Anne & Carol Hu on recordkeeping rsch

• traditional archival techniques for description:
  • archivist's knowledge about the medium in which the materials were created and functions records served
  • an understanding arrangement and structure of records
  • identification of record creators and major topics
• relying on the above is problematic in the description of elect. computer conferences (archivists know less about origin and contents of conferences, nature and function of records)
• this paper describes a research project designed to explore new and rigorous data gathering methods to complement traditional approaches for the archival desc. of computer conferences of historical value
• 1992 Bentley Historical Library conducted research to determine whether public conferences at U of M have archival value for documenting the intellectual, social and cultural life of a major acad. institution
• the project described in the paper was a small pilot study, distinct from the main thrust of the project, to explore new approaches to the description of computer conferences.
• Bentley project archivists determined that conferences have value and need to be documented, their import is often not documented in the administrative record
• need to develop thorough and helpful finding aids that reveal more information about conferences
• the traditional techniques for archi. desc. need to be enhanced for use with elect. conferences because:
  • applying approaches to screen data more difficult that paper
  • need to develop better narratives and inventories (more self-explanatory) of what happened at conferences for access by anyone from any computer
  • archivists might be subject to bias - favoring or believing a conference person or topic to be of more importance than another
  • these above reasons led the researchers to consider if using automated techniques would be more beneficial for analyzing and describing data from conferences
• researches developed a research project to explore in detail the use patterns and structure of one public computer conference, using observational data taken unobtrusively from active files of the online conference itself
  • tested the applicability of traditional archival description paradigms into the new medium of computer conferences - enhanced version of trad. archiv. description (not a new methdology for)
  • researchers chose the conference Wing:Span to study.

Literature Review 

• no research lit. exists in AS on this form of elec. communication or on the application of statistical or other analytical techniques in archival descriptive processes of computer conference records
• although some have written about the need to apply basic archi. principles to new technology, but amplified; little to say about methods fro describing conferences for the purposes of historical documentation

Methodology 

• hypotheses:
  • certain individ. would dominate conference
  • number of participants, their responses and items responded to would be narrow
  • indivd. that dominated conference would also dominate initiation of items
  • there would be considerable variation in the activity of the conference at different times of year
  • individual items could be classified by type, and type would exhibit a life cycle for analogous types of traditional records (on going, administrative, issues of topical interest

Data Collection (this is a meta - description of the process of collection; describes and defines and explains why particular methods were used)

• researchers explain why they chose this conference
• method selected for data collection was to examine what information was available from the system-generated item descriptor lists
• they gathered information in vivo and manually coded and tallied the number of discussion items and participant levels in the related discussion responses
• they analyzed the data graphically, using exploratory data analysis (used when a researcher is not sure what to expect from the data to reveal preliminary patterns that can be further examined)
• counts were made of each of the participant's responses, topical items responded to and items initiated
• developed decision rules for data that might present problems for coding; divided collection activity between themselves

Data Analysis (also meta description - they really define and explain how the statistical methods used work, in order to explain their results )

• frequency counts of total responses were graphed
• used Analysis of Variance to look for differences among groups as the samples involve the same people measured over time - tests if the differences that occur are greater than what would occur randomly
• explored patterns of participation over time - information contained in the item descriptor lists to count the number of items initiated by month
• created a scatterplot to study participation patterns and dominant individuals
• used a statistical test of correlation and regression (a prediction measure - can predict the number of responses made from any given number of items responded to)
• (I am not going to detail findings here, see paper)

Findings about Wing:Span that could be used for Archival Description (these could not be discerned intuitively or by the help of summary item descriptions)

• the organizer of the conference dominated discussion and initiation of items
• some difference occurs in usage  - times of year
• wing:span deals primarily with topical issues related to women rather than ongoing administration or games
• few items exhibited life cycles similar to trad. archi. materials
• item initiators do not correspond well with item responders - not a good predictor of dominance in this conference
• strong correlation for individuals between number of responses and number of items responded to; can discern roles of participants
• participation was surprisingly broad

Generalizable findings for Archival Desc. of Computer Conferences

• key topics in terms of longevity and activity patterns
• degree of internal structure and arrangement of conference
• breadth of participation - whole conference and individuals
• dominant players among respondents or initiators
• degree of relationship between respondents and initiators - helps archivists make inferences about the purposes for which items are being initiated
• difference in usage over time periods
• points at which participants join or drop out
• apart from first two items, none of this information could have been obtained with trad, archival unstructured observation, nor could  software itself generate info.

Conclusions 

• this systematic approach might enhance archi. description of computer conf. of historical value and might also be helpful for other forms of electronic information; perhaps a starting point for prepackaged statistical or rule based automation system
• would need though to be replicated