Merging HUs: Addressing the Schism Between Theory and Practice in Information Science

May 27, 2014

Author: B. Jane  Scales

Beginning a project in ATLAS.ti entails the creation of the Hermeneutic Unit (HU), the “container,” which holds everything related to the project:  primary documents, quotations, codes, families, memos, comments, and analysis generated by the various query and exploration tools.

In this brief paper, I examine the structure of two topically-related HUs, before merging them. My objective is to determine the following: a) How does one decide to merge two projects into one HU?  b) What conceptual and technical issues should one consider before merging two HUs?  c) How should one prepare the HUs for a successful merging?  Finally, I review the new HU, and identify any remaining issues to be addressed to prepare it for further use and exploration.

Two HUs – The Theoretical and the Practical

The two HUs I discuss facilitate my work with library reference issues.  Both are related to the evaluation of online reference services offered by Washington State University libraries. However, each HU was created at a different time, and for a different purpose.

In a previous blog post, “Reviewing Article Abstracts: Content Analysis and Synthesis,“ I relay my experience using ATLAS.ti to assist with a literature review.  This project, Literature_Review.HU, helped me quickly get a handle on issues related to the assessment of online reference chat transcripts.  Reference chat transcripts are the text product of instant messages between librarians, and the students who request online assistance via the Libraries’ web site.

To review the existing literature on the topic of reference chat transcript analysis, I coded abstracts from a set of thirty-nine of the most relevant and recent literature exploring this topic.

As a result, I quickly got a good overview of the theoretical frameworks others have used to assess these interactions. Subsequently, I created the second project, Library_Transcripts.HU, in order to evaluate a set of eighty-five reference chat transcripts.  See Figure 1.

Figure 1. An example of an anonymised reference chat transcript.

Figure 1. An example of an anonymised reference chat transcript.

After some reflection, and well after I had completed the second HU, I realized that I could have easily placed all of the PDs of both projects into one, and developed a more comprehensive examination of both the practical assessment of the transcripts, as well as the theoretical issues related to that assessment.

This lack of foresight became more apparent after reading articles such as Cibangu’s 2013 publication, “A Memo of Qualitative Research for Information Science: Toward a Theory Construction.” Cibangu comments on the continued call for more theoretical development and exploration within information science that have occurred over the past thirty years. In his conclusion, Cibangu states:

The trends and signs observed in information science’s qualitative work, from 2010-2011, call for a stronger engagement with qualitative research’s discussions. Theory is the tissue that connects our works and areas of study, and fills the gaps found in literature(s). Theory should not be understood as a wholesale flight from reality, but a tool with which to gain a tighter grasp of reality and a richer body of work in information science.

Objectives for Merging the HUs

My thought in joining both HUs, was to bring the theoretical and practical codes together, and explore their relationships.  By merging the two HUs, I hoped to use the themes presented in the professional literature (Literature_Review.HU) and compare them to the data coded within the Libraries’ actual chat transcripts.  This process should help the assessment progress from being merely descriptive in nature, and to being rooted or linked to the literature, and, thereby, to potentially appropriate theories.

Admittedly, throwing all the PDs, codes, and code families from these two projects together into one does not guarantee a more theoretical analysis of the data. However, it will facilitate a closer examination and likelihood that the more “practical” or descriptive elements can be linked to literature that has already explored qualitative analysis of these types of transcripts.

One can compare codes and code families from these two projects in Figure 2 and Figure 3.  The list of codes in Figure 2 is from the Literature_Review.HU, and illustrates topics and theories explored by information professionals when assessing reference chat transcripts.

The codes shown in Figure 3 are from the Library_Transcripts.HU.  They are essentially labels assigned to the transcripts describing status, type of question, and the type of links used to answer the questions.  However, these descriptive codes stand in isolation without any framework or context to help process their meaning, or inform some action.

Figure 2. Example codes from Literature_Review.HU

Figure 2. Example codes from Literature_Review.HU

Figure 3. Example codes from Library_Transcripts.HU

Figure 3. Example codes from Library_Transcripts.HU

For example, some of the codes within Literature_Review.HU are linked to passages describing the use of reference chat transcripts to inform the training of library staff.  Likewise, there are coded transcripts in Library_Transcripts.HU, perhaps ones in which library staff could have responded better, which may provide the material on which staff need to be trained.  By linking items from the different HUs, I could potentially develop a training plan for library staff closely based on and informed by the professional literature.

The Merging Process

I was not familiar with the Merge HU process in ATLAS.ti, however after reviewing the ATLAS.ti manual, I learned that it was mainly intended to facilitate team work.

While I’ve worked collaboratively with colleagues on projects that involved ATLAS.ti, I have never done it in a way that required HUs to be merged at any point.   I can see why the Merge function is important for projects with multiple researchers cross checking codes, and other activities to verify or challenge the research process.  Savin-Baden and Major discuss collaborative research as a means of improving the quality of analysis.  “Triangulation” and “Member checking” might be other processes that call for more than one individual editing an HU.

Reviewing the ATLAS.ti 7 manual (pages 154-165), we see that there are many options in how one chooses to merge HUs.  The first step is to look at the two projects and determine what the priorities of the merger are.  I carefully read through the options and determined that I really didn’t want to “merge” items within the two HUs, but to add items of HU to another.  For example, I did not see an overlap of codes or code families in the two projects, that could be collapsed together.  The codes and code families, at least at this point, needed to remain separate.  This simplified things quite a bit.

I made Library_Transcripts.HU my “target” project, and saved a copy of it before I started in case something went wrong.  Under “Project,” I selected “Merge with HU,” and set all of the options to “Add.”  See Figure 4.

Figure 4. Merging Wizard - adding one HU to another

Figure 4. Merging Wizard – adding one HU to another

I hit the “Finish” button, and after a second the Merge Report opened on the screen.  See Figure 5.

Figure 5. Section from the Merge Report

Figure 5. Section from the Merge Report

Outcomes of the HU Merging

I started with two HUs, similar in topic, but different in their focus.  The new HU contains 86 primary documents, 420 quotations, 162 codes, and 419 codings.  Figure 6 shows how the new project contains PDs from both original projects. (Figure 6 shows the PDs numbered up to 90; however, there were only 85 transcripts. Three transcripts had been removed earlier, but the original numbering of the PDs remained.)

Figure 6. PDs from the two original HUs are now in the new Merged.HU

Figure 6. PDs from the two original HUs are now in the new Merged.HU

Figure 7. Code manager in the Merged.HU.

Figure 7. Code manager in the Merged.HU.

Developing the Question Type Codes in Merged.HU

I now had all my data in one HU, however, this was not the end of the process. Instead, the new environment forced me to consider how particular groups of codes from the merged HUs could be related. Before I could start discovering these relationships, however, I needed to make a second pass through my coding and better develop and clarify codes in the article abstracts (contained in P90: Virtual Reference.docx).

I wrote a memo describing this process:

Figure 8. Memo documenting my next steps

Figure 8. Memo documenting my next steps

I developed two sets of “questions” codes from previously identified quotations in the literature PD. One set of the codes, I labeled “Q Theory.” These codes related to issues regarding the establishment of question codes or schemas used to analyze chat reference transcripts. Among the newly developed codes were:

Q Theory: question types

Q Theory: query clarification

Q Theory: limitation of analytic categories

Q Theory: disambiguation

Q Theory: conversation analysis

Q Theory: coding by question causes

Because the analysis was getting more in depth, I decided to import the full text of relevant articles into the Merged.HU. This allowed me to probe these codes a bit more and link them to relevant passages within the literature. After completing some of this coding, adding comments to the codes, and incorporating quotations from some of the articles used initially in Reviewing Article Abstracts: Content Analysis and Synthesis I began working on a network view of how some of the Q Theory codes related to the question types I had initially developed to analyze our libraries’ chat transcripts. I also created a code family to organize the WSU-based question codes, and determined one of the codes belonged more within the STATUS code family. The new code family was labelled WSU Question Codes, and included eight codes. See Figure 9.

Figure 9. Network view of the WSU Question Codes within the context of other Coding Schemas

Figure 9. Network view of the WSU Question Codes within the context of other Coding Schemas

I was a bit surprised to learn that there were many different methods of categorizing reference questions, but I now had some links to look closer at the methods and reasons others had used to tackle this part of chat transcript analyses. As much as anything, it illustrated to me the lack of standardization. Most academic libraries will adopt the categories that have the most practical application to their institutions.

The exercise of piecing together the various codes, family codes, and quotations in Figure 9 prompts me to look a bit more closely at how other researchers have organized their chat transcript analyses. Perhaps by examining other theories that produced these different question type schemas, I can learn additional methods and comparisons from the literature that I might have missed. In doing this, I begin to address some of the criticisms of contemporary qualitative research in information literacy outlined by Cibangu.

At any rate, the visualization of ideas and theories represented in Figure 9 is valuable because I can easily return to the project, and not waste time retracing my steps and re-discovering these issues. My next step will be to write a Memo on how the ideas and relationships represented in Figure 9 will inform my research.

Alternatives to Merging HUs

Merging or combining the two projects was not difficult. However, I could have avoided the need to merge HUs if I had planned a little better.  Rather than creating a new HU for the two projects, I could have built one and developed my codes a bit more efficiently.  In the future, I hope to think more conceptually about the research before starting.

Still, I now have a more comprehensive project, which promises a richer potential for analysis and improved library services. By doing so, I can hopefully remain more mindful of the theoretical issues others have failed to address.

For the information professional, the importance of planning and thinking beyond the moment and one’s immediate research needs, is an approach increasingly called for in the literature.  Moving research from the more limited practical realm, into a broader consideration of theory can result in better formulated and rigorous findings from which others may more readily benefit.  Referring to an article by Michael Harris, Glazier and Power suggest information practitioners think more “…holistically about problems under consideration.”

Works Cited

Cibangu, S. K. (2013). “A Memo of Qualitative Research for Information Science: Toward Theory Construction.” Journal of Documentation 69(2): 194-213.

Friese, S. (2013). ATLAS.ti 7.1 User Manual. A. t. S. S. Development. Berlin.

Glazier, J. D. and R. R. Power (1992). Qualitative Research in Information Management. Englewood, CO, Libraries Unlimited, Inc. p. 195.

Harris, M. H. (1986). “The Dialectic of Defeat: Antinomies in Research in Library and Information Science.”  Library Trends 34(Winter). pp. 515-31.

Savin-Baden, M. and C. H. Major.  (2013).  Qualitative Research: The Essential Guide to Theory and Practice.  pp. 476-78.

About the Author

Jane article 1 PhotoB. Jane Scales is the Reference Team Leader, and E-Projects Librarian at the Washington State University Libraries. She holds a bachelor’s degree in Russian Language from Indiana University, a master’s in German Language and Literature from Ohio State University, and a master’s in information science (MLIS) from the University of Kentucky. Her research focus includes information literacy, online learning theories, and academic reference services.