Whether beginning with live observation or videotaped recording, research based on systematic observation of behavior can be regarded as progressing through five phases. First, investigators form research questions and identify behaviors of interest. Second, they develop and pilot coding schemes that name and describe those behaviors. Third, guided by their coding schemes and whatever recording devices are available and appropriate, they record data. Fourth, investigators represent the recorded data in ways that facilitate analysis. And fifth, they analyze those data so as to address the questions that motivated the investigation initially. This talk focuses primarily on the fourth and fifth of these steps.

Computerized procedures and tools for data collection seem relatively well developed (e.g., Noldus’ Observer). In contrast, general as opposed to specific computer programs or packages for reduction and analysis of observational data are rare. Bakeman and Quera [1, 2], reasoning that lack of a common format for sequential data has impeded the development of general purpose analytic programs, have defined a sequential data interchange standard (SDIS). For ease of use and to reflect procedures that have been used and found useful historically, five data forms are defined by SDIS. The five are event, state, timed event, interval, and multi-event sequences. Almost always behavioral sequences recorded by investigators can be expressed in one or another of these five forms.

Any standardization of sequential data that became widely used would prove advantageous. We think SDIS is a simple, useful, and flexible candidate. Observers can easily record data in this format directly or, when investigators possess recording equipment that uses a different format or wish to analyze data from an existing archive, it is a simple matter to write programs (e.g., in Basic or Pascal) that reformat existing data into the SDIS format.

Once sequential data are expressed using SDIS conventions, the considerable power of programs like GSEQ [2] can be brought to bear, including its ability to modify existing codes (e.g., recoding or lumping existing codes), create new ones (e.g., combining existing codes using standard and, or, and not logical operations; or defining new codes keyed to onsets and offsets of existing ones, such as the five seconds after the onset of crying, or the two intervals after the onset of smiling), and produce any number of cross-classified counts (e.g., the number of seconds within five seconds of the onset of infant crying that contain an onset of maternal comforting). The remainder of this talk considers capabilities and limitations of GSEQ and the feasibility, once data are expressed in SDIS format, of writing additional, specialized programs to analyze SDIS data.

References

1. Bakeman, R.; Quera, V. (1992). SDIS: A sequential data interchange standard. Behavior Research Methods, Instruments, and Computers, 24, 554-559.
2. Bakeman, R.; Quera, V. (1995). Analyzing Interaction: Sequential Analysis with SDIS and GSEQ. New York: Cambridge University Press.