What is Adaptive Learning?

In education, adaptive learning tools are built around the interaction between teachers, students, and automated technologies, often addressing the imbalance between teachers’ limited capacity and students’ high demands.

A large group of students might need individual attention the instructor cannot possibly provide on their own, but adaptive learning courseware can deliver personalized learning activities using a model designed to measure student knowledge states. Within a subject-appropriate framework, it adapts activities to match students’ strengths, weaknesses, proficiencies, and knowledge gaps.

How does adaptive learning work?

Most adaptive learning functionality follows a similar process:

  1. Set goals for the student’s work session (like a homework assignment) and/or the whole educational context (like a course). This frequently involves an instructor creating a syllabus or assigning some learning objectives.
  2. Deliver assessment questions relevant to those goals.
  3. Estimate the student’s knowledge state based on the correctness of their answers, the difficulty of the questions, and/or the alignment of the questions to a knowledge graph of skills, concepts, or learning objectives.
  4. Decide what activity should come next based on the student’s estimated knowledge state—whether that means delivering easier or harder questions, providing an instructional intervention, remediating a gap in prerequisite knowledge, suggesting the student speak to their teacher, or some other activity.

The cycle of steps (two to four) then continues until each student reaches the goals set in step one.

(*Learn more about how Wiley’s proficiency model estimates student knowledge here)

What are the benefits?

Even in early days, the most immediate benefit of adaptive learning was personalization at scale: Students can receive something like personalized attention without requiring 10 times as many instructors. Adaptive learning tools act as “virtual teaching assistants,” engaging each student with the material most likely to improve their learning outcomes.

When students can receive assessment and instruction that meets them at their current knowledge state—which may differ significantly from that of their peers—they’re more likely to reach higher levels of achievement. Students requiring more practice, instruction, and remediation to prerequisite skills can receive additional support, while more advanced students in the same cohort can engage with challenging problems.

Learning science studies have shown that adaptive learning generally causes “all boats to rise” and that computer-assisted personalization at scale can be at least as effective as human-run group tutoring sessions in pedagogically appropriate contexts.

What are common challenges?

How does adaptive learning courseware shift instructors’ and students’ mindset?

For instructors, using adaptive technology typically means giving up step-by-step control. For the courseware to successfully adapt to a student’s needs, it can’t just deliver pre-selected materials in an instructor’s pre-set order. Successful courseware implementations of adaptive learning aim for a middle ground, where instructors choose the goals, content areas, and/or pedagogical functions for each student activity, but the adaptive technology assigns variable amounts and types of content to students as needed. The adaptive learning courseware should also provide easily interpretable analytics and calls to action so the instructor understands each student’s learning and what kinds of activities or 1:1 attention would yield the best results.

For students, adaptive learning activities may be different from their experience with “traditional homework.” Working on assignments differing in length and type from that of their peers can feel arbitrary and punitive without an explanation of the benefits of this arrangement. Additionally, students typically have a fear of submitting wrong answers since that would traditionally penalize their final grade. Adaptive learning usually reframes this penalty: Students are expected to get wrong answers as they learn and can achieve an excellent final grade on an adaptive activity even after a poor start, provided they put in the effort to learn from their mistakes. We find that students tend to respond positively to this “growth mindset” orientation.

How do Wiley and Knewton Alta approach adaptive learning?

Our two solutions support mastery-based learning in different ways:

Evaluation Guide for Adaptive Learning Courseware

Assess your needs and evaluate your courseware options with our free guide

Assess your needs and evaluate your courseware options with our free guide

Academic rigor, reporting, accessibility, affordability – there’s a lot to consider when evaluating courseware. That’s why we asked instructors who use Knewton Alta what crucial questions they asked—or wish they’d asked—when considering which courseware to adopt.

Download our courseware evaluation guide to help you think through some of your most pressing concerns.

Download guide
Assess your needs and evaluate your courseware options with our free guide

Knerd Story – Josh Guillemette, Valencia College

Josh Guillemette
Statistical Methods and Math
Valencia College

How long have you been using Knewton Alta?

Five years.

How are you teaching?

I teach in all modalities. Fully online and then what we call blended learning, where we meet face-to-face once a week, and the rest of the instruction is online. It’s almost like flipped. Most of the instruction is online, and then when students come to me, I just answer their questions and help them with their homework and send them off again unstuck.

And then I actually do have one class that’s fully face-to-face. We meet twice a week, all six of the students in the class because of social distancing.

Why did you start using Knewton Alta?

My Wiley rep stopped by my office and told me about the product. She explained it was adaptive and took students’ answers and figures out where they’re at and guides them along the path.

The price ($44) was appealing too. At the time we were using Pearson’s My Math Lab (MML), and a lot of people still use it. It’s around $104 at the bookstore. So, it’s a difference of $60. I thought if Knewton Alta is even half as good, it’s worth checking out. So, I got permission to do a pilot and I got four other faculty members to do it with me.

Two of us had half of our classes use Pearson, and half of them used Knewton Alta. And then I had two more Pearson, but they gave the same final, and then I had another person doing all Knewton Alta. So, there was quite a smattering of implementations, and we were able to collect and analyze some data.

For first-year statistics, we use the CAOS (Comprehensive Assessment of Statistics), which is a peer-reviewed statistical reasoning instrument for the final. I did an analysis and, obviously, the teaching made the most difference. But they were all really good instructors, so there was only a little difference. There wasn’t really much difference, statistically, between what the Knewton Alta folks did and the Pearson folks did. And Knewton Alta was $60 less. I think what I would say now is that because it was a pilot, we didn’t really understand fully how to implement Knewton Alta, and if we did it again the results would probably be different.

Why do I keep using it?

The cost was a lot less, and the students seem to be a lot happier. And, again, just using it has been a really positive experience.

I get unsolicited happy notes from students who are using the Knewton Alta. They’ll say things like, “Thank you so much. The adaptive thing takes a little bit of getting used to, but I really felt like it helped my learning and I really learned stuff. And the system is really good.

I never get that from MML. If I do get any comments about MML it’s usually, “It’s broken, help me fix it!” And so that’s been frustrating. Over the last couple of years (I don’t know if it’s the pandemic, or scaling, or what has happened) it just seems like their MML infrastructure hasn’t really kept up.

They seem to have a lot more technical difficulties, and I don’t have the same issues with Knewton Alta.

Student Feedback

We use a lot of MML in the math department, which means some students come in to my class with a MML background or mindset. They think, “I do the problems. I get them, it checks a box. I’m done. It’s 20 problems. That’s it.

Over the years I’ve been using Knewton Alta, I’ve gotten a lot better at explaining how it works to the students. I explain…that Knewton Alta is a little bit different. It adapts to the information that you give it. And now that I explain things, I don’t get as many questions or students complaining. If I do have a student who seems to be a little frustrated, it tends to be because they don’t know how many questions there are when they go into the system. And I always tell them there’s a recommended range up in the corner, and most of the time it’s 8 to 12.

Man at computerAnd I tell them if you go past 24 questions, you need to get a tutor, reach out to me, do something— because you don’t want to just spin your wheels. And most of them don’t. I get a lot of positive feedback, and what they say mostly is, “I really feel like I learned the material.”

One of the other things that I like, and I think students like it too, is the number of questions in Knewton Alta. In MML, you can put a review in there, but there are only so many questions, and then the review is over. But the review center in Knewton Alta is infinite, it keeps pulling questions from the test pool over and over again, forever. You can practice for as long as you want.

Students will ask when they should stop. I tell them, “When you’re getting 80% of the questions correct, then take the exam, take the certification. But if you’re not getting 80% correct, then you’re not ready and you need to keep going.” And I think students really like that aspect as well. And maybe that fits into the “I really have to learn it” mindset, as opposed to just figure out what the question is asking, check the right box, and move on.

How do you implement Knewton Alta?

I try to have students do something every day. So, there’s a due date and typically they’ll do three Knewton Alta assignments a week, sometimes two if they’re a little harder or longer. And then the week when we’re doing means, standard deviations, and Z scores, because it’s really just super basic calculations, they do four.

There is some level of instruction that I give them before they get started, whether it be a video or a Canvas page (which includes some classroom notes). I have 98 videos I’ve made for the stats class, and I can put a video in front of each of the lessons.

Students watch those things and then they’ll do the lesson. And then if they miss three or four questions in a row, it forces you to do the instruction again in Knewton Alta. If a student is struggling, I’ll remind them about the review instruction button. I tell them it might give them a slightly different point of view from what I’ve done in class, which might be helpful. And I know that Knewton Alta has made a change so now you can force the instruction first, before they get started. I don’t know how effective that is for most of my students. I think the original thought process behind how they set up Knewton Alta was the just-in-time instruction. Students typically dive in, get stuck and then try to figure it out.

So, I structured the class like that. Like, “Here’s some instruction. Here’s the homework. Go.”

I’ll get some students who will watch every minute of video twice, take lots of notes and then do the homework. But that’s the exception. The bell curve, the 68% in the middle, they’re just jumping in and when they get stuck, then they go back and read stuff until they feel like they’re not stuck. And then they’ll get stuck again. And then they’ll watch something. They go to, they get stuck, and it just I think that works for a lot of students.

In my math liberal arts class, I have fewer videos. And so I’m wondering about turning on the instruction beforehand for places where I haven’t made enough videos yet. So, that’s something that I’ve been thinking about so I’m glad that feature is there, but the original way that they set it up, like four years ago, I think that works for most students. And then the fact that it forces them to review instruction when they get four questions in a row wrong. Students need somebody to say, “Hey, this is how you’re supposed to do it!” And I think the students appreciate that because they’re not spinning their wheels forever.

I think that at the end of the day, you just have to meet students where they are. A lot of students just want to jump in. And so whether that’s good for them or not, I think you have to let them do it and then have a safety net underneath them.

Students come in at different levels. A lot of the students have never had stats, but then I have students who have had AP stats and they may remember some of that. So, they blow through the first module pretty easily. And there are students who are taking business calculus and their quantitative reasoning is at a level where it’s pretty quick for them to catch onto stuff. So, they don’t really struggle. And so for them, certainly not having to watch everything and read everything is helpful.

Is the adaptive nature of the program important to you?

Yes. Before the pandemic, they had a real-life Knerd camp up at New York. I got to go up there and talk to the data scientists about how the adaptive algorithm works, how they tried to make sure it never becomes a black box, and that there’s a person who has their hand on the wheel a little bit.

And it was a really good talk for me to understand how the algorithm works for the students. And it’s a big appeal for me. I really want the students to learn the material and develop some proficiency as opposed to “Do 20 problems, hit submit, and get 10 out of 10.” I really appreciate that aspect of Knewton Alta.

We’re allowed to adopt two textbooks for any one course on our campus. What initially drove the adoption of Knewton Alta as the second resource for statistics was the cost. And I think, being an advocate for it, that was a big driver. I think secondarily, the adaptive learning and the fact that it’s good software.

The Knewton Alta statistics course uses the Open Stax book. I make available on Canvas the PDF for that OpenStax book. But most of the students don’t read it. I have this really funny anecdote that I think summarizes textbooks for students. I asked a group of stat students in face-to-face class, “When do you read the stats book?” And this girl (who felt really comfortable) said, “When the professor sucks!”

I think that sums up the feeling of lots of students. If they don’t understand how the professor is teaching, that’s when they go to the textbook.

Most students would rather watch a video, read a short piece in Canvas, and try the homework, as opposed to read a chapter in a math textbook and try to follow that. And some of the textbooks are written really well, but still students’ level of motivation for reading is pretty low. But with the videos and the short bite-sized Canvas pages, they can read something, do an example, watch a video, solve a problem. That speaks to how millennials want to learn, expect to learn, how they learn.

That’s what they expect. And so that’s how I try to set it up. Knewton Alta does a pretty good job with that and there are a lot of videos and other resources. And the instruction that I’ve seen is pretty bite-sized.

Room for Improvement

My one “sad face” comment is that when I do the assessment, especially in this statistics class, a lot of the questions are still static. They’re not algorithmically generated. There are only a few of each learning objective to pick from. And if wanted to let students redo it, or if you wanted to pool questions, it just makes it a little harder. And so that is the one thing that is easier in MML; I can just grab a question from a learning objective, see the little red numbers that I know that are going to change for every single student, and I have less of a worry about academic integrity, students Googling stuff, that kind of thing. And in an age where I can do much less proctoring with my online students, that becomes a little bit bigger of a concern. It is the one thing that would make my life easier. Literally the only thing that they don’t do.

Knerd Story – Dina Yagodich, Frederick Community College

Dina Yagodich
Mathematics & Statistics
Frederick Community College

How long have you been using Knewton Alta?

I’ve been using it since fall 2019.

Why did you start using Knewton Alta?

I was teaching statistics for the first time. Two of my colleagues had started using Knewton Alta so I copied their course over and really liked the way it worked. In statistics we have students coming in at very different levels, and Knewton Alta helps get everybody to the same level.

For some of our classes we’ve also had experience using Aleks, which is also adaptive. But Aleks starts by assuming you’re at the bottom, and Knewton Alta starts by assuming you’re at a higher level.  A lot of students already had statistics in high school, so using Knewton Alta made the homework less frustrating for them because they can get through it pretty quickly, but students still get what they want or what they need.

Logo Description automatically generatedWe decided in the spring 2020 semester that everybody was going to use Knewton Alta in fall 2020. We had a handful of instructors using it as a pilot, and we were in the process of training everybody, when COVID hit. Looking back, it probably couldn’t have happened at a better point because Knewton Alta gave students more interactive help since nobody was in the classroom for the rest of the spring 2020 semester. And we had very few classes face-to-face in the fall of 2020, so it worked out well. And even our adjunct faculty didn’t have too much trouble getting used to it.
Now we’re using it mostly in college algebra, pre-calculus, and calculus one (as well as the liberal arts math class).

We’re able to come up with standard courses that everybody can just use. Some of us integrate Knewton Alta with Blackboard, some don’t. So far, we’re finding Alta is working very well.  It’s kind of shocking in two years to see how fast you can switch. A lot of it is cost driven. We’ve been trying to do open source, but for the homework piece it just wasn’t there for college algebra, statistics, the pre-calculus. So, we’re trying to go to the OpenStax textbook along with Knewton for homework.

Another thing we like about Alta, or Wiley, is that they have a policy that you have access until you pass. So, if somebody fails the class, all we have to do is ask our rep for new code, so that student doesn’t have to purchase the software again. That’s a big deal because not all of our students are successful.

How important is the adaptive nature of Knewton Alta in terms of influencing your decision to use it?

Very important for stats and pre-calculus because we know our students come in with such different prerequisite skills and an adaptive homework system lets students who know the stuff not spend too much time, and it allows students who really need extra time to get more homework. The big issue for all of us (me included) was getting rid of the control of which questions were asked. But I got over it quickly. So, I take control of my paper and pencil quizzes and my exams. I don’t use Knewton Alta for exams, but a lot of instructors use Alta for all the pieces.

I found that there was a stronger correlation between homework scores and exam scores with adaptive homework than there was with a non-adaptive system.

There are pieces of it that are still not perfect. But overall, I think it’s a better experience for students. It meets there where they’re at.

Student Feedback

After an exam, I have students answer anonymous survey questions. One of the questions was: “how has your experience with Knewton Alta, did it help you prepare for the test?”
Here are some of the student responses (from pre-calculus):

Then somebody else who used it last semester:

Calculus one students:

The class where we’re noticing the most issues is college algebra and supporting people coming in at the algebra one level.

In the spring (2022) we are going to try one class taught with Knewton Alta and a class taught with Aleks, because with that, starting at the top might not be the right mix for that specific class.

How do you support students using Knewton Alta?

We’re really good with explaining how to do the homework now. The first day of class, I tell them if you use this like normal homework where you just guess to try to see what the answer is, and then figure out what the answer is. You will fall into a pit of despair, and I don’t want that!

We make sure students know that instead, they should click on instruction when they need some help and that they don’t get penalized for doing that. So as long as the instructor really is clear about that, I think usually students quickly get the hang of how to use this software.

There are a couple kinds of questions where students have frustrations, but overall, I think students are getting good value for their money.

It can take a while to get some of the adjunct professors used to Knewton Alta homework. Some of them were concerned that Alta let students get to 100%, because they’re used to students getting around 87% on the homework. And we said yes, we actually want them to get to 100% mastery because then you will find that their test scores are also very good. The students who do it (the homework) are going to get 100% and you’re going to see that understanding carry through on exam.

When I look at student usage in Knewton Alta I look at the hours that they’re spending on homework and make sure that it’s a reasonable amount. If I see somebody who’s spending way too much time that’s when I’ll jump in. Or if somebody is zipping through them too quickly, I can ask them if they are looking things up to help them? Because that’s not going to work for exams!

How are you implementing Knewton Alta?

I have homework which opens on Saturday and is due every Friday. On Tuesday. I have paper and pencil quizzes based on the previous week’s material. I usually make the quizzes very tough because we spend the class period before going over any questions students have. So, my quizzes aren’t so much an assessment as they are a way of putting together all the different topics from the previous week. And students can see the kinds of formatting of questions that I’m going to ask on my exam.

The exams are paper, pencil, and written similarly to the quizzes. But they’re all based on the Knewton Alta homework. So, I go through Knewton Alta, before I write the test, to make sure I’m using similar vocabulary. The point of me asking students to do homework is so they do well on the assessments, not just to improve their grade.

Especially at the beginning of the semester, if people haven’t gotten to 100% in first week, I email them and let them know I extended their homework deadline an extra day. I let students know I want them to get to 100%. If somebody worked on the homework past due date and I can see that, I change the due date, so it matches their mastery.

So, a lot of my students get 100% on homework. I used to make homework worth 10%, now make it 30%. If I was back in a face-to-face classroom, I would probably make it 20%. I try to spread out the grade, so it’s not so test heavy. We don’t proctor our exams. And I figure if somebody who’s going to cheat with their way through the class, they’d have to pay somebody a lot of money to do all the Knewton Alta homework!

As the semester goes on, sometimes people only get to 87% mastery but that’s what is recorded. So over Thanksgiving, I’ll probably open up all the homework again and say, you can go back and improve your score on any of the sections to help students get ready for the final. Some will do that—although usually the ones who don’t need it, but I open it up to everyone.

Other feedback

One thing that could be improved is that it’s not super easy for me to hand off a class ‘as-is’ to an adjunct. We have gone through all the instruction for all the different objectives and copied all of the video links and embedded them into our Blackboard course. There’s a list of videos that students can watch ahead of time which helps, but it doesn’t help instructors to better know what’s going on and also to get a feel for what their students are seeing. Before, we could just look at the textbook and flip through it, but that doesn’t work anymore. Which is better for students, but instructors need more support.

So that’s the piece I think Wiley needs to improve on to make it easier for us to assign this to adjunct faculty without doing the prep work that we had to do.

They also have a lot of supplemental worksheets and PowerPoint slides, which I always give my students. Some love doing worksheets for extra practice, but it’s not easy to download them. They’re in different places and it’s not intuitive to just jump in and find them.

How is course content delivered?

I’m teaching a structured remote class, so I teach it more as a flipped classroom. I’ve taught calculus one before online, so I have my own calculus videos, which are specifically aligned to the textbook. So, students have my videos, my class notes, the OpenStax textbook, which they can link to online, or they can buy it for $26.

During my synchronous remote time we go over quiz questions, or if students have questions on homework. If I have notice in Knewton Alta that students are struggling on a topic, I’ll pull up those kinds of questions and we can do them together as a class. We’re finding at the community college that for many students structured remote is exactly what they need. They don’t want purely online, because they want some time when they know they have a professor in front of them.

Have you looked at relationship between Knewton Alta and grades?

I haven’t taught pre-calc in both formats. This semester, I’m going to compare how my Calc one students did in the Pearson homework and their grades, and then this semester with my Knewton Alta homework and their test grades.

So, I’m going to do a correlation to see if there is a stronger correlation between those two, because they would both be my classes. It will be nice to be able to compare two different classes and two different semesters with everything the same, except for the homework system.

What’s New in Knewton Alta – May 2020

Wiley is committed to a culture of continuous improvement with Knewton Alta, which is why we are always engaging instructors and students to gather feedback and optimize the learning experience moving forward.

As always, we are very cautious with our product releases to ensure minimal disruption during the semester, while continuing to provide an ever-improving experience for you and your students.

Below you will find information on enhancements made to Knewton Alta during the month of May. If you haven’t looked at Knewton Alta in a while, you’re missing something special.

Improved Student Context and Visibility into Assignment History

This update will help students (and instructors) to better understand their performance on an adaptive assignment and path to mastery, while providing transparency into the type of work that has been completed.

Learn more on how instructors can monitor student progress here.

Learn more about how students can track their own progress here.

Improved Course Copy Functionality

Getting started is now easier than ever!

Our team has been head-down looking for opportunities to make improvements to the course building experience in preparation for Fall. A number of usability improvements have shipped in the past month and we’re topping it off with a boost to copying courses. When copying a Knewton Alta course, instructors now have more options to configure their new course such as setting your institution or changing the dates as they are creating it so you can hit the ground running.

Doubling Down on Grading Accuracy!

A large percentage of Knewton Alta questions are free response, especially in our math courses, and we’ve just released a new update to our answer parsing service to provide better symbolic equivalence checking.

Students submit answers in a variety of complex — but equivalent — formats, and our new parsing engine leverages the active development community around sympy, a python-based computer algebra system, to better catch tricky cases where complex student answers may not appear to match the format that a subject matter expert would use.

Check out the Knewton Blog each month to see the improvements we’ve made to Knewton Alta.

Knewton launches altapass, an all-access pricing option, making alta even more affordable for students

With altapass, students can access multiple alta courses within a single subject area for $79.95; Knewton lowers price of a single-course alta subscription to $39.95

New pricing options available to students for Fall 2019

NEW YORK — Jan. 15, 2019 — Knewton, the world’s leader in AI-driven teaching and learning, today launched altapass, an all-access pricing offer for alta, the company’s adaptive learning courseware for U.S. higher education. With altapass, students can access multiple alta products across a single subject area for up to two years, and for unlimited use, for $79.95. Additionally, Knewton has reduced the price of a single-course alta subscription from $44 to $39.95.

By introducing the new pricing options, Knewton is making alta even more affordable and accessible, helping to put achievement within reach for the students who need it most.

Students wishing to purchase altapass for Fall 2019 may do so beginning Aug. 1 at Knewton.com. Students may still purchase access to a single alta course via monthly subscription for $9.95 per month.

At launch, altapass will be available across all 36 alta products in the following subject areas:

Knewton brings alta to scale with altapass

Knewton’s effort to make alta more accessible and affordable comes one year after the product’s successful introduction in the U.S. higher education market. Launched in January 2018, alta was used by instructors at more than 250 colleges and universities during the Fall 2018 term.

“It’s clear that we have something special with alta. Now, we’re making it even more affordable and accessible, so that the students who need alta the most can benefit from its impact on learning outcomes,” said Brian Kibby, CEO of Knewton. “We’ve turned the cost structure of our company into a competitive advantage — not just for Knewton, but for students looking for better results at an affordable price.”

“By keeping alta’s pricing simple and consistent across subject areas, we’re taking a lot of the mystery out of the cost of course materials for students. We’re also making alta more affordable for the high number of students who are using alta in more than one course in a single subject area,” said Heather Shelstad, Knewton’s VP of Marketing. “We’re giving students the power to decide which purchasing plan is right for them, and helping them save money no matter which option they choose.”

Knewton releases new insights into student usage, engagement and performance with alta

To provide fresh insight into how alta makes an impact on learning outcomes, Knewton’s data science team released a series of findings regarding student usage, engagement and performance during the Fall 2018 term. They include:

Knewton recently published the results of an independent study of alta’s effectiveness led by the Center for Research and Reform in Education at Johns Hopkins University. The study’s findings drew a link between alta and improved student performance across student ability levels, classrooms and institutions.

“Knewton is an outcomes company,” added Kibby. “While access and affordability represent a key part of alta’s value proposition, there’s nothing more important than its ability to deliver results for students and instructors. We’re going to keep challenging ourselves to set a new standard for transparency regarding those results.”

What’s new in alta for 2019

There are a lot of things that set Knewton apart as a company. One of the most important for our customers is our culture of continuous improvement.

Leaving well enough alone? It’s just not in our DNA.

That’s why, since we launched alta in 2018, we’ve been pounding the pavement, asking every student and instructor we can get in front of how we can make their alta experience better. Our goal: making an even bigger impact on learning outcomes by continually optimizing their alta journey.

Of course, improving the experience of students and instructors requires more than just listening. It takes action. That’s why, over the summer, we began releasing a series of features and enhancements that address the feedback we received.

Below, you can find a handy guide to all of the enhancements we’ve made to alta since launch. If you haven’t taken a look at alta over the past few months, perhaps it’s time for a fresh look.

General improvements for instructors

New functionality and product design makes alta more powerful — and useful — than ever.

Enhancements to alta’s LMS integration

Making learning more accessible is one of the key reasons why we built alta. By enhancing alta’s learning management system integration capabilities, we’re taking accessibility to new heights.

(Even better) Service and support

We’re setting a new standard for the customer experience by bringing service and support into the 21st century.

Testing and assessment enhancements

We’ve made a number of improvements to how instructors prepare and deliver quizzes and exams in alta — and given students new ways to prepare for their exams.

Content enhancements

We’ve made a number improvements to alta’s content with the goal of providing a deeper, more flexible learning experience that leads to better outcomes.

If you have questions about any of alta’s newest features, you can always reach out to us at support@knewton.com. A Knerd will get in touch with you ASAP to walk you through what’s new and answer any questions you may have.

Wishing everyone in our Knerd community lots of success in the semester ahead!

Johns Hopkins University analysis draws link between Knewton’s alta and improved student outcomes

Earlier this year, Knewton presented an efficacy analysis of alta, Knewton’s adaptive learning courseware for higher education, developed by our data science team. From our perspective, the results of our internal analysis strongly suggested a causal link between alta and improved student performance.

Because even the best-intentioned researchers can introduce unconscious biases when making analytical choices, we knew that our in-house analysis could only be part of the story of alta’s effectiveness. As a data science team comprised largely of academics, we have a deep appreciation for the value of independent reproduction of scientific results.

To get a fresh, unbiased perspective on alta’s impact, we shared our fully anonymized Fall 2017 data with the Center for Research and Reform in Education at Johns Hopkins University (JHU). More specifically, we asked JHU to assess the impact of demonstrating concept proficiency by completing alta assignments — as well as alta usage in general — on student outcomes like quiz and test scores, future assignment completion and retention.

Key findings

To gain an understanding of the study’s key findings and conclusions, we invite you to read JHU’s complete analysis of alta’s impact on learning outcomes.

Knewton’s commitment to impact and transparency

These analyses — performed by different teams, in different ways, and across different time periods — represent both a fundamental piece of scientific research and are key to our efforts toward transparency and continuous, data-driven improvement.

Now, the conversation around alta’s impact must expand to include things like direct feedback from instructors, student surveys, user research, and case studies from a variety of classroom settings. There’s a lot of work to be done!

While conducting a conversation that, by design, never ends is in some ways daunting, it keeps us connected to the experiences and results of our users. And from that perspective alone, this endeavor has been a valuable one.

Saving Money with Beautiful Soup and Hashing

Back in the summer of 2017, Knewton’s alta images had an expensive problem. Whether the images were of parabolas, molecules, or supply and demand curves, they were all missing two important things: alt text and long descriptions.

Why do our images need alt text and long descriptions?

As part of being ADA Compliant, alta’s images need alt text and long descriptions to make the images accessible to screen readers. This is important so that our blind or visually impaired students can use assistive technology like screen readers to interact with our visual content.

An accessible image needs alternative text (alt text) and possibly also a long description, which the screen reader can read out to the user. Alternative text is typically brief (we limit ours to 255 characters) and should always be included on accessible images. Long descriptions are used to describe more complex images, like a detailed diagram.

Alt text and long descriptions are added to images via HTML attributes. Here’s an example image:

Fluffy gray cat belonging to a coworker.

Let’s say that the HTML tag for this image is:

<img src="cat.jpg" alt="Fluffy gray cat" longdesc="cat.html"/>

By including alt and longdesc attributes in the image tag, screen readers can read out “Fluffy gray cat” and the contents of cat.html to the user.

The idea is that a visually impaired student can get all the information they need about an image via its alt text and long description. Take this image for example:

This figure shows two curves. The first curve is marked in blue and passes through the points (negative 1, 2), (0, 1), and (1, 1 over 2). The second curve is marked in red and passes through the points (negative 1, 3), (0, 1), and (1, 1 over 3). Attribution: Image by OpenStax Intermediate Algebra is licensed under Creative Commons Attribution 4.0 International License. Download for free here.

This image’s alt text is very specific:

This figure shows two curves. The first curve is marked in blue and passes through the points (negative 1, 2), (0, 1), and (1, 1 over 2). The second curve is marked in red and passes through the points (negative 1, 3), (0, 1), and (1, 1 over 3).

(Note: Due to limitations there are no alt text and long descriptions for images in this blog post, but we’ve included image captions as an alternative for screen readers.)

Unsurprisingly, writing alt text and long descriptions for thousands of images gets expensive fast. If there were only a way for us to not start from scratch…

Scrape OpenStax, save money

A good chunk of Knewton’s content is curated from the open source OpenStax textbooks.

Greg, our Senior Manager of Content, was staring at OpenStax textbooks online — as senior managers of content are wont to do — when it hit him: OpenStax includes alt text with their images. We could scrape out alt text from OpenStax and match them with the images used in our courses! To Greg, this sounded like an ideal Hack Day project.

Knewton holds “Hack Days” a few times a year, in which Knerds (the Knewton employees) get to work on whatever project we wanted. For the August 2017 Hack Day, Greg and I teamed up to make his OpenStax-scraping, money-saving dream happen. Our solution had two steps:

  1. Scrape OpenStax textbooks for images and their associated alt text.
  2. Associate the scraped alt text with their images in our content management system’s database.

Finding images with Beautiful Soup

Conveniently, each OpenStax textbook has a downloadable zip, containing all of the textbook’s HTML and image files.

I wrote a Python script to walk through all the directories of the unzipped book, looking for HTML files. Then it became a straightforward application of Beautiful Soup, a popular Python HTML parser.

>>> import codecs
>>> from bs4 import BeautifulSoup
# file_path is an HTML file's path
>>> page = codecs.open(file_path, 'r', 'utf-8')
>>> soup = BeautifulSoup(page.read(), 'html.parser')
>>> image_tags = soup.find_all('img')

Notice how simple it is to find all img tags. Once I got the HTML content, I just needed two lines:

  1. Create a “soup” using the HTML.
  2. Use the soup’s find_all method.

For each tag in image_tags, Beautiful Soup makes it easy to extract the altand src attributes.

>>> tag
<img alt="Cute puppy" src="puppy.jpg"/>
>>> tag['alt']
'Cute puppy'
>>> tag['src']

The alt attribute contains, well, the alt text. The src attribute contains the file path to the image, which will be used in the next section.

Matching up images with hashing

Now that I have scraped all the alt text in an OpenStax book, I need to match them up with images in our content management system (CMS). This is where hashing comes in: two identical images have the same hash, while two different images have different hashes. If an alt text’s corresponding OpenStax image has a hash that matches the hash of an image in our CMS, then we can apply that alt text to that image in the CMS.

In Python, hashing an image is a matter of using the image’s file path (which was scraped with Beautiful Soup) to read the image’s bytes, and then feeding the bytes into one of Python’s built-in hashing algorithms.

>>> import hashlib
>>> with open(image_path, 'rb') as image:
...     image_bytes = image.read()
>>> hashlib.sha1(image_bytes).hexdigest()

That hash there is of this image below. Try hashing it yourself with the same hashing algorithm and you should see the same result. (This is assuming that the image’s compression has not changed since this post’s publishing.)

Cartoon illustration of a laptop.

What about the long descriptions?

You’ll notice that I glossed over long descriptions in the last few sections. Did we scrape for that at all? Yes and no. OpenStax image tags do not have longdesc attributes. However, we did end up repurposing many OpenStax alt texts as long descriptions because they were so detailed and, well, long.

We also did not use every OpenStax alt text verbatim, as our team of subject matter experts sometimes improved upon them or shortened them to fit within our 255 character limit.

How much money did we save?

We haven’t calculated exactly how much money we saved (we’ve been busy building out alta instead 😉). But if we do a back-of-the-envelope calculation:

Therefore, thousands of images times tens of dollars per image equals tens of thousands of dollars saved! Not too shabby for a hack day project that I coded in a day.

What’s new in alta for Fall 2018

Ahh, the dawn of a new academic year. When the slate is wiped clean and we are given a new opportunity to help students achieve their goals.

There are new courses to be taught. Fresh faces in your classroom. Great new alta features just waiting to be used. With that in mind, we thought we’d provide a handy round-up of everything that’s new in alta for the Fall 2018 term.

If you have any questions, please reach out to us — we’ll have a knerd get in touch with you ASAP to walk you through what’s new and answer any questions you may have.

Now, on to the enhancements!

In-app support via chat

Ever wanted support to feel a little less like sending a message in a bottle and a little more like a two-way conversation with a friend? Check out alta’s new support chat feature (that we think represents the most advanced support infrastructure in the industry):

What does all of this add up to? More immediate responses, more proactive support, and an elegant support experience that’s on par with what you’ve come to expect from alta.

Courses and sections

Have you ever wished that you could manage multiple sections of your alta course — while keeping all the core elements consistent?

With our Courses and Sections update, you can create sections based on your original alta course. Alta will carry over the original course learning objectives, content, and settings to the new sections. (Instructors teaching the new sections may add coursework or modify due dates for homework, but don’t worry: the important stuff will stay the same.)

Courses and Sections is kind of a big deal, so we’ve dedicated an entire blog post to it. If you have questions or would like to learn more, we recommend checking it out.

Improved adaptivity in non-quantitative courses

After gathering feedback from instructors and analyzing how students performed using alta in their Economics courses, we recognized an opportunity to improve the student learning experience.

Over the summer, we released updates to how our adaptive engine measures student progress toward mastery in non-quantitative courses. We’ve also made the learning objectives in these courses more granular to allow us to help students gain proficiency with greater precision.

Desmos graphing questions in for Math, Econ and Stats products

We’re always seeking new ways to present content and assessment in order to help students achieve mastery. That’s why we’re proud to announce that alta products in Math, Economics and Statistics will feature graphing questions from Desmos.

Desmos provides a powerful platform for presenting assessment questions in the form of a graph. The flexibility of Desmos allows us to deliver higher-order comprehension questions and provide better support for graphing questions.

Expect between 25-50% of content within each alta product in these subject areas to feature Desmos.

…and content updates across the board

While we’re continually refining alta’s content to make sure that it’s effective in helping to improve learning outcomes, over the summer, we completed a front-to-back sweep of our content to make sure that everything is in tip-top shape.

Think of it as a little “summer cleaning,” if you will.

There are lots of things to love about alta, and chief among them is the fact alta is always getting better. You can expect to hear more from us soon about the next set of exciting enhancements to alta.

In the meantime, we wish everyone the best of luck with your alta journey this semester!