Student perceptions of laboratory classroom activities and experimental physics practice
Dimitri R. Dounas-Frazer, Kimme S. Johnson, Soojin E. Park, Jacob T. Stanley, H. J. Lewandowski
aa r X i v : . [ phy s i c s . e d - ph ] A ug Student perceptions of laboratory classroom activities and experimental physics practice
Dimitri R. Dounas-Frazer
Department of Physics and Astronomy, Western Washington University, Bellingham, WA 98225, USA andSMATE, Western Washington University, Bellingham, WA 98225, USA
Kimme S. Johnson
Woodring College of Education, Western Washington University, Bellingham, WA 98225, USA
Soojin E. Park
Department of Anthropology, Western Washington University, Bellingham, WA 98225, USA andWoodring College of Education, Western Washington University, Bellingham, WA 98225, USA
Jacob T. Stanley
BioFrontiers Institute, University of Colorado Boulder, Boulder, CO 80309, USA
H. J. Lewandowski
Department of Physics, University of Colorado Boulder, Boulder, CO 80309, USA andJILA, National Institute of Standards and Technology and University of Colorado Boulder, Boulder, CO 80309, USA
We report results from a study designed to identify links between undergraduate students’ views about exper-imental physics and their engagement in multiweek projects in lab courses. Using surveys and interviews, weexplored whether students perceived particular classroom activities to be features of experimental physics prac-tice. We focused on 18 activities, including maintaining lab notebooks, fabricating parts, and asking others forhelp. Interviewees identified activities related to project execution as intrinsic to experimental physics practicebased on high prevalence of those activities in interviewees’ own projects. Fabrication-oriented activities wereidentified as conditional features of experimentation based on differences between projects, which intervieweesattributed to variations in project resources. Interpersonal activities were also viewed as conditional featuresof experimentation, dependent upon one’s status as novice or expert. Our findings suggest that students’ viewsabout experimental physics are shaped by firsthand experiences of their own projects and secondhand experi-ences of those of others. . INTRODUCTION AND BACKGROUND
Undergraduate physics lab courses are characterized by avariety of learning objectives [1], including developing pro-ficiency with troubleshooting [2], modeling [3], and techni-cal writing [4–8]. National interview studies with instructorshave found that some instructors want students to developsophisticated views about experimental physics, such as thebelief that ‘nothing works the first time’ when conducting ex-periments [9] or the perception of experimentation as an itera-tive process [10]. Skills- and views-based learning objectivesmay be intertwined [11], suggesting that student engagementin experimentation can simultaneously develop students’ pro-ficiency with physics skills while shaping their perceptionsabout physics practice. Relatedly, the goal of this paper isto identify links between students’ engagement in multiweekfinal projects and their views about experimental physics.For several years, researchers have used the ColoradoLearning Attitudes about Science Survey for ExperimentalPhysics (E-CLASS) to gain insight into students’ views aboutexperimental physics. Developed by Zwickl et al. [12], E-CLASS is a Likert-style survey that probes students’ ideasabout what experimentation entails and their perceptions ofexperimental physics as doable or enjoyable. Analysis of E-CLASS scores shows that students’ views are different fromthose of practicing experimental physicists [13]. In an inves-tigation of students’ rationale for their responses to E-CLASSitems, Hu et al. [14] found that students’ views can be nega-tively impacted by their engagement in highly guided lab ac-tivities in which instructions can be followed without under-standing the relevant physics concepts. Consistent with thefindings of Hu et al., Wilcox and Lewandowski [15] demon-strated that students’ post-instruction E-CLASS scores aremore consistent with expert-like responses in courses thatincorporate open-ended activities compared to those that donot. To explain their results, Wilcox and Lewandowski hy-pothesized that “open-ended activities may provide greateropportunities for the students to engage authentically in theprocess of experimental physics” (p. 020132-6).Wilcox and Lewandowski’s hypothesis has support in thephysics education literature. Irving and Sayre investigatedthe experiences of students in an advanced lab course withmultiweek experiments. In the advanced lab course, studentsworked on long and difficult experiments during which theyengaged in activities that align with the authentic practice ofphysics. Irving and Sayre argued the course simulated the ex-periences of participating in a practicing community of physi-cists, thus supporting students to develop knowledge aboutphysics practice [16]. Quan and Elby studied the experiencesof students working on semester-long projects in a researchcourse for first-year physics majors. In the research course,projects spanned theoretical and experimental topics, and stu-dents participated in a regular seminar in which they reflectedon their experiences working on projects. Quan and Elbyshowed that, for some students, participating in authentic re-search can lead to coupled shifts in their confidence and their view of science as an endeavor in which novices can makemeaningful contributions [17].Additional research is needed to fully explore the land-scape of mechanisms that explain how students’ views aboutexperimental physics are shaped by their engagement in au-thentic experimentation. Here, we describe a study that com-plements prior work [15–17] by examining students’ percep-tions of the authenticity of specific activities during the finalproject portion of an upper-level lab course.In our study, we align our definition of experimentalphysics with Ford’s [18] definitions of scientific performanceand practice. Drawing on the Next Generation Science Stan-dards [19] and philosophical work by Rouse [20], Ford de-fines performances as the constituent activities of scientificpractice, and scientific practice as a set of connected per-formances whose collective purpose is to explain nature bet-ter [18]. Ford’s work is a useful lens through which to inves-tigate physics projects, as demonstrated by Quan et al. [21].Following Ford’s lead, we view experimental physics prac-tice as comprised of connected performances, such as build-ing apparatus or analyzing data, whose purpose is to explainthe physical world better. Although Ford’s notions of per-formance and practice did not inform our study design, theyprovide language for articulating our research questions:Q1. When completing projects in a lab course, which ac-tivities do students perceive to be constituent perfor-mances of experimental physics practice, and why?Q2. How does participation in projects inform students’ideas about what experimental physics practice entails?
II. CONTEXT, PARTICIPANTS, AND METHODS
To probe students’ thoughts about their final projects,we collected survey and interview data from undergraduatephysics students enrolled in upper-level optics and lasers labcourses. The courses were required for some physics bache-lor’s degree tracks at a private, selective, Christian, Predom-inantly White Institution (PWI) in the Midwestern UnitedStates. Typical enrollment in each course was about 20 stu-dents per course. Averaged over five years, 13% of coursecompleters were women, and 87% were men; 6% were stu-dents of color, and 94% were white [22].The optics and lasers lab courses were similar to one an-other in content and format. Learning objectives included de-veloping students’ competence with optics- and lasers-relatedtopics and skills. Each course was divided into two seven-week halves. In the first half, students completed weeklyguided lab activities. In the second half, they worked ingroups of two to four students to complete projects, such asbuilding a plasmon laser or achieving single-photon interfer-ence. Groups were assigned based on students’ shared inter-est in a topic. Projects culminated in written reports and oralpresentations. We have previously analyzed data from thispopulation in other studies [23, 24], and a detailed descrip-tion of the course contexts can be found in Ref. [24].Data collection was led by authors DRDF, JTS, and HJL.
ABLE I. Activities emerged from analysis of student responses to weekly reflection prompts. The degree to which students perceivedeach activity to be a feature of experimental physics was probed during a post-project Likert-style survey, and students’ rationale for surveyresponses was probed in follow-up interviews. Columns represent groupings that emerged during analysis of survey and interview data.Execution-oriented Interpersonal Fabrication-oriented Propagation-orientedSetting up equipment Asking a supervisor for help Fabricating parts and materials Reading scientific papersTroubleshooting problems Asking peers for help Building electronics Reading technical data sheetsMaintaining a lab notebook Confirming results with an expert Writing code to interface with equipment Writing lab reportsAnalyzing data Dividing labor among team members Writing code to simulate results Presenting results orallyMaking decisions as a team Reflecting on progress
We collected data from one instance each of the optics andlasers courses. Out of 36 total students, 35 agreed to par-ticipate in our study. Demographics of research participantsclosely match those of course completers.We collected data using free-response surveys, a Likert-style survey, and post-instruction interviews. While projectswere ongoing, we administered weekly free-response sur-veys that prompted participants to reflect on their goals, chal-lenges, and successes; for more details, see Ref. [24]. Eachweek, authors DRDF and JTS read through student reflec-tions and collaboratively generated a summary of students’progress on their projects. Through this process, we identified18 activities that were common topics of reflection (Table I).After identifying the activities in Table I, we created athree-point Likert-style survey that prompted participants toevaluate the extent to which each activity is a feature of ex-perimental physics research: not a feature , might be a fea-ture , or definitely a feature of experimental physics research .The Likert-style survey was administered at the end of thesemester, and post-project interviews were conducted shortlythereafter. Interviews with students in the optics lab wereconducted in person by HJL, and interviews with students inthe lasers lab were conducted remotely by DRDF. During in-terviews, the interviewer asked participants to explain theirrationale for each response on the Likert-style survey.Analysis of interview transcripts was led by authors DRDF,KSJ, and SEP in consultation with HJL. The unit of analysiswas a participant’s explanation for their response to a singleitem on the Likert-style survey. Our goal was to characterizewhich activities were perceived as features of experimentalphysics, and why. We engaged in two iterations of collabo-rative coding during which we coded 630 transcript excerpts.Collaborative coding consisted of multiple coders simultane-ously evaluating a transcript excerpt and reaching consensuson an interpretation. Throughout both iterations, we regularlydiscussed our methods and interpretations with other physicseducation researchers [25] in a deliberate effort to incorporatedialogue into our process for generating claims [26].In the first iteration of coding, KSJ and SEP collaborativelyidentified emergent themes and developed an initial codebookconsisting of code definitions, inclusion criteria, and exem-plars [27]. DRDF played a supervisory role. Preliminary re-sults were presented at a conference [28, 29], after which wereflected on questions and suggestions that arose during the presentations. In the second iteration of coding, DRDF, KSJ,and SEP revised the codebook by redefining some emergentcodes. Selected codenames, definitions, and exemplars fromthe revised codebook are listed here: • Necessary: interviewee referred to an activity as aninevitable, necessary, or required part of research.“You’re going to take data if you’re doing research, sothen you’d have to analyze it for it to be useful.” • Not necessary: interviewee referred to an activity asimportant, helpful, or common, but not a necessary partof research. “You don’t always have to give [oral pre-sentations], but it’s commonly something to do.” • Status-dependent: interviewee refers to their own orothers’ status as a novice or expert experimentalist.“You might be the expert . . . People could check yourwork, but I don’t know necessarily that they’ll be anybetter at understanding it than you will be.” • Resource-dependent: interviewee referred to availabil-ity of resources. “We had to build a lot of them [elec-tronics] here to try to save money, but if you have themoney, it’s nice not to have to build everything.” • Secondhand experience: interviewee referred to oth-ers’ experiences in the course or related contexts. “Ididn’t have to do that [write code to simulate results],but I know other groups that had to do that.” • Firsthand experience: interviewee referred to their ownexperiences in the course or related contexts. “We hadto fabricate our diode laser gain material.”The revised codebook differed from the initial one in severalways. For example, although the initial codebook includedcodes related to students’ context-dependent and experience-oriented explanations, it did not distinguish between status-and resource-dependence or between secondhand and first-hand experiences. Finally, DRDF, KSJ, and SEP collabora-tively recoded the data by applying the revised codebook toall 630 transcript excerpts, and they discussed findings andinterpretations with the other coauthors.
III. RESULTS
Data analysis revealed the following patterns (Table I): • Four activities were almost unanimously perceived as definitely a feature of experimental physics research ;because these activities relate to the execution of a re-earch project, we labeled them execution-oriented . • Five activities accounted for almost all status-dependent codes; because these activities involve inter-actions among people, we labeled them interpersonal . • Four activities accounted for almost all resource-dependent and secondhand codes; because these ac-tivities involve creating apparatus, we labeled them fabrication-oriented . • The five remaining activities were not characterized byobvious coding patterns, but they almost all relate topropagation of scientific knowledge to and from theproject team; we labeled them propagation-oriented .Most activities were identified as definitely or might be a fea-ture of experimental physics research by all participants. Sixactivities—including three fabrication-oriented activities—were identified as not a feature by one or two participants.For all but one activity, a majority of participants referredto firsthand experiences . The exception was writing code tosimulate results, a fabrication-oriented activity for which onlyabout a third of participants referred to firsthand experiences . Execution-oriented activities were more frequently de-scribed as necessary than those in other categories, and, onthe Likert-style survey, they were identified as definitely afeature of experimental physics research by almost all par-ticipants. Compared to other activities, troubleshooting wasmost frequently described as necessary . Consider the follow-ing response from a student we call Brittany:“Troubleshooting problems with equipment. Huge. Wehad so many problems. I said ‘definitely a feature’ be-cause troubleshooting problems is one of the biggestparts of lab projects that I’ve done, no matter whatproject it was. You know, you have an idea, and ofcourse it’s not going to come off exactly right. So that’sbig. And then if you have something running for a longtime, it won’t stay running at the top of its game for itsentire life. You’re going to have to fix it sometimes.”Brittany described troubleshooting as “huge” and “one of thebiggest parts of lab projects,” and she referred to her firsthandexperience encountering problems on her final project and allother projects. Brittany framed the need to troubleshoot as aninevitable part of experimentation (cf. Ref. [9]), indicatingthat she viewed troubleshooting problems with equipment toto be necessary for experimental physics.Compared to other activities, troubleshooting and main-taining a lab notebook were most frequently described as def-initely a feature of experimental physics research . Considerthe following response from a student we call Ashley:“[Maintaining a lab notebook] is typically a very bigpart of research because you need to show proof. Es-pecially, like, if you’re in industry, you definitely needproof that you did things at certain times so that com-peting companies know that you did do it first. Or, ifyou’re in academia, you need that notebook for whenyou graduate from your group, for people coming in touse your work, and having a notebook that’s clear is al-ways helpful. Even in writing this [lab report], there’s things, as I run through the notebook, that I hadn’t re-alized group members had done that I needed to incor-porate, so that’s important.”Ashley described lab notebooks as “very big,” “always help-ful,” and “important” in corporate, academic, and educationalsettings, and she referred to her firsthand experience relyingon a notebook when writing a report for her project. Ashelysaid that notebooks are needed to provide timestamped evi-dence of milestones and to facilitate knowledge transfer be-tween group members, indicating that she viewed maintain-ing a lab notebook to be necessary for experimental physics.
Interpersonal activities were identified as definitely a fea-ture of experimental physics research by a majority of partic-ipants, and they accounted for almost all instances in whichparticipants discussed the status-dependent nature of an activ-ity. Compared to other activities, asking a supervisor for helpwas most frequently described as status-dependent . Considerthe following response from a student we call Michael:“Asking a supervisor for help I said was just maybe.Some people could be way better at this than I am. Andso therefore they know exactly what they’re supposedto do, and after the supervisor tells them right at thebeginning, ‘You’re going to want to do this work,’ theycan just go. And there have been times at the lab where[the project team] have gone a week without talkingto our professor about [the project] because we have avery clear idea of where we want to go. So—and also,if you were at the top of your field, doing research thatno one has done anything like before, then you may nothave a supervisor to ask for help.”Michael referred to his firsthand experience working on hisproject without input from his professor to illustrate his viewthat it is possible to conduct research without asking a super-visor for help. According to Michael, people with high re-search competence are unlikely to ask a supervisor for help,and someone who is “at the top” of their field may not beable to do so, indicating that he viewed asking a supervisorfor help as a status-dependent aspect of experimentation.
Fabrication-oriented activities were identified as mightbe a feature of experimental physics research by a majorityof participants, and they accounted for almost all instances inwhich participants discussed the resource-dependent natureof an activity or referred to their secondhand experiences .Compared to all other activities, fabrication was most fre-quently described as resource-dependent . Consider the fol-lowing response from a student we call Brandon:“Fabricating parts I said ‘might be’ because you mightbe able to just find a company that’s manufacturing thething, the equipment you need, so you might be able tojust buy it. Or, if you can’t find it, or you find some-thing that’s close but not exactly—you might have tokind of edit it or go into a shop and actually build it.Like, I know one lab group has done that a whole lot.”Brandon referred to a secondhand experience in which agroup built parts of their apparatus to illustrate his view thatcommercial availability of equipment can inform whether orot experimentation involves fabrication. Thus, Brandon’sview is consistent with the notion that fabricating parts or ma-terials is a resource-dependent aspect of experimentation.
Propagation-oriented activities were identified as defi-nitely a feature of experimental physics by a majority of par-ticipants. However, we did not notice other obvious patternsin code assignments. For example, while reading scientificpapers was described as necessary by a majority of partici-pants, presenting results orally was more frequently describedas not necessary than any other activity. Consider the follow-ing response from a student we call Logan:“[Presenting orally] can be helpful for explaining yourresults to other people, but I wouldn’t say it’s abso-lutely essential since you can always just publish . . . ”Logan reasoned that, because oral presentations are not theonly mechanism for propagating research findings, presentingresults orally is not necessary for experimental physics.
IV. DISCUSSION AND LIMITATIONS
In response to research question Q1, almost all partici-pants viewed the activities in Table I to be constitutive per-formances of experimental physics practice, sometimes con-ditionally. Execution-oriented activities were often viewedas necessary for experimentation, whereas interpersonal andfabrication-oriented activities were viewed as conditional as-pects of experimental physics, depending on the novice or ex-pert status of team members and the availability of commer-cial apparatus. Participants’ views on propagation-orientedactivities were mixed, but not because propagation of knowl-edge itself was viewed as disconnected from experimentalphysics practice. Rather, participants acknowledged a vari-ety of avenues through which research findings can be sharedwith others, some of which may be prioritized over others.In response to Q2, participants regularly referred to theirfirsthand experiences working on projects to illustrate orjustify their views about the role of an activity in experi-mentation. Because some projects did not involve buildingparts or writing code, participants’ views about fabrication-oriented activities were often informed by secondhand expe-riences and perceived to be conditional aspects of experimen-tation. Thus, participants’ views about experimental physicswere shaped by their own experiences working on multiweekprojects and their perceptions of their peers’ experiences.Our study design constrains the generalizability of ourfindings. We investigated students’ views about only a sub-set of activities that are relevant to experimentation and thatarose in a particular educational context. Moreover, whitemen are more overrepresented in the courses we studied thanamong physics bachelor’s degree recipients in the UnitedStates, which likely contributes to homogeneity of the viewsreported here. Indeed, as measured by E-CLASS, on average,women’s and men’s views about experimental physics differin some respects, and nonbinary people’s views have not beenexplored [30]. Hence, we cannot achieve probabilistic gener-alization to all students or all lab courses. Instead, we strive for theoretical generalization [31], which, for us, involves in-ferring some plausible mechanisms through which projectsmay influence students’ views about experimental physics.One such mechanism is repeated firsthand experiencewith an activity, similar to Brittany’s experiences with trou-bleshooting. Brittany’s view of troubleshooting as inevitableis consistent with a common stance among electronics labinstructors: troubleshooting does not need to be explicitlytaught or assessed in lab courses because ‘nothing worksthe first time,’ and the need to troubleshoot arises organi-cally [2, 9]. Irving and Sayre [16] argued that students areaccountable to a different kind of physics knowledge in labcourses than in other courses. It is plausible that studentscome to view troubleshooting as inherent to experimentationbecause technical problems are frequent, and students are im-plicitly accountable for learning how to troubleshoot them.Engaging groups in unique projects may be another mecha-nism that shapes students’ views about experimentation. Irv-ing and Sayre [16] argued that students develop experiment-specific expertise when groups work on distinct projects at thesame time. Similarly, we find that students notice experiment-specific approaches to experimentation. Comparing and con-trasting their own approaches to those of others could giverise to combinations of firsthand and secondhand experiencesthrough which students develop nuanced ideas about exper-imental physics as comprising context-dependent combina-tions of performances. Future work could explore the impactof working on different projects at the same time on students’views about experimental physics.Although we did not engage with Ford’s [18] idea thatpractice comprises performances that are connected andwhose purpose is to explain nature better, we see hints thatsome participants viewed some activities as purposeful. Ash-ley recognized that maintaining a lab notebook plays mul-tiple roles in the experimental process, in alignment withwhat Hoehn and Lewandowski refer to as ‘writing as pro-fessionalization’ [4]. Ashley’s perception of lab notebooksas “always helpful” is in contrast to work by Stanley andLewandowski [6], who found that many graduate students donot view lab notebooks as a beneficial part of their undergrad-uate lab courses. However, evaluating student responses likeAshley’s using Ford’s notion of purposefulness is beyond thescope of the present work. Future work could explore howfinal projects in lab course support students to view perfor-mances as connected or purposeful (cf. Ref. [21]).
ACKNOWLEDGMENTS
S. Lavender and L. Torres helped with undergraduate re-search logistics, C. Hoyt helped collect data, L. Kiepurahelped transcribe interviews, and the WWU PER Group(especially A. Boudreaux, R. Barber DeGraaff, and T. Lê)helped interpret results. This material is based on work sup-ported by the NSF under Grant Nos. 1726045, 1323101,1734006, and 1208930, and by the Washington NASA SpaceGrant Consortium under Grant No. NNX14AR60A.
1] J. Kozminski, N. Beverly, D. Deardorff, R. Dietz,M. Eblen-Zayas, R. Hobbs, H. Lewandowski, S. Lin-daas, A. Reagan, R. Tagg, J. Williams, and B. Zwickl,
AAPT Recommendations for the Undergraduate Physics Laboratory Curriculum (AAPT, College Park, MD, 2014).[2] D. R. Dounas-Frazer and H. J. Lewandowski, Electronicslab instructors’ approaches to troubleshooting instruction,Phys. Rev. Phys. Educ. Res. , 010102 (2017).[3] D. R. Dounas-Frazer, L. Ríos, B. Pollard, J. T. Stan-ley, and H. J. Lewandowski, Characterizing lab in-structors’ self-reported learning goals to inform devel-opment of an experimental modeling skills assessment,Phys. Rev. Phys. Educ. Res. , 020118 (2018).[4] J. R. Hoehn and H. J. Lewandowski, Frame-work of goals for writing in physics lab classes,Phys. Rev. Phys. Educ. Res. , 010125 (2020).[5] J. T. Stanley and H. J. Lewandowski, Recommendations forthe use of notebooks in upper-division physics lab courses,American Journal of Physics , 45 (2018).[6] J. T. Stanley and H. J. Lewandowski, Lab notebooksas scientific communication: Investigating develop-ment from undergraduate courses to graduate research,Phys. Rev. Phys. Educ. Res. , 020129 (2016).[7] C. L. Ramey II, The Transformation of an Upper-Division Laband Comparative Analyses of Two Scientific Writing Activities ,Master’s thesis, Texas Tech University (2018).[8] P. Coppens, J. Van den Bossche, and M. De Cock, Goalsof lab work in electronics: Student and staff ideas,International Journal of Electrical Engineering Education , 124 (2016).[9] D. R. Dounas-Frazer and H. J. Lewandowski, Nothing worksthe first time: An expert experimental physics epistemology,in Physics Education Research Conference 2016 , PER Confer-ence (Sacramento, CA, 2016) pp. 100–103.[10] D. Dounas-Frazer, J. T. Stanley, and H. J.Lewandowski, Instructor perspectives on itera-tion during upper-division optics lab activities, in
Physics Education Research Conference 2017 , PER Con-ference (Cincinnati, OH, 2017) pp. 116–119.[11] B. R. Wilcox and H. J. Lewandowski, Developing skillsversus reinforcing concepts in physics labs: Insight froma survey of students’ beliefs about experimental physics,Phys. Rev. Phys. Educ. Res. , 010108 (2017).[12] B. M. Zwickl, T. Hirokawa, N. Finkelstein, and H. J.Lewandowski, Epistemology and expectations survey aboutexperimental physics: Development and initial results,Phys. Rev. ST Phys. Educ. Res. , 010120 (2014).[13] B. R. Wilcox and H. J. Lewandowski, Students’views about the nature of experimental physics,Phys. Rev. Phys. Educ. Res. , 020110 (2017).[14] D. Hu, B. M. Zwickl, B. R. Wilcox, and H. J. Lewandowski,Qualitative investigation of students’ views about experimentalphysics, Phys. Rev. Phys. Educ. Res. , 020134 (2017).[15] B. R. Wilcox and H. J. Lewandowski, Open-ended versus guided laboratory activities:impacton students’ beliefs about experimental physics,Phys. Rev. Phys. Educ. Res. , 020132 (2016).[16] P. W. Irving and E. C. Sayre, Conditions for building a community of practice in an advanced physics laboratory,Phys. Rev. ST Phys. Educ. Res. , 010109 (2014).[17] G. M. Quan and A. Elby, Connecting self-efficacy and viewsabout the nature of science in undergraduate research experi-ences, Phys. Rev. Phys. Educ. Res. , 020140 (2016).[18] M. J. Ford, Educational implications of choosing “practice"to describe science in the Next Generation Science Standards,Science Education , 1041 (2015).[19] NGSS Lead States, Next Generation Science Standards: For States, By States (The National Academies Press, Washington, DC, 2013).[20] J. Rouse, Practice theory, in