Skip to content

January 2010 Workshop

Posted by Administrator on 31 January 2010, 4:06 pm

The January 2010 workshop included a group of four doctoral candidates from the PhD program in Work-based Learning Leadership from The Wharton School of the University of Pennsylvania, a doctoral candidate in Anthropology from Florida International University, another doctoral candidate from Drexel University’s doctoral program in Nursing, and a faculty member from East Carolina University.   The goal of the workshop was to teach the program’s functions of data organization, coding, data exploration, and networking with a particular emphasis on how to conduct a well-managed and organized analysis process.   Participants and instructor agreed that the program allows for so much that it is essential to have in place a system to guide the analyst throughout the analysis process.  This system could rest upon the following set of principles:

a.  Make good use of the code family tool.  All codes should be part of code families, and code families should clearly represent either emergent or a priori themes.   Emergent codes should be inserted into families as soon as they are created.  For this, work with the code family manager always opened so that you can immediately insert the new code into its corresponding family (or create a new code family if necessary).

b.  Be consistent in the use of comments and memos.  Determine what kinds of things will be written in comments and what kinds of things will be written in memos.  The rule is that comments are like footnotes (thanks to Dr. Megan Plyler for suggesting this term) of specific elements within each object of the hermeneutic unit.  Thus, what is written as a comment should only apply to that individual element (e.g., an individual primary document, individual quotation, individual code).    On the other hand,  memos have an applicability (or a potential of) to the analysis project as a whole.  In other words, memos can be descriptions, methodological or theoretical reflections, and interpretations that help to make sense of the data and that contribute to answer the study’s research questions.

An example of the difference between comments and memos can be seen in the case of what is written in relation to a quotation.  A comment of a quotation could be a note about the emotional state of the interviewee when referring to something in particular,  while a memo should be a statement about how that particular quotation helps to understand the research problem.  Of course, it is not always easy to distinguish between what applies to the individual element only and what applies to the research analysis as a whole.  However,  making an effort to be consistent can be very useful, especially when the hermeneutic unit increases in complexity.

c.  Be consistent in terms of the types of codes you use.  It is a good idea to clearly determine  the level of abstraction or generalizability at which you will code.  You may want to determine that what you may think of as broad codes (e.g., participants’ definition of leadership) should instead be a code family, and keep codes to represent the more specific manifestations of how participants define leadership.   Also, the group concluded that it makes sense to periodically check for the quality of the coding structure.  This can be done by always keeping the code family manager (or the code family object explorer) opened.

d.  Try to correlate code families with memo families.  If memos are elaborations, reflections, and interpretations of what is described through codes, it makes sense that they are organized in such a way as to correlate with the themes of the study represented in the code family structure.  This can be done by naming memo families similarly to how code families are named.  Of course, it is essential to be flexible.   Thus, we could well have individual memos or memo families that do not match or correspond to the code structure.  An example of this would be the methodological memos in which the analyst describes and discusses the use of ATLAS.ti in the analysis, or broader methodological issues.

More principles for an organized use of ATLAS.ti will be discussed in later posts.  Thanks!.

Ricardo.

Filed under Coding, Family organization, Memos, Workshops.
2 Comments

Santiago Workshop

Posted by Administrator on 9 January 2010, 7:58 pm

In December 2009,  I taught a 16-hour workshop in the Psychology graduate program of the Universidad Diego Portales, Santiago, Chile.  Participants included psychologists, sociologists, social workers, philosophers, and anthropologists working in universities, non-governmental organizations, and governmental agencies.  Most of them had not used ATLAS.ti before, while others had some previous experience with the software.  Although the group of 14 was diverse in terms of topical areas of interest, they shared a methodological interest in grounded-theory or grounded theory-like approaches to research and analysis.

This workshop was particularly important because it allowed me to reestablish contact with the Chilean community of social researchers using software for qualitative data analysis.  I had conducted a series of workshops there in 2004 and 2005, introducing researchers to ATLAS.ti.   Although the workshop was planned as an introductory course,  we were able to discuss some of the strategies of coding and analysis that I have been developing in my previous workshops.  In particular, we discussed the need to clearly specify levels of generalizability and abstraction in the coding process and to complement these different levels with the use of aggregation tools such as code families, super-codes, and super-families.  I also paid special attention to the integration of free and attached memos, coding, code family organization, and the use of weak links as tools for discovery.   One of the participants,  a philosopher working from the perspective of argumentation theory, was particularly motivated in developing code-to-code and hyperlink relations representing some of the common relationships found in his field.  Thank you to all participants in this workshop!

Ricardo.

Filed under Workshops.
Comment

Code Family Networks as Heuristic Devices

Posted by Administrator on 2 December 2009, 6:16 pm

The code family network is a powerful heuristic device.   Among other things, it helps to visualize findings by theme.  Prerequisites for this approach to work well include having a) well designed and consistent code families representing the study’s emerging themes, b) all codes assigned to corresponding code families, and c) codes representing individual findings.   By examining the code family network, and manually rearranging codes in space in order to better represent categories or sub-themes, the analyst will be able to make sense of what the data is telling.  The description of these findings needs to be immediately written in the network’s comment editor or in a memo.   If this is done well, the code family network will be an effective tool in the discovery of themes and in making sense of data.

Filed under Uncategorized.
Comment

Hermeneutic Circles and Coding

Posted by Administrator on 19 November 2009, 6:58 pm

Neal Rosenburg, doctoral candidate in Nursing, proposed a very interesting idea in the November workshop.  He is using a hermeneutic approach in his research on stigma and HIV/AIDS.   One of the components of this approach is the use of “hermeneutic circles”, whereby the group of students conducting their dissertations under the guidance of the same instructor reflect upon their own texts in reference to their colleague’s texts.   Neal’s point is that the hermeneutic circle can function as a form of quality control in coding.  This is an interesting alternative when there is only one coder and coding is conducted following an openly inductive approach.  Neal, would you like to further develop this point?  Thanks.

Ricardo.

Filed under Uncategorized.
Comment

Primary Document Editing

Posted by Administrator on 15 November 2009, 12:19 pm

In the November workshop we discussed some applications of editing primary documents (RTF documents) within the Hermeneutic Unit.  ATLAS.ti allows the user to edit RTF documents, including adding/deleting text and highlighting text with different colors.  We concluded that an interesting use of the editing tool would be to show different emphases in the speech through colors.  For instance, in in-depth interviews, the analyst could represent with different colors different emotional states the interviewee experiences when telling a story, such as euphoria, sadness, confusion,  and fear.  In this case, the transcriber would have to  note on the transcription when the interviewee experiences changes in emotional state.  If this is not done, the researcher can assign into the Hermeneutic Unit the audio version of the interview and  establish associations  between the audio passage and its corresponding passage in the transcription.  That would allow the researcher to identify emotional changes in the speech, and represent th0se changes in the text through the use of colors.

Filed under Uncategorized.
Comment

November Workshop

Posted by Administrator on 14 November 2009, 11:28 am

Participants in the November workshop included an emergency medicine  medical doctor from Yale University conducting qualitative research on Latino youth traffic safety and injury prevention, a Nursing doctoral candidate from the University of Missouri at St. Louis conducting research on stigma and HIV/AIDS in Africa using a hermeneutic approach, and an economic anthropologist from East Carolina University conducting ethnographic research on transnational migration and work in the U.S., Mexico, and Honduras.   Participants learned to design and implement an analysis project using ATLAS.ti’s basic functions, including coding, writing comments for each object of the Hermeneutic Unit, writing memos, creating weak link and strong link networks, and producing outputs.  We also discussed coding audio, graphic, and video sources of data.  All participants had text data sources (interviews, focus groups, and fieldnotes) and two of them had graphic data sources (pictures and drawings).

A number of issues were discussed during the day and a half of the training.  An interesting discussion had to do with starting the systematic analysis of the text with writing memos instead of coding.   Participants agreed that it makes sense to write introductory memos even before thinking about how to code the data.  These introductory memos can include reflections on assumptions, biases,  theoretical knowledge of the phenomenon under study,  knowledge of the research population, and preliminary interpretations of the study object based on the data collection experience (in the case when the analyst actually collected the primary data).  These introductory memos serve as guides for analysis and interpretation.   We also discussed the use of a-priori codes representing  guiding hypotheses or research questions as boundaries for data exploration.  The group agreed that it makes sense to have these codes (not too many) as a way to set the boundaries for the discovery of meanings in the text.  Through this process of discovery the original guiding hypotheses will be verified or rejected, and new emerging hypotheses will be developed.  Of course,  the group also agreed that the researcher is free to determine how structured coding will be, and that has to do with the methodological approach that guides the study.  In future postings I will further elaborate on these topics and others that were discussed in this training.  Thank you!

Filed under Workshops.
Comment

Low-Abstraction Coding

Posted by Administrator on 26 October 2009, 11:34 am

When coding inductively, my recommendation is to do so at a very low level of abstraction.  That is, emergent codes will be worded in such a way as to represent as specifically as possible the situation or process that it is being coded.  As a result of this approach, each code will have a low level of generalizability (i.e., it cannot be applied to many cases).  The advantage of this approach to coding is that each code will directly represent a given finding, making it unnecessary to reexplore the HU’s quotations in search for specific findings, something that should be done when coding with more abstract codes.  Of course, one of the consequences of this approach is that we will end up with a very high number of codes.  This fact, per se, is not necessarily negative.  It would be a handicap if codes are not organized in a coherent way using the code family tool.  Indeed, my recommendation is that newly created codes be immediately assigned to code families.   In order to do that, just keep the code family manager opened while coding and assign codes to families as soon as they are created.  Following this approach, code families can represent emergent themes or findings,  all of which are illustrated by the individual codes.

Filed under Uncategorized.
2 Comments

Thematic coding

Posted by Administrator on 22 September 2009, 5:13 pm

ATLAS.ti organizes codes alphabetically.  This means that when you open the code list you will see codes ordered alphabetically and not in any hierarchical structure.   This form of code organization may result problematic when you have a long list of codes and you want to code according to themes.  How can you solve this problem?  As always, code family organization is a critical step in data management with ATLAS.ti.   If you have good families, families that make sense from a thematic point of view, then you can code using the object explorer function (…tools/object explorer).  Once you open the object explorer, open the code families tree and keep it opened.  Then, when coding just look for the code family that best represents the theme you are looking for.  Once you find it, open that family and select the code(s) of your choice.  Following, just drag the code to the text and drop it anywhere in the main window.  Of course, a necessary previous step is to select the segment of the text you want to code.

Again, this way of coding requires that you have a solid code family structure.  In future postings I will discuss some strategies to create strong code families that make sense from a thematic point of view.  Thanks!

Ricardo.

Filed under Coding, Family organization.
Comment

Memos as Spaces for Preliminary Reflection

Posted by Administrator on 17 September 2009, 9:26 pm

One of the issues discussed during the September workshops had to do with using memos as spaces for a priori reflections about the study.  These are reflections that take place before actually starting the systematic analysis of the data.  What kinds of things can be written is such memos?  Assumptions, biases, theoretical reflections, and anything else that might seem worth making explicit before exploring the data.  One of the workshop participants talked about writing in a priori memos what she learned while visiting the community for the first time, even before starting to collect data.  Another participant talked about making explicit in these memos her assumptions based on the review of the literature and her general (or specific) knowledge of the object of study.   The interesting thing about writing memos making theoretical assumptions explicit is that they can be contrasted with emerging findings and/or used to guide later interpretations of those findings (also in the form of memos).

Ricardo.

Filed under Memos.
Comment

Themes From the August and September Trainings

Posted by Administrator on 15 September 2009, 11:13 am

Participants in my August and  September workshops included a public health researcher, a clinical psychologist doing her post-doctoral work, a sociologist of religion,  an anthropologist working in Africa on micro-finance projects,  and two professors of the Japanese language.   They all do great research using qualitative and mixed-methods methodological approaches.  We had the opportunity to discuss the application of ATLAS.ti to rapid ethnographic assessment instruments, focus group research,  ethnographic research closed to the grounded theory tradition, and survey research.

One of the things we discussed in all three workshops had to do with how to approach the need to check for inter-rater reliability in qualitative research using ATLAS.ti.  One of the consensus was that checking for inter-rater reliability is not a necessity from a qualitative/interpretive perspective, but that it becomes almost a requirement when using qualitative methods within the context of quali-quanti research teams in which reporting numbers is a priority.  It may also become a necessity when funders require numbers to prove a set of pre-existing hypotheses and utilize qualitative research to provide accounts to support those hypotheses.

In the September 12-13 workshop, we concluded that inter-rater reliability can be checked with ATLAS.ti when using a theoretical coding strategy (deductive/more structured than unstructured) and in that case operational definitions of codes are critical.  This requires that each code be clearly defined jointly by the coding team using the comment editor.  However, we also concluded that using a grounded  coding strategy (inductive/more unstructured than structured),  a strategy in which discovery of meanings not verification of pre-existing hypotheses is the goal, checking for inter-rater reliability can be nonsensical,  a distortion of the nature of qualitative/interpretive research.   The group agreed that, luckily enough, ATLAS.ti allows for both approaches.

Other interesting issues discussed in these workshops included: triangulation diaries-interviews-focus groups-photographs; the critical role of memos as spaces for description, analysis, and interpretation (following Harry Wolcott’s data transformation model); and the importance of family organization in terms of primary documents, codes, and memos to allow for rich and well integrated analysis.

Finally, a lesson learned from these three workshops is that teaching ATLAS.ti is never a purely technical matter.  Indeed, it is always methodological, it always requires a rich discussion of the  reasons why (not only the means).

Thanks!  Ricardo.

Filed under Inter-rater reliability, Workshops.
2 Comments
« Older Entries