the future. But this debate is worth unpacking. The lack of alignment of instruction and assessment, as well as schedule, structure, and support services is a huge problem in education. It leads to poor student performance and frustration on the part of educators.
Rather than alignment around an age cohort, the new bet is alignment around every student, individual learning progressions that leverage student interests.
Dan Ingvarson, who has been working with Australia's National Schools Interoperability Program, and now runs his own Education Technology Consulting firm (Aprina.net) said, "I have seen digital and non-digital point solutions, which cause change, but not at scale," noting the presence of many barriers and constraints in the traditional system in both countries.
Bror Saxberg, Kaplan Inc. CLO, said student-centered learning is not just about differentiating levels, it's personalization around motivation, a topic that warrants more primary research.
In Getting Smart (2011), I argued that customization would lead to productivity (more learning per hour) and improved motivation would lead to more time learning. We opened Smart Cities (2014) with a description of next generation learning, which is blended, personalized, and competency-based. Early models and trials suggest this direction is promising .
The emerging vision (e.g., Nellie Mae's student-centered, Hewlett's deeper learning) is a hunch that student agency will produce better education outcomes. It suggests more student-control over path and more more interest-driven learning. Acton Academy and Big Picture Learning are relatively low tech versions of student-centered learning. Summit Public Schools is more of an engineered solution with a pedagogy mixing playlists and projects.
Some experts consider the regime of standardized testing to be the primary barrier to more student-centered learning. But Joel Rose, New Classrooms, said, "This is not an accountability challenge, this is a design challenge." Saxberg agreed on the need for more design engineered solutions (i.e. defined outcomes and learner experiences supported by a learning platform which facilitates individual learning progressions). "The barrier is R&D," according to Rose who thinks philanthropic and impact investors should invest in more next generation learning models.
A Path Forward
Wilhoit complimented the state consortia on the development of better summative assessments. However, it's ironic that consortia delivered better but bigger tests in the school year likely to go down as the beginning of the end of big summative. The protest over longer tests comes at a time when it's increasingly clear that good schools know how every student is doing in every subject every day.
To, "Drive down the social, political, reporting and educational reasons for these high stakes tests is a way of joining many smaller, immediate, informative and safe digital components into an ecosystem," said Ingvarson. The group identified five of these next steps for the Center, states, and funders.
1. Disaggregate data uses. With the rise of the Internet, most sectors left the age of data poverty 20 years ago. In education we continue to use big year end tests to improve learning, manage student matriculation, evaluate educators, and check school quality. Now that schools are connected and most students are learning online part of the day, learners and teachers are benefiting from more formative feedback. Like good schools, states and districts can build assessment plans that use discrete data sources for different purposes, an approach likely to make both psychometricians and lawyers happy. Assessments shrink when stakes are reduced. A 10 minute quiz can confirm unit level mastery for an individual student, while it takes a week to verify everything about a school full of students.
The biggest opportunity is for assessment frameworks that support competency-based learning sequences. Districts and networks of schools could develop assessment systems that according to iNACOL , "Measure individual student growth along personalized learning progressions," and, "Use multiple measures of learning, including performance-based assessments." A state could create an innovation zone to pilot the use of on-demand (or frequently scheduled) end of course demonstrations of learning to manage student progress thus reducing the need for end of year exams.
2. Tagging formative. New standards and platforms are making it easier to combine formative data by linking to standards. A few emerging examples include:
Agilix Buzz , used by the EAA in Detroit, links instructional units to multiple forms of assessments. Buzz allows teachers and curriculum specialists to easily align instructional objects at any level (folders, units, lessons, activities, etc.) to state, common core, or any customized set(s) of standards. This mapping facilitates: 1) mastery reporting and analytics, 2) provision of actionable data to teachers for personalized intervention, and 3) formative assessments that, based upon individual performance, provides every student a personalized learning path through the curriculum.
Instructional partners of Houston ISD are required to use the IMS Global Thin Common Cartridge, which enables content interoperability using Learning Tools Interoperability (LTI). Learning platform itslearning combines formative assessment items in a mastery tracking tool which drives a recommendation engine. A recent Education Week post suggests that other districts and some vendors appreciate Houston's leadership on interoperability standards.
MasteryConnect allows teachers to build standards-aligned assessments (using multiple item types and drawing from three item banks) and to add new assessments to a master tracker.
Edmodo's free assessment tool, Snapshot, allows teachers to easily and frequently administer standards-aligned assessments (using multiple item types from three high-quality item banks) to help gauge mastery, and, recommends high-quality resources (aligned to specific standards) to help students improve their mastery.
Scootle is an Australian library of tagged learning resources (an example of tagged resources a state or network could share).
It's likely that that we'll see several K-12 badge systems that support competency-based learning system. There are several possible tagging schemes that avoid fragmenting the standards while supporting idiosyncratic instructional approaches.
"Tagging has to be considered carefully," advocates Ingvarson who notes he has seen several generations of tagging go down the drain with the introduction of new standards. He believes, "Creating a robust concepts layer where content, learning standards and assessments can all move independently yet continue to work makes a more sustainable model," and "The conceptual tagging model is an example of interoperable infrastructure that is needed to make things work together."
Data analysis and achievement analysis tools are getting better and provide some help combining assessment from different sources. Several feature data visualization (e.g., BrightBytes, Schoolzilla, EdElements).
3. Assessment solutions. Robust learning platforms with embedded assessments (New Classrooms, DreamBox, i-Ready, Realizeit) are typically a closed adaptive system often limited to one subject. They provide personalized learning experiences with continuous performance feedback. While student-driven and leveraging gamification, most of these systems aren't interest-driven. Some are likely to become more robust allowing new learning experiences (e.g., open content, teacher developed assessments).
In addition to EdTech startups, school networks are an important source of innovation in school models and learning platforms (e.g., Summit Public Schools, New Tech Network, Brooklyn Lab). Regional new and transformed school grant programs (like those supported by NGLC in six cities) not only create great new school options, they advance creative uses of formative assessment.
4. Post hoc data barrage. Without going through the trouble of tagging data or putting up with the limitations of a closed system, it is increasingly possible to subject to correlate data sequences from different digital learning applications after the fact by analyzing thousands of data points.
IMS's Caliper Analytics standards support both post hoc and real-time data feeds for millions of students daily. In May, IMS will launch the Caliper developer community featuring Caliper API implementations in six different programming languages.
5. Learner profiles. As described in Data Backpacks and MyData Initiative, parents and students should benefit from a portable mobile learner profile and portfolio. In addition to being about to download school information, a learner profile could capture information from tutors, summer camps, and learning games.
While the mobile "backpack" would provide instant access to progress monitoring tools, these next-gen apps won't need to carry lots of data around. Rob Abel said linked data (JSON Linked Data), "Is possible for the student, teacher, parent to see all of the relevant outcomes, dashboards from different systems in one application without any data actually moving." As a result, the same dashboard can show up in several 'backpack' applications without any data actually moving the same way Google Maps can show up on any web page.
Backpacks will be a mash-up. IMS's Learning Tools Interoperability (LTI) was designed to be such a mash-up language for educational apps. There are currently 40 Learning Platforms that can accept LTI apps. Next steps will require vendors to make use of more advanced LTI features. Abel thinks that will, "Enable a very fluid definition of a 'learning profile' that reflects what the student actually does versus some preconceived one-size-fits-all profile."
Curt Allen, Agilix, thinks real time actionable data would create a "firestorm of demand."
Imagine if parents had access to powerful visual tools on their phone that told their student's learning level and rate of growth.
6. Infrastructure for interoperability. This last one is the layer which adds the tools to link up the others. Ingvarson described the Australian Data Protection Hub, a manifestation of a government (or NGO) owned service providing enablement with enforced privacy and simple mechanisms for vendors to implement. "This glue reduces the steps for each player in the above set to actually achieve something. Rather than just talking about how nice it would be if we were able to put the right data in the right place at the right time, we now see a government funded open source tool set to help do that." Given recent failures and campaigns, it's hard to figure how this layer gets built in the US. But with a couple of capable NGOs, a group of states could make progress.
Ingvarson summarized his description of the solution this way:
Disaggregate: The where and when freedom to undertake activities which could be used as assessments.
Tagging: This concept is the conceptual framework - how do we align the values we receive back from disparate systems.
Assessment solutions: The rise of the ecosystem of tools to enable sometimes large, sometimes small activities which can be considered as assessments.
Post-hoc analysis: The process to make sense of the data, which has been returned in any format to determine which contributes to being relevant to each of the pillars of the current high stakes test.
Learner profile: Meeting the needs of some of the participants for the feedback which high stakes tests meet.
Infrastructure for interoperability: Partnerships of those with the duty of care over the student's digital representation.
As outlined in their paper, Wilhoit is confident that developments like those outlined above will mean a gradual reduction in the size of and reliance in heavyweight summative assessment to smaller, less frequent tests that check the validity of local assessments. Ingvarson said, "This will enable a richer and more diverse field for teachers and therefore as above employ them to do more of what is right for the students and less of what is right for the test."
At iNACOL, Wilhoit said this all means a different discourse between students and teachers, it encourages authentic conversations and collaborative dialogue about student learning. For learning leaders, it means being ready to say, "We don't give up on mastery, we provide the supports for the student to get there."
For more, check out: