Researchers Using Video Face Options and Challenges
June 10, 2010
Video technology is transforming research in the learning sciences.
Video provides rich records of student and teacher interactions: eye gaze, gesture, body posture and proximity, content of talk, tone of voice, facial expressions, and use of physical artifacts. Video also captures between-person processes such as the alignment and maintenance of joint attention.
Video technologies allow researchers to collect, share, study, present, and archive detailed cases of teachers’ practice. Video libraries support teaching, learning, and intensive study of those practices.
UW-Madison education professor Sharon Derry chaired a national task force of scholars in writing guidelines for researchers who use video and who want to better design formal and informal learning environments. Some findings from their work, funded by the National Science Foundation, recently appeared in The Journal of the Learning Sciences. Researchers face several challenges, including deciding which elements of a complex environment should be video recorded; which analytical frameworks and practices are available and appropriate; what technological tools are available and which must be developed; and how research protocols can encourage data sharing while protecting subjects’ privacy rights.
Selective emphasis shapes all phases of video research, determining which events are brought into focus for deeper analysis. The process involves (a) planning a study, (b) shooting original footage, (c) choosing one or more clips from a body of such footage, (d) focusing on the selected video appropriately, depending on research goals, (e) developing final products for presentation, and (f) archiving and curating video and related products.
Most research using video involves detailed analysis of selected clips. Researchers are often concerned with closely describing and accounting for the relative frequency of a type of event. Inductive approaches apply when a minimally edited video corpus is collected and/or investigated with broad questions in mind but without a strong orienting theory. A deductive approach is required when the researcher has a strong theory and clear research questions.
Researchers taking a narrative approach may work more or less deductively or inductively. Those with a strong theory or set of research questions may delineate the terrain in advance by designing interview or protocols before stepping into the research setting.
The Trends in International Mathematics and Science Study (TIMMS) video project represents a way in which these two categories of selection overlap conceptually. The huge TIMMS video corpus has been used for professional development purposes and to support cross cultural studies of teaching. But “sampling representativeness” has been an overriding concern in selecting for both purposes.
A discovery-oriented approach to video analysis aims to reveal unanticipated findings. A top-down approach, meanwhile, uses records to identify and code events determined before data collection. The task force guidelines suggest that researchers should focus first on theory-driven questions and develop concrete plans for a first pass through the video records. Having good questions helps maintain perspective and prevents the researcher from getting lost in detail.
At the same time, the researcher should anticipate new discoveries and be ready to articulate questions that can be refined and tested during multiple passes through the data. An explicit multistage analysis can strengthen the likelihood of generating findings that are reliable and valid.
In the early stages of analysis it can help to share a video segment with a group of other researchers to gather multiple interpretations of the events, to specify dimensions for analysis, and to brainstorm potential issues to investigate further.
Technologies for sharing and reporting video research include tools for analysis, tools for supporting case development and sharing, models for sharing video in reports of research, metadata schemas, collaboratories and virtual repositories, and practices addressing legal and ethical issues related to video sharing.
Unfortunately, most video technologies support only certain phases of the workflow. At least 10 different functions of video research are supported by different tools. That means researchers must investigate available tools and acquire or develop a toolkit that meets their budgets, that can be accessed or supported from their sites, that supports their research practices and goals, and that involves them in a supportive user community.
Researchers need to consider different ways to share and report video data together with print and other media. For example, a few minutes of a single-camera video record can be provided on a compact disc and included with a printed issue of a journal. Relevant video clips can be posted to, and accessed from, a website, and site visitors may add commentary. Sharing video data among learning scientists could help accelerate the growth of scientific understanding of learning and teaching. Multiple researchers could gain access to video data records that now tend to reside on the shelves, on DVDs, and on hard disks of individual researchers.
Confidentiality concerns arise whenever researchers collect or share recordings in which viewers may recognize individual subjects. Confidentiality can be protected by restricting access to the video and to personal information such as the names of the participants or the schools in which data were collected.
What parts of a video corpus should be a public resource and for what purposes are controversial questions. A major difficulty is that video data, once collected, might be used in multiple ways, and it may not be possible, at the time that a recording is made, to specify what these uses might be. Furthermore, often it is not known at the time of collection even who will be studying the data. As video data records travel further from the research project in which they were collected, the types of users and uses may expand in unpredictable ways. So it’s important to register extensive metadata on the contextual aspects of video data capture and attributions for subsequent citation and source tracing.
Today’s video researchers must strive to become adaptive experts with knowledge of the many issues covered in these guidelines, ranging from analysis to technology to ethics. But no one researcher can do it all. The very objects of video research must include programs of work that will create infrastructures allowing us to become an adaptive, distributed, collaborative, expert community.