Wednesday, April 27, 2016

A Situative "Roadmap" for Synergy in Motivating Family STEM Engagement

By Dan Hickey
This is the third post about my collaboration with Iridescent Inc., a science education non-profit in LA. This new post describes how a key assumption in newer "situative" theories of motivation can resolve the tensions between prior empiricist and constructivist approaches. When combined with Design Based Research methods, this assumption can result in a coherent "roadmap" towards synergy across the three approaches. I contend that such a roadmap can help Iridescent and other informal STEM innovators find a route that takes them from current levels of engagement to much higher levels of engagement, both in terms of quantity and quality.

This post could use some work and some trimming but I need to get it up for my class and colleagues and get on to other things. Will try to clean it up soon

In my first post in this series, I described Iridescent's two primary programs and my evaluation of one of the programs involving Engineering Design Challenges at their Curiosity Machine website. Iridescent staff hosted weekly Curiosity Classes for children and parents and encouraged families to continue working on and submitting design challenges from home. We found that only about a third of the families reported submitting from home, and fewer still began working on new challenges from home.

In my second post in this series, I elaborated on three different approaches that Iridescent might take to increase family engagement with the Design Challenges during, between, and after the weekly Curiosity Classes. These approaches follow from three "grand theories" of learning that many educational researchers embrace. While different researchers use different labels, I will call them empiricism, constructivism, and situativity. The second post showed how the first two strategies were "orthogonal." Specifically, empiricist and constructivist perspectives prescribe very different guidelines for motivating engagement. These approaches are appealing because they provide educators and innovators a clear path forward, and promise to work in most any learning context. Unfortunately, these two paths go in opposite directions. This is because empiricist perspective characterize engagement in terms of the ways individuals behave, while constructivist perspectives characterize engagement in terms of the way typical humans think and process information. Thus, the incentives, rewards, and competition suggested by empiricist perspectives are likely to undermine the intrinsically-motivated learning that constructivist perspectives assume is needed to support meaningful cognitive engagement in the short term and create self-determined learners in the long term. 

The Search for Synergy
I have been fascinated by this debate for my entire career. In articles published in 1997 and 2003, I explored whether newer situative approaches to motivation might offer a way around this tension and reveal new sources of synergy. To reiterate, situative approaches characterize engagement as engaged participation. This means that engagement is characterized as increasingly successful participation in the social and cultural practices that define "communities of practice." In the second post, I summarized the implications of situative perspectives for helping Iridescence motivate extended family engagement. This revealed that situative perspectives do not lead to the more straightforward and general paths that follow naturally from empiricist and constructivist perspectives.

In this new post, I explore how a specific aspect of situativity and "design-based" research methods from the Learning Sciences can provide a practical road map that Iridescent could use to find the synergy needed to draw from all three approaches. In addition to supporting their own programs and families, I believe that following this roadmap can also result in design principles that should be useful for helping motivate engagement in museums, afterschool programs, websites, and and other entities that support informal science education.

Combining Individual and Social Activity
Before I can create this roadmap for synergy across motivational practice, I have to make a big decision. It turns out that there are two very different ways of combining the activity of individuals with the activity of social groups. To extend the roadmap metaphor, if we do not address this issue from the outset, it will be difficult or impossible to follow the map. More specifically, the evidence of different types of engagement will point in different directions. So guidance is needed to inform smaller decisions along the way.

For researchers working in the empiricist and constructivist perspectives, reconciling individual and social activity is not been particularly problematic. As I argued in my 2003 paper, researchers in these traditions typically embrace an "aggregative" reconciliation where social activity is simply characterized using aggregated notions and evidence from individual activity. Hence, behavioral theorists Like Sigrid Glenn working in the empiricist tradition characterize social activity using "meta-contingenices" to represent the aggregated effects of the incentives for individual activity. In contrast, researchers working in the constructivist tradition like Albert Bandura characterize social activity as "collective efficacy" which represents the sum of the efficacy of individuals. Unfortunately, this actually exacerbates the problem of having to choose one perspective over the other. Obviously, the two resulting characterizations of social activity are incompatible. And from a situative perspective they are both highly incomplete. As such, an aggregative reconciliation works against synergy.

A very different kind of reconciliation follows from under-appreciated assumption about situative theories. Jim Greeno introduced this issue in a 1993 chapter and then further it in his his 1998 presidential address to the American Psychological Association. His so-called "situative synthesis" argued that a situative approach treats social and cultural activity as "primary" and treats both the behavior of individuals (in empiricist perspectives) and the way that humans typically process information (in constructivist perspectives) as "secondary." In Iridescent's' case, this means that they can treat the way that children and families respond to motivational practice that follow from empiricist perspectives and rationalist perspectives as "special cases" of engaged participation.

What this means in practice is innovators are free to try out different motivational practices, while looking at the consequences of those practices on the entire range of engagement. Importantly, this approach provides guidance for resolving the many questions that will come up along the way when the evidence is unclear or contradictory.  Returning to the roadmap metaphor, the situative synthesis helps guide the decision making process when several different routes towards the same goal are apparent. By treating social and cultural practices as primary, innovators can focus first on engaged social participation and then second on individual behavior and cognition.

An Example of the Situative Synthesis with a Leaderboard
As I explored in the second post, many working in the empiricist tradition would suggest that Iridescent explore some sort of competition or gamification to motivate family engagement. Even though Skinner and other strict behaviorists argued against competition, it is an increasingly popular strategy for motivating engagement, particularly with the rise of market-oriented educational reforms. Indeed, some sort of leaderboard for each Design Challenge is one of the motivational strategies that Iridescent has considered. Meanwhile, most working in the constructivist tradition would vehemently oppose competition or gamification. An alternative approach would first gather some baseline evidence of all three forms of engagement. The team would then design some sort of leaderboard that they believe is most likely to motivate desirable forms of engagement without causing negative consequences. Then then introduce the new practice and see what happens. Specifically they would look at how the leaderboard works and begin gathering evidence of engagement.

After the first implementation, it is certain that additional refinements would be needed to make a leaderboard work in this context as had been envisioned. These refinements would be informed by further consideration of the suggestions from each of the three perspectives (summarized on the previous post). As such, it makes sense to postpone any formal study of engagement until the innovation was working well. In the meantime, informal assessments of the behavior, intrinsically motivated learning, and engaged participation can be made along the way to help inform those refinements. In doing so, the larger project learns how to gather valid evidence of engagement. This prepares the project for the formal evaluation needed to validate the innovation and the program as whole.

As long as the team keeps track of the logic behind the design of the leaderboard, the team is then nicely poised to examine the engagement that results, make refinements according, and try again. In doing so they can develop motivational design principles that will be support coherence for their team going forward. Importantly, this will also result in motivational design principles that can be used by other innovators facing similar challenges. This principled iterative refinement is a signature research method for the Learning Sciences known as Design Based Research. DBR dates back to the early 1990s when Ann Brown began conducting "design experiments." DBR has evolved significantly over the years and has been somewhat controversial (because of concerns over generalizability). From my perspective, the debate has been settled by the specific reference to DBR in recently educational RFPs in major federal agencies including the National Science Foundation (particularly the Cyberlearning initiative) and Institute of Museum and Library Services (particularly the STEM Expert Facilitation of Family Learning in Libraries and Museums initiative)

The Challenges of Assessing Engagement
As introduced in the previous post, each type of engagement is typically assessed with a different method and studied using a different unit of analysis. Each method and unit has advantages and disadvantages; synergy means taking advantages of the strengths and minimizing the weakness of each, as follows:

Measuring behavioral engagement is rather straightforward: In our example, Iridescent would examine whether children and families completed, started, and submitted more Design Challenges from home after the leaderboard was introduced. They might also look for other behavioral indicators of potential negative consequences, such as a drop off in engagement as the leader's score became impossibly high for most families to accomplish. Because the unit of analysis is typically an individual person, behavioral engagement can be studies with relatively few participant. Thus, it might be possible to introduce the leaderboard in the middle of a program and compare engagement in each family in the first half with the second half.

The measurement of behavior can be better understood using the metaphor that follows from the first of three "worldviews" advanced by the philosopher Karl Popper. These worldviews are consistent with our three "grand theories" of learning. Empiricist perspectives are consistent with Poppers "mechanistic" worldview. This leads to the metaphor of the learner as a machine. When designing machines, inventors and engineers can make changes to a single machine and see what happens. The results are typically easily measurable and often speak for themselves.

Assessing self-determination is more challenging. This is because a given behavior might be the result of the range of forces arrayed along this table (borrowed from Learning Snippets blog). Returning back to the example from the previous blog, imagine if Iridescent offered families $20 credit at Amazon for every Design Challenge they submitted. If there were some sort of criteria and evaluation of quality, it is quite possible that the incentivized submissions would appear reasonably similar to the ones that families submitted without the incentive. However, if we administered a survey self-determination (examples here, all based on the table below) we would almost certainly find that the cognitive engagement in the incentivized families falls to the left of cognitive engagement in the non-incentivized families. Indeed, it is easy to imagine some children in the incentivized families reporting "amotivated" engagement (all the way to the left of the table) as they crank out yet another Design Challenge for their share of the incentive. Similarly, it is difficult to imagine children in the non-incentivized families demonstrating external or introjected regulation (because they would just say they did not want to do it and in most families that would be fine).

This example illustrates why self-report assessment dominates studies of self-determination. This also illustrates why the typical unit of analysis is groups of individuals. Constructivist perspectives are consistent with Popper's second "organismic" worldview. This leads to the metaphor of learners as growing plants. Comparing self-determination under different conditions is a lot like comparing the effectiveness of two fertilizers. The agronomist applies the treatment to two similar plots of plants and then compares average yield of the two plots.

Following this example, our hypothesis about incentives could be tested by having two groups of families complete design challenges with and without incentives, and having parents and children complete an online self-report survey with each challenge. Unfortunately, such a study would take a lot of time and require enough families (i.e., 30 per group) to allow statistical tests to distinguish between random variation and systematic differences. Particularly with smaller samples, random assignment and other procedures are needed to control for potential confounds. Furthermore, by the time the data is gathered and analyzed, it might be too late to act on the results. This makes self-report measures difficult to use in the kind of iterative design-based refinements described above. This is one of the reasons I stopped using self-report measures in my own research once I completed my dissertation in 1995. However, I returned to self-report surveys in my studies of the Quest Atlantis educational videogame once I figured out they could be reframed using the situative synthesis as describe below

Observing engaged participation is both harder and easier. Because participation and the resulting identities are highly contextual, it impossible to measure or assess them accurately. They can only be understood using interpretive methods like discourse analysis that take into account the specific social and cultural context in which that participation takes place and where identities are negotiated. This requires a lot of very training to do well. I know this because I was not trained in this manner and so I really struggle to do it well. The reason it is so difficult is because the unit of analysis in of engaged participation is activity in context. Situative theories are consistent with Popper's third worldview, contextualism. This worldview of a historical event as an organizing metaphor for learners. There is no "accurate" or even "complete" understanding of a given historical event. Rather, one's interpretation of an event must take into account the context in which the event occurred. Perhaps more importantly, a "full" interpretation of a historical event must also take into account the context from which it is being interpreted.

Fortunately, in other ways, it can be quite easy to observe engaged participation. To reiterate from the first post, we are concerned here with disciplinary engagement. Specifically we are concerned with whether or not children and parents are participating successfully (and with increasing success) in the engineering design practices (curiosity, creativity, and persistence) at the core of Iridescent's mission. There are lots of indicators of this engagement. Persistence is probably the easiest, as indicated by the number of times a child or a family redesign their device to meet and exceed the specific goal of the challenges. All other things being equal, more persistence = more persistence. Similarly, creativity is displayed by the use of novel solutions that were not presented or suggested in the instructional materials. Thus, more creativity = more creativity. The practice of curiosity may might simply be indicated by the number of Challenges the family pursues at home. One idea to go further might add links to additional disciplinary resources such as more advanced video or devices. Families that explore those links after submitting a challenge (and perhaps while waiting for feedback from the online mentor) are clearly practicing curiosity more than families who do not.

As elaborated in the first post, the other key aspect of disciplinary engagement concerns disciplinary concepts. This is the stuff that disciplinary experts know, independent of context. In the design challenges, these are concepts like load and tension that are introduced in the inspirational and instructional videos. Engaged participation is about using disciplinary concepts to participate more successfully in disciplinary practices. This what I observed in my first post that I watched when I visited one of the prior family evening program. I watched a dad gently help use help is daughters use the concepts of thrust and angle as they persistently experimented with the number and direction of exhaust straws in the Spinning Machine design challenge. While the girls were "discovering" these concepts in their devices, they were learning how to use them to participate more successfully in the engineering design practices. Importantly, they were also learning the technically correct labels for those concepts.

A key point about engaged participation is that experienced educators and designers (who are knowledgeable about a particularly educational context) are mostly likely to "know it when they see it." This is because they are in the best position to consider all of the contextual factors that influence any interpreted. This knowledge is needed to interpret observations with enough confidence to inform curricular designs. For example, a particular level of engaged participation in an advantaged family with professional parents is not nearly as convincing evidence of success as that same level of engagement in a disadvantaged single-parent family. Furthermore, the same incentive is likely to have a different impact on the advantaged family than a disadvantaged family. Focusing primarily on such contextual observations is paramount in designing effective curricular for particular contexts. I believe that this is what Iridescent has been doing all along, and explains why their programs are already so engaging for their targeted learners.

Applying the Situative Synthesis to Iridescent's Design Challenges
To reiterate, the situative synthesis treats individual behavior and individual cognition as "special cases" of engaged participation. This means that further refinements of the Design Challenges and the Curiosity Machine should focus primarily on engaged participation and only secondarily on behavioral and cognitive engagement. This has several big advantages. The first is that it allows innovators to use engaged participation to "settle" the debates. Consider for example, if a leaderboard increases measured behavioral engagement while decreasing assessed cognitive engagement. Given the prior research on incentives, this actually seems quite likely. If the introduction of the leaderboard clearly increases engaged participation as well, this would presumably trump a small decrease in cognitive engagement.

Perhaps the biggest advantage of the situative synthesis is that it overcomes the methodological equivalent of "teaching to the test." When one targets behavioral or cognitive engagement directly, it is hard to know if the resulting behavior or cognition will "transfer" to other settings and contexts. This is particularly an issue with behavioral interventions such as incentives. Hundreds of studies have shown that free-choice engagement is diminished when incentives are withdrawn. Going back to our example $20 incentive for each incentive, I am convinced by the prior research is quite convincing that we would see a dramatic increase in behavioral engagement when the incentive was available and an equally dramatic decrease when the incentive was withdrawn. While not as dramatic, a similar "test-prep" phenomenon is possible with practices for supporting self-determination and intrinsic motivation. Providing optimal challenge, offering useful performance feedback, and helping learners related to others are all likely strategies for fostering self-determination and intrinsic motivation during the design challenges. Whether or not these desirable forms of cognitive engagement generalize to other (perhaps less supportive) STEM opportunities is far less certain.

While proponents of empiricist and constructivist interventions have conducted studies showing generalizability of increases in behavioral and cognitive engagement. But these studies were carried out in controlled laboratory contexts and a find them unconvincing for effects in more particular contexts, More importantly, I am convinced that evidence of behavioral and cognitive engagement that follows from interventions targeting engaged participation are by definition more likely to transfer to other contexts. This is because these changes are indirect "residual" effects of the intervention. In other words, evidence of behavioral or cognitive engagement from situative refinements is itself evidence of generalizability and transfer. 

An Example of the Situative Synthesis with Educational Videogames
As I mentioned above, I stopped using self-report assessments of motivation after I completed my dissertation. My evaluation of The Adventures of Jasper Woodbury math problem solving videos uncovered some interesting effects on self-reported interest and intrinsic motivation. Yet I was left frustrated that I could not do anything with those results to improve Jasper Woodbury or the way teachers used it. From 1995 to 2005, I explored the situative synthesis described above as it related to educational assessment in studies of educational multimedia (including the GenScope genetics program and three programs from the NASA-funded Classroom of the Future). From 2005 to 2010 I extended that program of research to assessment studies of the Quest Atlantis educational video game with Sasha Barab from 2005-2010.

In 2009 and 2010, I extended the situative assessment research in Quest Atlantis to include a quasi-experimental study of incentives and their impact on engaged participation and cognitive and behavioral engagement. Studies carried out with Eun Ju Kwon and Michael Filsecker compared 100 sixth-graders playing two different versions of the Quest Atlantis Taiga game. The game took place in a virtual national park where students served as apprentice rangers examining water quality for the park ranger (played by their teacher). The inquiry challenge in this game was figuring who and what was responsible for declining fish stocks, One version of the game included a public "leaderboard" that tracked players progress through the game stages. The leaderboard also displayed status according to the ranger's evaluations of the "field reports" that players submitted at each stage. The incentivized game also included digital badges awarded for higher levels of player status; the badges promised the earner special powers in the game. The other version of the game did away with both the leader board and the badges and instead appealed to intrinsic forms of motivation for excelling in the game (i.e., satisfying curiosity, saving the park, etc., consistent with Malone and Lepper 1987 guidelines)

As reported in the 2014 paper with Michael, we studied engaged participation of the students in their role of apprentice to the park ranger by coding the content of each student's field reports. Specifically, we examined (a) how many of a dozen targeted disciplinary concepts students used and (b) whether they used those concepts appropriately. It turned out that players in the incentivized condition used more of the concepts and did so more accurately. We also studied self-reported intrinsic motivation as students progressed through each of the four game stages, and studied changes in personal interest in scientific inquiry and water quality with pre and post surveys. Students in the incentivized condition reported higher levels of intrinsic motivation and relatively larger gains in personal interest than the non-incentivized students. Some, but not all of the difference were statistically significant. Unfortunately, a software problem precluded a planned comparison of behavioral engagement by looking at total time-on-task and logins from home. Given that all of the prior research suggested increased behavioral engagement in the incentivized condition, we were able to convince the reviewers that that we had sufficiently searched for the hypothesized negative consequences of the incentive practices and did not find them.

Naturalistic vs. Interventionist Research Goals
One of the big takeaways for me from the Quest Atlantis studies was the fundamental tension between naturalistic studies of the way the world is and interventionist studies focused on changing the world. It took two annual cycles with significant federal funding to refine the experimental design and instrumentation. For example, we struggled to maximize the motivational appeal of the incentives without confounding the experimental design. Case in point, one of the most motivating features of the in-game badges was giving players new tools that would help them solve the STEM challenge. But giving those players an actual advantage would have introduced a serious confound in the experimental design. We also found the peer  reviewers were reluctant to allow us to include interactions in the changes of interest from pre-test to the post-test that were on the margins of statistical significance, even though all of the interactions pointed to increased (or smaller decreases) in interest in the incentivized condition. More generally, many of the resources that might have gone into further refining the incentive practice using informal indicators of engagement were consumed by the laborious efforts to more formally observe, assess, and measure engagement.

In retrospect, I view the challenges I encountered as particularly worrisome for the field. This is because I believe I was well-prepared to deal with them. I had encountered similar tensions with the situative assessment studies, and had been engaged in a long running debate with the Educational Psychologist Richard Walker about the importance of "what is" versus "what might be" in terms of situative theories of motivation. In a number of papers, Richard took issue with my argument for the pragmatic value of locating goals and values primarily in the social context and only secondarily as internalized by individuals. Ultimately it is a rather philosophical difference. I agree with Richard that goals and values are transformed as they are internalized and then again as they are externalized in achievement contexts. My point was simply that we can assume that those transformations are carried out using socially constructed knowledge to justify focusing primarily (if not entirely) on situated models of engagement when attempting to transform educational practices.

This tension between naturalistic experiments & observations and interventionist design-based research is not going away any time soon. But as mentioned above, many federal agencies are making explicit reference to DBR in their requests for educational research proposals. Particularly promising is the definition of DBIR (design-based implementation research) by Bill Penuel and Barry Fishman and colleagues that extend these ideas to broader systemic transformation. What is emerging across the Learning Sciences and beyond are numerous "road maps" that include design principles that appear more productive for engaging and assessing learning than what came before.

The Challenge of Program Evaluation
Ultimately the challenge may lie in whether or not the field can define credible models for evaluating informal and formal educational programs and curricula. Most funding agencies, and particularly federal agencies, are still under enormous pressure to eventually "prove" that interventions "work" in rigorous randomized clinical trials. Such studies are very expensive to carry out. True experimental design also present a fundamental challenge to generalizability. Very few schools are led and structured in ways that make it possible to randomly assign students to experimental and control conditions. And in many cases, the most appropriate control conditions requires withholding an obviously valuable treatment (such as feedback).

A 2003 paper with Steve Zuiker argued that conventional "objective" approaches to program evaluations are very problematic, and particularly so when carried out with by external partners. They tend not to provide evidence and information that is useful for advancing projects and informing the field. Rather, we argued that program evaluation should be a natural extension of the design-based iterative refinements of educational innovations. Specifically we argued that what many innovators to can be characterized as design-based iterative iterative refinements that align immediate and close engagement with proximal indicators of engagement and learning. The impact of those refinements can then be evaluated objectively with distal (and possibly "remote") indicators of engagement and learning. I am still trying to figure out just how this would map to Iridescent's Family Science program. Clearly the completion and submission of Design Challenges from home represents a distal indicator of engagement. While remote-level indicators of engagement are hard to obtain in informal settings, it seems that continued work on Design Challenges after the Family Science program is over might work. At proximal, distal, and remote levels, indicators of transferable learning can likely be obtained by assessing the amount and precision with which families use the disciplinary concepts in their description of their devices and their reflections on their Design Challenges. Having these levels of indicators can provide the objectivity necessary to validate claims regarding efficacy, regardless of whether one is working with external program evaluators.

In closing, I will mention new alternative that looks work exploring is is the Value Creation framework that was recently advanced be Etienne Wenger and his wife Beverly Trayner. Given that is an explicit program evaluation assessment framework that builds directly on Wenger's foundational work on learning communities, it might have quite a bit to offer.  I hope to explore this potential in a future post,

Bandura, A. (2000). Exercise of human agency through collective efficacy. Current directions in psychological science, 9(3), 75-78.

Glenn, S. S. (1988). Contingencies and metacontingencies: Toward a synthesis of behavior analysis and cultural materialism. The Behavior Analyst, 11(2), 161.

Greeno, J. G., Smith, D. R., & Moore, J. L. (1992). Transfer of situated learning. In D. Detterman & R. J. Sternberg (Eds.), Transfer on trial: Intelligence, cognition, and instruction (pp. 99-167). Norwood, NJ: Ablex.

Greeno, J. G. (1998). The situativity of knowing, learning, and research. American Psychologist, 53(1), 5-25

Filsecker, M., & Hickey, D. T. (2014). A multilevel analysis of the effects of external rewards on elementary students' motivation, engagement and learning in an educational game. Computers & Education, 75, 136-148.

Hickey, D. T. (2003). Engaged participation versus marginal nonparticipation: A stridently sociocultural approach to acheivement motivation. The Elementary School Journal, 401-429.

Hickey, D. T., & Zuiker, S. J. (2003). A new perspective for evaluating innovative science programs. Science Education, 87(4), 539-563.

Malone, T. & Lepper, M. (1987). Making learning fun: A taxonomy of intrinsic motivations of learning. In R. E. Snow and M. J. Farr (Eds.) Aptitude, learning, and instruction: Vol. 3. Cognition and affective process analyses (pp. 223-253). Hillsdale, NJ: Lawrence Erlbaum.

Penuel, W. R., Fishman, B. J., Cheng, B. H., & Sabelli, N. (2011). Organizing research and development at the intersection of learning, implementation, and design. Educational Researcher, 40(7), 331-337.

Wenger, E., Trayner, B., & de Laat, M. (2011). Promoting and assessing value creation in communities and networks: A conceptual framework. The Netherlands: Ruud de Moor Centrum.

No comments:

Post a Comment