Coding Unit and Coding Schemes

A code in qualitative inquiry is most often a word or short phrase that symbolically assigns a summative, salient, essence-capturing, and/or evocative attribute for a portion of language-based or visual data.                -Saldana, 2009, p. 3

Advantages of the use of video in education research include the capacity to replay footage in order to review observational data, the ability to use multiple-coders to increase coding reliability, and the means to capture breadth and depth of observations not afforded by naturalistic observations. Video-analysis includes several stages which are largely similar to qualitative content analysis. In this section we will talk about coding considerations and some of the prevalent coding schemes for video-analysis.

Video Survey Research

Video survey research (see Stigler, Gallimore, & Hiebert, 2000) combines qualitative videotaping techniques with quantitative national survey research techniques. The TIMSS 1999 Video Study (Jacobs et al., 2003) is an example. It involves a large number of video samples (nationally/ internationally representative) and uses mixed-methods research analysis. Such studies involve multiple coders distributed geographically and diverse in their backgrounds. To ensure consistency and reliability, such studies demand developing a detailed and systematic coding manual describing the coding units and schemes, and definitions.

Coding Unit

The coding unit specifies the basic entity or phenomenon that is to be analyzed during a video episode. Defining the coding unit is considered very important, as without it there can be differences in the coding decisions across episodes and also within episodes for different coders. The coding unit should be one that affords understanding of the phenomena being studied without adding repetitiveness or confusion to it. Lack of a coding unit also poses issues for comparability of findings with other similar studies. Some of the coding schemes that are used more often in video survey research studies are discussed below. The decision to use one over the other is usually informed by research questions, researchers’ pragmatic knowledge, and/or existing theories.

Coding Schemes

Time-based Coding:

This approach divides the video episode into time-intervals. Each of these time-interval streams are then coded for occurrence or quality of behavior(s) in question. For example, the MQI (Mathematics Quality of Instruction) instrument (Hill et al., 2008; Hill, Umland, Litke, & Kapitula, 2012) uses time intervals of 5 minutes to then apply a score on the five MQI elements. For more details, please refer to http://isites.harvard.edu/icb/icb.do?keyword=mqi_training&pageid=icb.page394700

  • Time-based coding is effective for measuring discrete behaviors (e.g., occurrence coding- absence or presence of student questioning) or frequently-occurring behaviors (e.g., ranking coding- ranking teacher’s explanation of mathematical ideas).
  • It is not useful when coding infrequently-occurring behaviors or events that are long in duration (e.g., coding a discussion of a math problem within a lesson).
  • It is crucial to thoughtfully decide and justify the time-interval selected. It should not be too long so as to notice the occurrence of the behavior too frequently (or more than once). It should not be so brief that the behavior extends into the next interval.

Event-based Coding:

Unlike time-based coding, event-based coding unit is independent of time. An event, defined by the behavior being studied, is the unit of analysis. The event could then be the entire video-episode or a portion of it, depending on the length of the naturally occurring behavior. For example, if the construct being studied is students’ interactions with their peers during collaborative work, the coding unit in the video-lesson would be any time the teacher allows for students to work in groups.

  • It is effective for measuring frequently and infrequently-occurring behaviors.
  • It is important to clearly specify how to recognize when the event begins and when it ends. This sometimes poses a challenge, as it may not always be clear when one event ends and the next one starts.

Coverage Coding:

Coverage coding was one of the types of coding used in the TIMSS (Third International Mathematics and Science Study, 1999) Video Study (Jacobs et al., 2003). Coverage-based coding divides the whole video-episode (or specified part) into mutually exclusive categories. For example, an entire mathematics lesson can be divided into non-overlapping periods of “1) no mathematical work, 2) mathematical organization or management, or 3) mathematical work” (Jacobs et al., 2003, p. 88). Each of the main categories may then have further sub-categories or codes.

  • Coverage coding is useful in that it allows partitioning of the entire episode into smaller, manageable instances of mutually exclusive activities.
  • Note that in quantitative video analysis, categories are required to be mutually exclusive, otherwise it will violate the assumption of independence required for most statistical procedures. In contrast, qualitative video analysis allows for co-occurring codes.

Ranking (Rating/ Scoring) Coding:

Ranking coding is used to measure the quality of a behavior being studied. For example, if the construct being studied is mathematical questions posed by the teacher in a classroom, then it might be useful to rank all the observed teachers based on the level of cognitive demand of the questions they pose (e.g., low, medium, high). Ranking coding is also widely used when comparing within the unit of analysis. The MQI instrument (CITE) also uses ranking coding to score each time-interval based on predetermined MQI elements. Additionally, the Mathematics Scan (M-Scan) instrument (CITE) uses a 1-7 rating to measure mathematical instructional quality.

Global (face-sheet) Coding:

Global coding (also referred as face-sheet coding) are the codes that apply to the entire video-episode, case, or lesson being coded. Global codes are usually helpful in identifying an overall characteristic of the video-episode, such as ethnicity, gender, socio-economic status, or achievement.

Occurrence or Frequency Coding:

Occurrence or frequency codes are marked anytime a behavior occurs during a video-episode, which can then tell us the total number of times a behavior occurred during the entire video-episode. Occurrence/ frequency coding is used widely in qualitative studies.

References

New M-Scan citation

Hill, H. C., Blunk, M. L., Charalambos, C. Y., Lewis, J. M., Phelps, G. C., Sleep, L., & Ball, D. L. (2008). Mathematical knowledge for teaching and the mathematical quality of instruction: An exploratory study. Cognition and Instruction, 26, 430-511.

Other Hill citation

Jacobs, J., Garnier, H., Gallimore, R., Hollingsworth, H., Givvin, K. B., Rust, K., . . . Stigler, J. (2003). Third International Mathematics and Science Study 1999 Video Study Technical Report Volume I: Mathematics, NCES (2003-012). Washington, DC: National Center for Education Statistics.

Ostrov, J. M., & Hart, E. (2014). Observational methods. In T. D. Little (Ed.), The Oxford handbook of quantitative methods, 1(1), 286-304. (?? Where is this in the text…?)

Saldaña, J. (2009). The coding manual for qualitative researchers. London: SAGE Publications Ltd.

Stigler, J. W., Gallimore, R., & Hiebert, J. (2000). Using video surveys to compare classrooms and teaching across cultures: Examples and lessons from the TIMSS Video Studies. Educational Psychologist, 35(2), 87-100.