66 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 3.40 Keyword in context search results • Double-click to show the search hit in the context of the document. If appropriate, you can now begin to code the data. See Chapter 4 for further detail. Figure 3.41 Viewing search hits in the documents You can also access the project-wide search from the ribbon: • Select the Search Project tab. • Enter ‘time’, for instance, as a search term. As you can see in Figure 3.42, I selected to search only in Documents and only the document content. When you open the Search tool, all options are activated. You can either uncheck all options that you don’t want or click on ‘Uncheck All’ and then select where you want to search.
EMBARKING ON THE JOURNEY – DATA AND PROJECT MANAGEMENT 67 Figure 3.42 Project search settings • Click on the Search button to run the search. REVIEW QUESTIONS 1 Which data file formats can you analyze with the support of ATLAS.ti? 2 When working with multimedia files, which formats do you need to choose when working across platforms (Windows/Mac). 3 When you want to analyze video data, what do you need to pay attention to when preparing them? Which options do you have when adding them to a project? 4 Where does ATLAS.ti store your data? 5 What are the reasons to change ATLAS.ti’s default location for storing data? 6 How do you set up a single-user project? 7 How can you organize project data? 8 How do you create a backup of a project? 9 What is a project snapshot? How can it be created and what can it be used for? 10 How do you transfer a project to a different computer? 11 How do you prepare and import survey data? 12 How do you prepare and import data from reference managers? 13 How can you create word clouds and for what purpose? What can you do to optimize the information you get from word clouds? What are possible next steps in the analysis? 14 Why is writing so important when analyzing qualitative data? 15 What type of memos can you begin to write at the beginning of your project? 16 What is the difference between comments and memos? When will you be using comments and when will you be using memos? FURTHER READING Birks, Melanie, Chapman, Ysanne and Francis, Karin (2008). Memoing in qualitative research: Probing data and processes. Journal of Research in Nursing, 13(1), 68–75. Charmaz, Kathy (2014). Constructing Grounded Theory: A Practical Guide through Qualitative Analysis. London: Sage. Di Gregorio, Silvana and Davidson, Judith (2008). Qualitative Research Design for Software Users. Maidenhead: Open University Press.
68 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Gibbs, G. (2002). Writing as analysis. Online QDA: http://onlineqda.hud.ac.uk/Intro_QDA/ writing_analysis.php. Richardson, Laural (2003). Writing: A method of inquiry, in Hesse-Biber, Sharlene N. and Leavy, Patricia (eds) Approaches to Qualitative Research: A Reader on Theory and Practice, chapter 22. Oxford: Oxford University Press. Saldaña, Johnny (2015). The Coding Manual for Qualitative Researchers, chapter 2. London: Sage.
Technical aspects of coding 4 ‘The only reason why we think of categorical thinking as more logically compelling is that we feel more at home in it, as something of our own making. It is our instrument for coming to grips with what we think of as essential aspect of being.’ Blanchette (2003: 118) Categorizing is something we do every day, and we learned it early. A child learns that there are humans, animals and plants. Some humans are family members; others are friends or strangers. Dogs and other animals can be distinguished by breed. There are golden retrievers, German Shepherds, terriers or boxers. Independent of breed, they can be perceived as best friends, playmates or threat; they can be guard dogs, service dogs or be seen as members of the family. Coding your data with ATLAS.ti is not much different from this. According to Freeman (2017), the main goal of categorization is to tag things to define or organize them. In the process of categorization, we compare data segments and look for similarities. All similar elements can be grouped under the same name. By naming something, we conceptualize and frame it at the same time. As shown in the above example, dogs can be categorized on the basis of their breed but also on the basis of their duties or what they mean to us. However, before you can begin to code data in a meaningful way, you first need to learn the technical aspects of it – a skill for your onward journey. Coding is a core function in ATLAS.ti that lets you ‘tell’ the software where the interesting things are in your data. Based
70 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI on this you will learn further skills, like developing a system for all the interesting things you observe: how to write notes on them, how to examine your data in depth, how to see patterns and relations using the various advanced analysis functions and how to write up your major insights and findings on the way. But this is for later chapters. First, let’s concentrate on the technical aspects of coding. Coding in a technical sense simply means assigning a label to a data segment. A better-known term these days is tagging. The goal of tagging is to find the things you tagged using the tag name. The software uses the words ‘code’ and ‘coding’, as almost all the other CAQDAS do because of the popularity of grounded theory at the time when the first programs were developed in the late 1980s and early 1990s. Coding in CAQDAS, however, is very different from grounded theory coding in the methodological sense (Friese, 2016b). I will discuss this in more detail in Chapter 5. If you are more comfortable with the idea of tagging, in what follows simply replace the terms ‘code’ and ‘coding’ in your mind with ‘tag’ and ‘tagging’. The data segment that you can code (or tag) can be as small as one character in a text document, a few pixels in an image file or less than a second in an audio or video file. In this chapter, you will learn the different ways of applying codes to text and multimedia data, modifying the length of a coded segment, renaming, deleting and merging codes and writing definitions for codes. To explain the various options, I will use the project bundle with the name Children & Happiness sample project (chapter 4) that you can find on the companion website. You can use it for the exercises, too, or you can use the project that you created when working through Chapter 3 or your own data set if it is not yet coded. Questions of how to build an efficient coding system and how to code to best utilize the advanced functions of the software are all discussed in Chapter 5. LEARNING OBJECTIVES When you have worked through this chapter thoroughly, you should know the procedures of coding with ATLAS.ti. That is: • How to apply codes to several types of data. • How to resize quotations. • How codes can be renamed, merged or removed. • How to write definitions for codes and why this is important. • How to export lists of codes and how to prepare a code book. • How to import an existing code list using the Excel import function. • How to handle PDF, image, audio, video and geo documents. • How to make use of quotation names and comments.
TECHNICAL ASPECTS OF CODING 71 SKILLS TRAININGS Skills training 4.1: coding with a new code Skills training 4.2: coding with two or more codes Skills training 4.3: list coding Skills training 4.4: coding via drag and drop Skills training 4.5: replacing a code Skills training 4.6: resizing the length of a quotation Skills training 4.7: unlinking and removing codes Skills training 4.8: writing quotation comments Skills training 4.9: coding with in-vivo codes Skills training 4.10: further coding-related options Skills training 4.11: importing a list of existing codes Skills training 4:12 exporting the code list for use in another project Skills training 4.13: focus group coding Skills training 4.14: coding a PDF document Skills training 4.15: working with audio and video files Skills training 4.16: coding an image Skills training 4.17: working with geo data For the exercises in Skills trainings 4.1–4.8 you will work with a few test codes that you will simply label test 1, test 2, test 3, test 4, A and B. Content does not matter at this point. The only exception is when I explain the use of in-vivo codes. SKILLS TRAINING 4.1 CODING WITH A NEW CODE • Open your own project that you created while working through Chapter 3. Alternatively, you can also use the Children & Happiness sample project for this chapter that you will find on the companion website. • Load the document D3: Belkin’s parenting blog discussion. • Select a text segment with the mouse, right-click and select Open Coding from the context menu. • Enter a name (here I’ve used ‘test 1’) and click Enter (see Figure 4.1). You have created your first quotation and code. Each is displayed both textually in the Project Explorer, the Quotation and Code Manager and graphically in the margin area. A bar marks the size of the coded segment (= quotation) and the code name appears next to it (Figure 4.2).
72 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.2 Display of codes and quotations in the margin area and the Project Explorer The separation of quotation and code has many advantages. As they are independent entities, you can comment on each code and on each quotation. Quotations can be linked not just to codes but also to each other and to memos. This is a prerequisite for the hyperlink function discussed in Chapter 7. It allows you to work directly at the data level without necessarily using codes. Let’s code three more segments using this method: • Highlight a different text segment, right-click, select the coding option from the context menu and enter ‘test 2’ as code name. • Code two more segments using ‘test 3’ and ‘test 4’ as codes. Overlap some of your codings and observe what happens in the margin area. Next, before showing you more variants of coding, I would like to explain how quotations and codes are referenced. Let’s look at the Project Explorer on the left-hand side (Figure 4.3). Expand the branches for Documents and Codes: Figure 4.1 Coding a data segment with a new code
TECHNICAL ASPECTS OF CODING 73 Figure 4.3 Project Explorer after having done some coding Quotation reference Under the main Documents branch, quotations are displayed on the first level and if coded, the codes are displayed underneath. When a quotation is created, ATLAS.ti automatically assigns an identifier to it. This identifier is built from the index of the document to which it belongs plus the first 70 characters of the text segment – for example, ‘3:2 However, the first year of motherhood was rough. I was only 25 and bec...’. In the Project Explorer only the first few words are displayed. For graphic, audio and video segments, the original file name of the document is chosen as identifier. The ID 3:2 means that the quotation is from document 3 and is the second one that was created in this document. The reason for the chronological numbering is to do with the fact that you do not necessarily code a document from the first line to the last. You will jump between passages and modify or delete some quotations during the coding process. A linear numeration would have to be updated with every single quotation that is inserted, which would take up unnecessary computational capacity. If you prefer your quotations to be numbered sequentially, open the Quotation Manager and sort all quotations by their start position. Then choose Renumber Quotations in the Quotation Manager’s ribbon. This can be useful, for example, when coding open-ended questions from survey data and you want to keep the cases in synch with the cases in the statistics program you are using. Start and end positions. The figures in brackets after the quotation name show the location (start and end positions) in the document. Currently the start and end positions
74 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI in text documents are given by number of characters – for example, (474:1034) or (96:362). When text editing becomes available again, paragraph numbering will be used as in previous versions. Do you remember what I said in Chapter 3, when I discussed good-practice rules for preparing transcripts? My suggestion was to break larger speech units into smaller paragraphs because of the way ATLAS.ti referenced quotations. However, ATLAS.ti might not reintroduce paragraph numbering until version 9. For image files, the reference consists of the coordinates of the upper-left and lower-right corners of the marked rectangular area. Audio and video quotes use a time reference (start position and duration). References for PDF quotations consist of page number and number of characters on the page for the start and end positions. Note that all PDF documents in ATLAS.ti start with page 1. When you add a book chapter that in the book begins on page 35, in ATLAS.ti page 35 will be page 1, page 36 will be page 2 and so on. Density. The density column in the Quotation Manager indicates the number of linked quotations, codes and memos. Figure 4.4 Quotation properties The quotation ID in combination with the start and end positions can be used when citing quotations in a report. For example: ‘I was happy before I had kids and am happy now. However, the first year of motherhood was rough. I was only 25 and becoming a mom forced me to grow up.’ (D3:1; start position:end position) The reference allows the reader of a report to trace the quote and to find it either in the original digital material or in a printed version of the coded data. Code reference All codes are listed under the main Codes branch in the Project Explorer. You can recognize codes by the green diamond symbol. The number behind the codes {1-0} means that the code has been applied once and that it has not yet been linked to other codes. If you open the Code Manager or the code list in the navigation area, you also find the terms ‘groundedness’ (how often a code is applied) and ‘density’ (the number of linkages to other codes).
TECHNICAL ASPECTS OF CODING 75 The density remains 0 until the researcher manually links codes to other codes, mostly in later stages of the analysis if relationships between codes are established. The density is not a value that is calculated by the software. It goes up when the researcher begins to link codes to each other (see Chapter 7). SKILLS TRAINING 4.2 CODING WITH TWO OR MORE CODES Let’s code a new segment with the two codes A and B: • Highlight a data segment. Right-click and select Open Coding. • Enter ‘A’ on the first line and ‘B’ on the second line. Click Create. The quotation is now coded with the codes A and B. Figure 4.5 Coding with two or more codes SKILLS TRAINING 4.3 LIST CODING If your code lists get longer, you won’t always want to scroll to find the code you are looking for. Also, you want to be reminded which codes you already have. The Codings dialog also offers list coding. If you start typing, the list of codes will be immediately filtered and only show you codes that contain the letter(s) you entered in the entry field: • Highlight a data segment, right-click and select List Coding. Type the letter ‘t’. The list of codes shortens and only your test codes are displayed (Figure 4.6).
76 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.6 List coding Another option to quickly filter the list of codes is to use the search field for the code list in the navigation panel or the Code Manager (see the next Skills training). SKILLS TRAINING 4.4 CODING VIA DRAG AND DROP At the beginning of the coding process, it is normal to generate a lot of new codes. But after a while, you will find similarities and repetitions in the data and want to apply codes that already exist. Your sample project should currently contain six codes. If you want to reuse them, you can either drag and drop them from the Project Explorer, the Code Browser or the Code Manager. If you want to code from the navigation panel, the recommended option is to use the Code Browser, as it only contains the codes. If your list of codes gets longer, it is easier to enter the first letters of a code into the search field rather than to scroll the list up and down. The advantage of using the Code Manager for drag-and-drop coding is that (1) you can edit the code comments (= code definitions), and (2) you can quickly access subsets of codes using code groups. When using smaller screens, I recommended docking the Code Manager and moving it into a separate tab group (see Chapter 2). Figure 4.7 Drag-and-drop coding to apply existing codes
TECHNICAL ASPECTS OF CODING 77 • To use drag-and-drop coding highlight a data segment, select one or more codes in the above-mentioned lists and drag the code onto the highlighted data segment (Figure 4.7). For text documents, there is no need to drop the code in the highlighted area. It just must be dropped somewhere on the document. Whatever is highlighted gets coded. Intuitively one often drags the code on top of the highlighted data segment. If you spend many hours coding, this gets very tiring on your eyes and requires more and more concentration, and the constant twisting action of your wrist in trying to hit the highlighted area may even strain the muscles. The software does not require you to do this when coding text documents. Thus, be gentle on yourself. You will find out over time and with more practice which option you prefer – to code from right to left or from left to right, or maybe make use of both options depending on the task at hand, or to use list coding. For regular coding, I prefer to code using the Code Manager because I like to use the group side panel to quickly navigate to a certain code group and its codes. If I refine my coding system and work on just one category to develop and apply subcategories, I prefer the Code Browser. It is quite common for more than one code to be used for coding a data segment. The segment might contain ambiguous information or different layers of meaning. You may want to code for content and for context, use of language or aspects of time. It is difficult and not very efficient to incorporate a variety of aspects into a single code name and it should not be done (see Chapter 5 for more information). You can later rely on the computer to find all kinds of combinations for you. SKILLS TRAINING 4.5 REPLACING A CODE If you want to replace a code (perhaps because you have changed your mind, found a more fitting code, used a code by accident or want to develop subcategories) then the easiest way is to use drag and drop. This time, however, you cannot drop the code just anywhere; you must drop it exactly on top of the one that you want to replace. • Let’s assume you want to replace the test 4 code in Figure 4.7 with test 2. Click on the ‘test 2’ code in the Code Manager or Code Browser, hold down the left mouse button, drag the code to the ‘test 4’ code in the margin area and drop it. Voilà – the code is exchanged. SKILLS TRAINING 4.6 RESIZING THE LENGTH OF A QUOTATION Resizing the length of a quotation is also quite a common procedure. Maybe you discover that a chosen segment is too large or that you forgot something and need to extend the length of a quotation. Both ways of modifying are possible. • Select the quotation by clicking on the quotation bar or code in the margin area and move the ‘handle’ in form of little orange circles to the right, to the left or up or down, depending on whether you want to shorten or lengthen the quotation (Figure 4.8). This applies to all media types expect geo quotations, as geo quotations have only one locator.
78 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.8 Resizing the length of a quotation SKILLS TRAINING 4.7 UNLINKING AND REMOVING CODES Just as you can erase a pencil mark in the margin of a paper document, you can ‘erase’ a code in the margin area on screen as well. The digital equivalent is called unlinking. It is mostly used when a segment is coded with more than one code. • To unlink a code, right-click on the code in the margin area and click on the option Unlink. When you unlink a code from a segment that only has one code, the bar for the quotation remains in the margin. If you want to ‘erase’ both the code and the quotation, you need to proceed as follows: • Mark the coded segment to be removed (e.g. by clicking on the code in the margin). • Move your pointer over the highlighted quotation, right-click and select Delete from the context menu. Note that removing a coded segment does not delete the text or other data material that you have coded. The quotation is an entity stored in the project file. You can think of all entities stored in the project file as a layer on top of your data material. The project file only accesses the data via a reference and loads it into the editor. The data themselves are not touched.
TECHNICAL ASPECTS OF CODING 79 SKILLS TRAINING 4.8 WRITING QUOTATION COMMENTS If you notice something interesting while coding and you want to write it down, use the quotation comment field. A lot of ATLAS.ti novices attach a memo instead of using the comment field. Memos, however, are more than mere comments (see Chapters 3, 5 and 6). Comments become part of the entity you write them for. Memos can either be freestanding or linked to one or more quotes, codes or other memos; they have a title, can be classified by type and you can also write a comment for each memo. When you want to comment on a data segment, all you need is the quotation comment. Using memos for this purpose means not taking advantage of the real potential of the comment feature. The fact that ATLAS.ti differentiates between comments and memos allows you to use different tools for analysis at different levels. What often happens when users use memos instead of comments is that they write the comment in the memo title. The memo itself remains empty. This makes it difficult to manage the list of memos because the memos do not have proper titles. Ultimately, users reject the memo function and find it not particularly useful. When you want to comment on a data segment: • Right-click on the quotation bar in the margin area, select the option Edit Comment from the context menu (Figure 4.9). • An editor will pop up. Write a comment on the selected data segment (Figure 4.9). • Save your comment by clicking on the Save button or use the key combination Ctrl+S. Close the editor. Figure 4.9 Writing a quotation comment After you have saved the comment, you will see a yellow Post-it note inside the blue quotation bar in the margin area and on the quotation icon elsewhere in the program. If you use an interpretive approach to analyze your data, you are likely to write lots of quotation comments. In this case, I recommend that you open the Quotation Manager, move it to a new tab group to the right of the document you are working on (see Skills training 2.6) and use the comment field in the Quotation Manager. In Chapter 5, I have prepared a skills training for this type of analysis, so you can practice it (see Skills training 5.6).
80 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI SKILLS TRAINING 4.9 CODING WITH IN-VIVO CODES The term ‘in-vivo code’ can be traced back to grounded theory methodology. In-vivo codes, according to Corbin and Strauss, are ‘concepts using the actual words of research participants rather than being named by the analyst’ (2008: 65). Technically, this means that when you highlight a text segment, the first 40 characters of the segment itself are used as a code label. In-vivo codes are especially useful at the beginning of the coding process when collecting ideas. As it is quick and easy to create them, this is what they are often used for in computer-assisted analysis. This first collection of ideas mostly takes place at the descriptive level. Description, again according to Corbin and Strauss (2008), is the basis for more abstract interpretation of data and embodies concepts, at least implicitly. Proper concept building, however, needs to happen at some point of the analysis. It does not make much sense to collect 200 in-vivo codes without developing them further. You just end up with a very long code list. With progressive analysis, some of the in-vivo codes used as a quick device for collecting things that you notice in the data should be merged and others renamed in more abstract terms (see the next skills training for how this works). I have seen projects gone mad – the champion being a project with over 16,000 codes. I will tell you in Chapter 5 what has gone wrong if you end up with thousands of codes. For now, as a rule of thumb, remember that a project should only have between 120 and 300 codes. There are projects that have fewer codes and others that have more, but 1,000 codes are in most cases already too many. Now let’s turn to the technical practicalities again. • Mark a text passage. Right-click and select the option In Vivo Coding from the context menu. Figure 4.10 shows an example based on the Children & Happiness project, where some text segments are collected in the form of in-vivo codes. The ideas that I collected have to do with positive effects of parenthood. Figure 4.10 Coding with in-vivo codes After collecting some more instances and reviewing them, I could think of better and shorter names as code labels. These first ideas could, for example, be developed further into a category called ‘Effects pos. of parenting’, with subcategories like ‘personal growth’,
TECHNICAL ASPECTS OF CODING 81 ‘fulfillment’ or ‘life purpose’. This would also require extending quotation boundaries to include more context. I generally do not recommend using the in-vivo coding option as there is a danger of generating too many codes too quickly and on the wrong level. If you want to work interpretively and close to the data, create quotations instead without applying codes. See Skills training 5.6 for further detail. SKILLS TRAINING 4.10 FURTHER CODING-RELATED OPTIONS There are many other options in the Code Manager, which you can access via the ribbon or the context menus. For instance, you can create new codes and rename, delete or merge codes. The splitting function will be explained in Chapter 5. Creating new codes • Click on the Free Code(s) button in the ribbon of the Code Manager or click on the drop-down arrow in the button New Entities and select New Code(s). • Enter a code name in the first entry field, and a second and third in the fields below (Figure 4.11). Then click on Create. Frequency and density are both zero in the code list as the codes are not yet linked to anything. Figure 4.11 Entering multiple codes at once
82 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Renaming codes The renaming option is a global option. This means that renaming a code word in the Code Manager affects all coded segments that use this code. • Mark a code word in the Code Manager, right-click and select Rename from the context menu (or click F2). Adding a color attribute to codes If you want to add color to your codes: • Select one or more codes in the Code Manager and click on the drop-down arrow of the Change Color button in the ribbon. • This opens a color palette. Select a color. Figure 4.12 Set code colors The color attribute is shown by a colored circle in front of the code label. This color is also used for code nodes in networks (see Chapter 7). Deleting codes (and other entities) The delete option also has global effects, at least within the boundaries of the project. This means that deleting a code removes it from the entire project and from everywhere in your documents, code groups or networks – anywhere it was used. This is how: • Select one or more code labels in the Code Manager, right-click and select Delete from the context menu. You can also click on the Delete Code(s) button in the ribbon of the Code Manager.
TECHNICAL ASPECTS OF CODING 83 Writing code definitions Comments can be written into the white field below the windows splitter, as we have seen in Chapter 2. When entering a comment (for codes this is in most cases a definition, a coding rule and possibly a sample quotation), the commented entity shows a Post-it note (see Figure 4.13). • Try it out: write a comment for one of your codes. When you are finished, click on another code: the definition is saved, and you will see a Post-it icon attached to the code icon and a tilde (~) after the code name. • If you double-click on a commented code in the margin area, an editor with the comment opens. Writing code definitions is, analytically speaking, a very important process (see also Fielding and Lee, 1998: 94ff.; Guest et al., 2012: 52–4). When I look at team projects, code definitions are never missing because it seems self-explanatory that in teams one must write down what each code means. This is not the case when analysts work alone. They typically reason that writing code definitions is not necessary, since they think they know what their codes mean. I can tell you that this is not always the case and that you cannot just take it for granted. What happens if you are ill and cannot work on your project for some time, or if you go on vacation, fly to a meeting or have other commitments that keep you away from your analysis? Returning to your data, you sit down, click on a code and ask yourself, ‘What is this code doing here? What was I thinking of when I created it?’. You try to remind yourself by browsing through the quotations and you begin to remember – but if only you had written a code definition at the time or at least some comments… So, you see, it does help to put some information into the code comment field. There are other benefits as well. Coding is a process and your thoughts about the data will evolve. The meaning you attached to a code at the beginning of an analysis may change over time. There is nothing unusual about that. If there is no definition for the code, you may not notice the subtle changes over time and in the end the code may have quotations that in fact no longer fit under a common label. ATLAS.ti makes it easy for you to track changing meanings or associations, since you see the code definition right in front of you in the Code Manager or Project Explorer as you use it. Furthermore, going through all your codes and defining them helps the analytic thought process. It forces you to draw clear-cut boundaries between codes. If you realize there are overlapping meanings, you can either merge the codes or make them more distinctive. Figure 4.13 Display of code with comment
84 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Code definitions also add transparency to your analysis. They are necessary for a third person to understand and follow your analytic ideas. Try it out, even if it means some extra work. I promise you will see that it is worth doing. Merging codes The Merge Codes function can be used to combine two or more codes with each other. This is a common procedure when you begin to clean up your code list after initial coding or when you work in a team and merge sub projects. You may realize that there are two codes that have different names but mean the same thing. Thus, the contents of the two codes, the coded quotations, can be combined under one code name. • Select all codes in the Code Manager that you want to merge. • Right-click and select Merge Codes from the context menu. • A window opens that lists all the codes that you have selected. Choose the one code that you want to keep and click on the Merge Codes button. All quotations of all other codes will be moved to the selected code. The labels of all other non-selected codes will be removed from the code list. This is a desired effect and nothing to worry about. You are not losing anything. On the contrary, you gain more analytic precision, and you move from the descriptive to a more abstract level of analysis. This is a prerequisite to use the further analysis tools of ATLAS.ti. As you can see in Figure 4.14, before merging, both codes, ‘test 1’ and ‘test 4’, coded two quotations. After merging, ‘test 1’ contains (2+2) four quotations. However, the numbers do not always add up like this; for instance, it would not be the case if both codes had been applied somewhere to the same quotation. If the codes have not yet been defined, you won’t be able to see later that the codes were merged unless you enter a comment manually. You can do this if it is important to leave an audit trail to trace your steps. Figure 4.14 The process of merging
TECHNICAL ASPECTS OF CODING 85 SKILLS TRAINING 4.11 IMPORTING A LIST OF EXISTING CODES If you already have a list of codes that you want to use in your project, you can create and import an Excel file with these codes, including code descriptions, code colors and code groups. This is a useful feature for multiple scenarios. You may already have some ideas for codes derived from the literature and may not want to enter each code individually in ATLAS.ti. You are a teacher and want to give your students an already prepared code list that they can use. You work in a team and have discussed in advance which codes you want to use and have already created a table with codes and their definitions. You want to reuse a list of codes from an earlier project. If you are developing ideas on paper or in an Excel spreadsheet in advance, be careful not to force the data into these codes when you start coding. Stay open-minded. If there is no matching code, add new codes or adapt existing ones. This is how you need to prepare the Excel file for import in ATLAS.ti: Figure 4.15 Preparing a code list for import in ATLAS.ti You can enter headings like Code, Code Definition, Code Group 1 and Code Group 2, but you do not have to. If you do not enter headings, the columns are interpreted in the following order: column 1: code name column 2: code description (comment) column 3: code group column 4: code group all subsequent columns: further code groups If you color the code name, this color is used in ATLAS.ti as code color. • To import the Excel file, select the Import & Export tab in the ribbon and then the Codebook (XLSX) button from the Import gallery. Activate or deactivate the option ‘My data contains headers’, depending on what your Excel file looks like.
86 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI SKILLS TRAINING 4.12 EXPORTING THE CODE LIST FOR REUSE IN ANOTHER PROJECT You can export your list of codes for different purposes. If you want to create a code book, the report should be nicely formatted. If you need your list of codes with definitions and code groups for another project, the formatting does not matter. What is important is that all information is transferred 1:1 to another project. In Skills training 8.4, I show you how to export your list of codes as code book for your report. If you want to export the list of codes with definitions, colors and code groups for reuse in another project: • Click on the Import & Export tab and select Codebook (XLSX) from the Export gallery. • Save the Excel file to your computer, on a server, in the cloud or on an external storage media. Figure 4.16 Exporting a code list for reuse In case you are wondering, the QDC format is an international standardized format for exchanging code lists and, from March 2019, for fully coded projects between different CAQDAS. You will find further information at www.qdasoftware.org/. SKILLS TRAINING 4.13 FOCUS GROUP CODING In the analysis of focus group transcripts, a first task is often to encode all speaker units. This is necessary if you later want to compare the answers of the various participants either on a case-by-case basis or by attributes such as gender, age and occupation. ATLAS.ti can do this job for you if you prepare the transcripts in such a way that each speaker can easily be identified. This means each speaker needs to have a unique identifier and, if possible, this identifier should be applied consistently throughout the entire transcript. ATLAS.ti offers two standard patterns for identifier, and if you have not yet prepared the transcript, I recommend using one of them. You can also define your own pattern or use GREP expressions in case there are typos or an inconsistent use of speaker or other identifiers. For instance, the regular expression: To+(m|n) matches ‘Toom’, ‘Ton’, or ‘Toon’ to find different spellings of the name ‘Tom’. The two standard patterns are as follows:
TECHNICAL ASPECTS OF CODING 87 Pattern 1: <speaker name:> Pattern 2: <@speaker name:> An example transcript following pattern 1 might look like this: Figure 4.17 or like this: Figure 4.18 It is also possible to add further information to each speaker like their gender, age group, educational level, etc. This way information will be automatically coded. Here is an example: Figure 4.19
88 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI The speaker unit of Alex will then be coded with Alex, gender male, age group 1, education level high school. Before the data are coded, you can change the code labels or add further codes. Recommendation. For readability, I recommend starting each speaker unit on a new line and entering an empty line between speaker units, but neither is required. When a pattern is recognized, the chosen code(s) is applied from the first letter of the pattern to the start of the next recognized pattern. Therefore, it does not matter whether a new unit starts on a new line or whether there is an empty line in-between. I recommend using the pattern ‘@speakername:’. This way, all speakers will be listed together in the code list as their codes all start with the prefix @. Further, I recommend adding the attribute codes for each speaker manually when auto coding the speakers and not into the transcript. Coding focus group data You can either use your own data for the following exercise or download the focus group sample documents from the companion website. • Create a new project and add the focus group transcript(s). • Right-click on a focus group transcript and select the option Focus Group Coding. • Select a pattern for recognizing speaker units and click Next. ATLAS.ti lists all identifiers that fit the selected pattern and adds the name(s) as code. These might also include items that you would not classify as speakers but that fit the pattern. If so, it is easiest to deselect all items and to mark only the speakers. Figure 4.20 Choosing speakers to code (pattern 1)
TECHNICAL ASPECTS OF CODING 89 If you plan to compare the answers of male and female respondents or, as in the example, the answers of teachers teaching different languages, you can add further codes. All codes must be separated by a semicolon. Since the speaker codes are a different type of code, I use special characters like @ or # as the first letter in the code name. In the code list, these codes are at the top, because the list is sorted alphabetically. If you use pattern 2 for your speaker IDs, all speaker codes already start with @, so you do not have to add it manually. Figure 4.21 Adding codes for each speaker unit Figure 4.22 Results of focus group coding
90 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI • Add more codes as needed and click on the button Code. When ATLAS.ti is done coding, a window pops up telling you how many codings have been created. (A ‘coding’ is the link between a code and a quotation.) • Open the Code Manager and load the focus group transcript to inspect the results. An example is shown in Figure 4.22. HANDLING OTHER MEDIA TYPES Now I will explain the differences in handling other file formats. The differences mostly relate to selecting data segments and the information included in reports. Other than that, the different data file formats are smoothly integrated into all functions of the software. When you click on a code that codes data in different formats, all quotations are listed and you can browse through them one by one. You can see a preview of the quotations in the quotation list or view them in the document context: audio and video quotations are played, coded image quotations are highlighted within the image file and the position of geo quotations is shown in a map. When you create reports of coded data segments, you can save them as rich text, PDF or an Excel file. Image quotations are included in rich text and PDF reports. In the future, reports might also hold the first frame of a video quotation. For audio and geo quotations, only the quotation reference is displayed in the reports. You can enrich this output by making extensive use of the option to rename the quotations and to write comments. SKILLS TRAINING 4.14 CODING A PDF DOCUMENT You can continue to work with the Children & Happiness project for the following exercise. • Load D1 by selecting it in the Project Explorer. Resize the editor if needed. When selecting a text segment in PDF documents, you need to place the cursor directly to the left of the first letter. If you place it too far to the left of the text, you will select a rectangular graphical area instead of the actual text segment. • Select a string of text and use any of the coding techniques described above (e.g. via the context menu or coding via drag and drop). • Try coding a few text passages in the PDF. • Now experiment with selecting a graphical image in the PDF. Select, for instance, the title and the subtitle and the box around them. • Modify one of your coded segments in the PDF document. SKILLS TRAINING 4.15 WORKING WITH AUDIO AND VIDEO FILES Below, I describe how to work with video files. The instructions also apply to working with audio files; the difference is that instead of the video image you see a white space.
TECHNICAL ASPECTS OF CODING 91 Adding audio/video files to a project As audio and video files can be quite sizable, you have the choice to link them to a project instead of adding them. This means they stay at their original location and are accessed from there. Preferably, these files should not be moved to a different location. If the files need to be moved, you need to re-link the files to your project. ATLAS.ti will alert you if there is an issue and a file can no longer be accessed. You will find a Repair Link button in the Document Manager under the Utilities tab. Although it is possible to code an audio or video segment straightaway in the same manner as coding a text or image segment (you highlight a section, right-click and select the option Open Coding), when working with video files I prefer to set the quotation first and then think about which code I want to attach. Video files are richer in information and setting the correct start and end positions is often not as clear-cut as selecting a quotation in text files. In addition, when working with audio and video files, I recommend adding descriptive information in the form of segment titles and comments in the Quotation Manager (see ‘First steps in analyzing video data’, below). Therefore, the quotation itself becomes more prominent. Display of video documents • Import the sample project video Iceland for Skills trainings 4.15–4.17 from the companion website or add a video file to your own project. For the latter, I recommend linking the video file to the project rather than importing it. See Skills training 3.1 for further information. • Load the video document. On the right-hand side, preview images are displayed. When you add a new video to a project, you do not at once see the preview images because they first need to be created. The size of the preview images is up to 50 × 50 pixels. To create these images, one frame per second is selected. Depending on the length of the video, this may take a few seconds or up to a few minutes. The media controls that allow you to start, stop and pause the video, and to skip forwards and backwards, can be accessed in several ways. They appear when you hover over the video. You can also access them in the document ribbon or from the context menu when you right-click inside the video pane. The ribbon and the context menu have an option to create a snapshot of a video frame. Snapshots are automatically inserted as new image documents. You will find volume and playback-rate controls in the document ribbon. For Skills trainings 4.15–4.17, you can use your own data or work with the sample project provided on the companion website (Skills trainings 4.15–4.17 (versions 1 and 2)). Version 1 only contains the documents; version 2 has quotations, codes and hyperlinks for doublechecking your own work.
92 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.23 Display of video data An easy way to start and stop the multimedia file is by pressing the space bar. The View tab in the document ribbon has a few further options. You can hide or display the video preview and audio waveform in both the timeline and the zoom line. The timeline can be displayed horizontally or vertically. Zooming the timeline You can zoom the timeline to select just the section of the audio or video that you want to see in the margin area: • The easiest way is to use the mouse-wheel zoom by holding down the Ctrl key. Another option is to move the two orange sliders that you see on the right- and left-hand sides of the timeline. Time indicators display the start and end positions and length of the selection (see Figure 4.24).
TECHNICAL ASPECTS OF CODING 93 Figure 4.24 Zooming the timeline Creating an audio or video quotation • Move your mouse pointer over the audio waveform or preview images. You see time indicators that show the start and end positions and total length of the segment. Left-click roughly where you want the quotation to start and drag the mouse pointer to the approximate end position. As soon as you let go of the mouse, a quotation is created, and you see the blue quotation boundary in the margin. Figure 4.25 Creating a video quotation • Start and end points can also be set via the keyboard. Click on the spacebar to start the video. Press the comma key (,) to set the start position and press the dot key (.) to set the end point. • To review a quotation: move over it with the mouse and click on the Play button or double-click on the quotation bar in the margin area. Figure 4.26 Reviewing a video quotation
94 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI If you are not quite happy with the start and end positions, an easy way of adjusting them is as follows: • Move the playhead to either the beginning or the end of the quotation and use the right and left arrow keys for fine-tuning the segment. Each click moves the playhead pointer in 1/1000 increments of the length of the video. This allows you to set the length of audio and video quotations quite precisely. • Once you have located the correct position, drag the orange slider onto the red line of the playhead. Repeat this for the second position as well if necessary. Display of video quotations The default name for image, audio and video quotations is the document name. You can, however, rename each quotation. A step-by-step instruction is provided below. All icons of commented quotations show a little Post-it note, and you see a tilde (~) in front of the quotation ID (see Figure 4.28). In version 2 of the sample project, several quotations have been linked to other quotations. Such links are called ‘hyperlinks’ in ATLAS.ti and are marked with the > and < signs (see Skills training 7.5). Figure 4.27 Adjusting the length of a quotation Figure 4.28 Renamed and commented video quotation in the Quotation Manager (see version 2 of the sample project)
TECHNICAL ASPECTS OF CODING 95 First steps in analyzing video data I recommend beginning your analytic work directly on the video file. I would not spend months transcribing a video file as some people do. Depending on the research question and the analytical approach, you could create a transcript of the conversations in the video and link it to the video file using the multimedia transcript option. However, it is not necessary to transcribe what you can see or hear, such as music, nature sounds, background noise, etc. This can be captured by quotations, codes and comments that have a direct link to the video data. Therefore, my suggestion is to add your video files to the ATLAS.ti project and begin your analytic work immediately. Making use of quotation names and comments As mentioned above, the name of a video quotation is its data file name. In most cases, this is not very useful, at least not for analytic purposes. I suggest doing two things: renaming the quotations and adding comments. You can rename each quotation in the process of creating them or open the Quotation Manager alongside the video document in a new tab group (see Skills training 2.6). I recommend the latter if your screen is big enough. • Option A: in the process of creating video quotations, right-click on the quotation bar in the margin area and select Rename. • Option B: right-click on the quotation in the Quotation Manager and select Rename or click F2. Figure 4.29 Renaming video quotations
96 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI There is no character limit. I suggest using the name as a title that summarizes the video quotation. Longer descriptions and interpretations can be written in the commentary of the quote: • Option A: right-click on a video quotation bar in the margin area or within the highlighted area on the preview images and select Edit comment. • Option B: in the Quotation Manager, write the comment into the comment field (see Figure 4.29). A quick way to access all commented quotations is by setting a filter in the Quotation Manager. • Open the Quotation Manager and select the Search & Filter tab in the ribbon. • Click on the drop-down arrow of the filter button and select Commented. While adding titles for each video segment and writing comments, you begin the analytic interpretive process. An additional benefit is that you can create meaningful text reports. Below, you can see a report of video quotations coded with the code ‘places that impressed’ (Figure 4.31). If you want to create the report yourself, import version 2 of the sample project. • Open the Code Manager and select the code ‘places that impressed’. • Click on the Report button in the ribbon. • Select the options shown in Figure 4.30: Quotations – Content, Comments, Hyperlinks. Adding codes Coding video quotation is no different from coding other media types: • Right-click on the highlighted quotation segment (or the quotation bar) and select Open Coding; or drag and drop an existing code from the navigation area on the left or from the Code Manager onto the quotation. Figure 4.30 Creating a report for video quotations by code
TECHNICAL ASPECTS OF CODING 97 Figure 4.31 Code report for video data SKILLS TRAINING 4.16 CODING AN IMAGE What you have just learned about working with video data can also be applied to images. When creating an image quotation, the quotation name is the same as the file name. As has been described above, you can rename and comment each image quotation in the process of analyzing. You can immediately code each selection that you make, or you can stay at the quotation level before assigning codes. This will depend on your methodological approach. In order to try it out, you can either add an image file to the project or create a snapshot of an image in the video file: • Move the playhead to the position that you want to capture, right-click on the video and select Capture video frame.
98 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.32 Creating a video snapshot The snapshot will be added as new image document to your project. • Load the new image document, select a rectangular area and go ahead and code it as described for text segments (Figure 4.33). Figure 4.33 Coding an image document SKILLS TRAINING 4.17 WORKING WITH GEO DATA Adding a geo document • To add a geo document to the project, go to the Home tab in the ribbon. Click on the down arrow of the Add Documents button and select New Geo Document.
TECHNICAL ASPECTS OF CODING 99 • The new document is loaded, and you see the world map. ATLAS.ti uses OpenStreetMap for this function. The default name for the document is ‘New Geo Document’. You may want to rename it, so it better describes your analytic intention. If you work with the sample project, name it ‘Iceland’. Figure 4.34 Finding a location on a geo document Let’s look for the place that I have captured above: click Query Address in the ribbon and enter ‘Gulfoss’. The map immediately displays the location that fits the search term and adds a place mark (Figure 4.34). Creating a geo quotation Creating a geo quotation is not much different from creating other types of quotations. The only difference is that the quotation is just one location on the map – the place mark – and not a region. • Right-click on the place mark and select Create Free Quotation. If you want to create a quotation for a different location on the map, left-click until you see a place mark and proceed as above. Geo quotations can be accessed in the margin area, the Quotation Manager, the Quotation Browser and the Project Explorer in the navigation area on the left-hand side – and from within networks. Geo quotations are displayed as follows:
100 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 4.35 Display of geo quotation in the Quotation Manager The quotation icon shows a place mark and the quotation name consists of the geo coordinates for degree of longitude and degree of latitude according to the World Geodetic System 1984 (WGS84). In addition, the Quotation Manager shows a preview of the geo location. As recommended for the analysis of video data, it is useful to rename and comment on geo quotations. • Change the name of the quotation to something more meaningful, like the place you have marked. Write something in the comment field to describe the location (e.g. excerpts from your field notes or information you have about the location). • Create a few more geo quotations, rename and comment them. See version 2 of the sample project. Coding geo quotations You can also create a quotation and link a code to it in one step, or code an existing quotation: • Set a place mark or select an existing one, right-click and select the option Open Coding (or click on the equivalent coding buttons in the Document ribbon). If you move your mouse over a quotation in the map, a text window opens displaying information about the quotation like the name, the author, when it was created, any codes and memos that are attached to it, the longitude and latitude and the address.
TECHNICAL ASPECTS OF CODING 101 Figure 4.36 Mouse-over information of geo quotations REVIEW QUESTIONS 1 Which coding options does ATLAS.ti offer? 2 What are in-vivo codes and what can they be used for? 3 Explain the quotation ID in the Quotation Manager. 4 What does ‘grounded’ and ‘density’ mean in the Code Manager? 5 Why is it important to write code definitions? 6 For which purpose can you use quotation comments? 7 How do you need to prepare a code list if you want to import it? 8 How can you export a code list for reuse in another project? 9 How can you code focus group data and what do you need to know about preparing focus group transcripts? 10 What do you need to know when coding PDFs? 11 How can you go about analyzing image, audio, video and geo documents? Which options are you likely to use more or in a different way compared to working with text data? FURTHER READING Freeman, Melissa (2017). Modes of Thinking for Qualitative Data Analysis, chapter 2. London: Routledge.
102 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Richards, Lyn (2009). Handling Qualitative Data: A Practical Guide, 2nd edn, chapter 5. London: Sage. Richards, Lyn and Morse, Janice M. (2013). Readme First for a User’s Guide to Qualitative Methods, 3rd edn, chapter 6. Thousand Oaks, CA: Sage. Saldaña, Johnny (2009). The Coding Manual for Qualitative Researchers. London: Sage.
Creating a coding scheme 5 ‘Coding means that we attach labels to segments of data that depict what each segment is about. Through coding, we raise analytic questions about our data […]. Coding distills data, sorts them, and gives us an analytic handle for making comparisons with other segments of data.’ (Charmaz, 2014: 4) ‘Coding is the strategy that moves data from diffuse and messy text to organized ideas about what is going on.’ (Richards and Morse, 2013: 167) Now is the time to gain your first hands-on experience of data content coding. You begin by exploring your data, discovering and collecting interesting things and putting your technical coding skills into practice. The aim of this part of the journey is to get acquainted with the terrain, observe, take notes and develop your first coding ideas. Perhaps you are already beginning to structure your thoughts about the things you are noticing or at least wondering how you can do this. I will explain how you can systematically build your coding system, why some paths are dead ends or detours and why some tools should be used in the way suggested to get a meaningful and understandable representation of the data terrain.
104 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI A code in ATLAS.ti can be a simple description, a concept, a category, a subcategory or a wildcard that modifies a link in a network. The software itself does not dictate how to use a code. It only provides this entity as an item in the toolbox. If you want to build a cupboard, then your toolbox may contain everything you need to build this cupboard out of a pile of wood. However, if you have no knowledge or experience, the pile of wood may never become a cupboard. If someone with experience stands at your side and supports you, then you are more likely to succeed. Twenty years ago, I stood in front of this toolbox for the first time and started coding data with ATLAS.ti. It helped me to analyze my data and I have not given up. But my projects from those early days look different from the projects that I analyze today. I could not use the full potential of the software. Through trial and error, I have learned over time how to get the most from the tools. In this chapter, I would like to assist you with your first attempts, so that you may build a beautiful cupboard (in the figurative sense) even if this is your first time working with ATLAS.ti. LEARNING OBJECTIVES In this chapter, I give best-practice advice and rules of thumb to help you build a wellstructured code system in a methodological and analytical sense. This means a code system that allows you to perform further analysis that goes beyond simple retrieval. You will learn how to build categories either by developing subcategories from broad codes or the other way around, from descriptive labels. In addition, you will learn about the pitfalls in building categories. On the methodological level, I introduce you to the difference between thematic and interpretive coding, and how to apply comments and memos in ATLAS.ti. We will start this chapter by playing a virtual jigsaw puzzle, as playing a puzzle has many similarities with coding. Even if you are coding data for the first time, you have already ‘coded’ many things in your life. You will learn how to apply these existing skills to a new area. SKILLS TRAININGS Skills training 5.1: noticing and collecting – coding data for content Skills training 5.2: retrieving all quotations of a code Skills training 5.3: developing subcategories Skills training 5.4: building categories from descriptive labels Skills training 5.5: defining categories on the ‘right’ level Skills training 5.6: comparing thematic to interpretive coding Skills training 5.7: writing research-question memos
CREATING A CODING SCHEME 105 LET’S DO A PUZZLE SO YOU REMEMBER HOW GOOD YOU ARE AT CATEGORIZING Imagine you are sitting at a table. On the table there are 1,000 pieces of a jigsaw puzzle with the picture side up (Figure 5.1). Now it’s your turn. Your task is to put the puzzle together. How do you go about it? Figure 5.1 Doing a puzzle Most people would answer that either they begin with the corners and the edges or they sort by colors or shapes. Let’s begin with the corners and edges. Why do you think that most people begin like that? These pieces are easy to recognize since they have at least one straight edge. When it comes to analyzing a project in ATLAS.ti, I likewise recommend you begin with what is easiest. In Chapter 3, you learned how to make use of already known characteristics, incorporating them into the file names for easy sorting and ordering of the documents within the project. I also showed you how to create document groups like male, female, location, age, etc. By starting to work on a project in this way, you have literally framed it (Figure 5.2). Figure 5.2 Start to lay the frame
106 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI As soon as the frame is laid, the next task is to examine the arrangement of the other parts in more detail. You could try to find those parts that belong together. But that’s probably tedious. A better strategy is to sort the parts by color and similarity in terms of what is visible on them (Figure 5.3). The sample puzzle here depicts a castle with a forest around it and a lake in the upper-right-hand corner (I know this as I have seen the lid of the box!). The next step is to take a closer look at one of the piles. Let’s look at the pieces of the puzzle that look like they belong to the castle. Approaching this strategically, one looks for puzzle pieces that belong to a part of the castle, like the roof, the towers, the windows or the battlements. In other words, one segments the castle into sub units (Figure 5.4). This process is repeated for all pre-sorted stacks until everything can be put together to complete the puzzle. Figure 5.3 Sorting the pieces of the puzzle Figure 5.4 Segmenting the castle into sub units
CREATING A CODING SCHEME 107 Pre-sorting the parts of the puzzle into piles of similar elements is like coding the data by major themes. Segmenting the piles into smaller sub units is like building subcategories. Even if you are not a seasoned puzzle solver, as mentioned in Chapter 4, you are nonetheless already very skilled at constructing subcategories. You do it every day and you learned the technique a long time ago when discovering the world as a child. You may have first realized that a certain animal is a dog and then used the word ‘dog’ for the various kinds of dogs. Later you learned that they are beagles, boxers, golden retrievers, poodles, mongrels, etc. Developing subcategories for your data is not much different. Let’s assume you have collected lots of quotations under a common label; the next step is to look through them, as I did with the castle pieces. After reading or looking at a few quotations, you will quickly notice where the commonalities are. Having coded the data, the software makes it easy to retrieve all segments that belong to one topic and to take a second, closer look. The aim is to develop subcategories so that you can bring some order to the pile and differentiate the various aspects of the topic area you are looking at. It is, however, not always that easy with computer-aided analysis. You do not necessarily know what the main topics are, and sometimes you start on the lower level, naming only the pieces of the puzzle. If that is the case, you will be generating very many codes in a very short time. If you notice this, you should stop coding and look at which of the codes can be aggregated so that more data segments can be collected in them. Technically, this means you will be merging codes (see Skills training 4.10). The goal is to sort and organize the data rather than just naming each element (of the puzzle). When you do a normal puzzle, the lid shows you what the finished puzzle should look like in the end. By comparison, in a qualitative research study, we usually do not have a template that shows what the result should be. The researcher probably has some ideas based on existing literature. But the answers to the research questions will only emerge through the analysis process. There is a certain kind of puzzle that is similar in strategy to the qualitative analysis approach: the so-called WASGIJ puzzle (‘jigsaw’ spelt backwards). The finished puzzle does not match the picture on the box. Instead, the player must assume the role of one of the persons on the cover and put himself in the place of that person. The final image corresponds to what this person sees from his or her perspective. The solution thus results from the process of putting together the parts. There are also jigsaw puzzles without a template for advanced players with experience. Those puzzles are comparable to a project where it is difficult to find existing literature or earlier research on the subject matter. Thus, you may have a challenging time developing detailed research questions based on previous knowledge, and the only option you have is to go into the field and start collecting data. Grounded theory studies often are like this. Like a puzzle without a template, such a project is proportionately more difficult than one which is guided by research questions. First ideas for coding can be derived from research questions, from theories, from the literature or from the interview guideline. Ideas for coding in the grounded theory sense can also emerge from the data, but this is not so easy for a beginning researcher. As Kelle and Kluge note: ‘novices in the field of social research have a particularly tough time following recommendations like “let theoretical concepts emerge from your data material”. For them, such attempts might result in drowning in data
108 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI material for months’ (2010: 19). Therefore, I recommend that you do not start with the most difficult kind of puzzle (= methodical approach) if qualitative data analysis is new to you. A thematic analysis is easier than a grounded theory study, just to give an example. Having pointed out the similarities between doing a puzzle and coding qualitative data, I should say that there are also some differences. Qualitative data are not yet broken down into parts. This happens only in the process of coding; in ATLAS.ti, this is the creation of quotation. Furthermore, I have said that when doing a puzzle, the individual parts are initially sorted into larger clusters. When coding data, the main topics may not be so obvious, and there is a process to develop them. Sometimes one encodes at the level of subcategories, or even lower, at the descriptive level. And only over time does it become clear which codes belong to a higher-order category. Let’s start with the first exercise where you are asked to code data yourself. This exercise will show you if you tend to build piles (you are a lumper) or if you stay very close to the data and name each interesting data segment first (you are a splitter). It may also be that you are both a lumper and a splitter (Weller and Romney, 1988; Guest, et al., 2012). In the second and third Skills trainings in this chapter, I show you what to do if you have collected lots of data segments under a main theme and, vice versa, what you need to do if you have coded the data very finely. SKILLS TRAINING 5.1 NOTICING AND COLLECTING – CODING DATA FOR CONTENT For the first few exercises in this chapter, I would like to ask you to open your own Children & Happiness project that you created in Chapter 3. For Skills trainings 5.2 and 5.3, I will provide specially prepared projects that you can download from the companion website. I hope that you are now ready and eager to begin with the coding work. Before you begin, I would like to give you a navigation instrument: the N–C–T model. Figure 5.5 Use the N–C–T model as your navigator
CREATING A CODING SCHEME 109 Noticing. This refers to the process of finding interesting things in the data when reading through transcripts, field notes, documents, reports, newspaper articles, etc.; or when viewing video material or images; or when listening to audio files. In order to capture these, the researcher may write down notes, mark the segments or attach preliminary codes. Codes may be derived inductively or deductively. At this point, the level of a code does not play a role. Codes may be descriptive or already conceptual. The important point is to mark those things that are interesting in the data and to name them. Collecting. Reading further, you will very likely notice a few things similar to some you have noticed before. They may even fit under the same code name. If a similar issue does not quite fit under the same heading as the first issue you noticed, you can simply rename the code to subsume the two. Even if the term is not yet the perfect code label, it does not matter. As you continue to collect similar data segments you will think of a better and more fitting label. Thinking about things. We need to think when noticing things, when coming up with good names for codes or when developing categories and subcategories. We need to do some more thinking when it comes to finding patterns and relations in the data. This mostly takes place after coding when asking, ‘How do the various parts of the puzzle fit together? How can we integrate the various aspects of the findings in order to develop a comprehensive picture of the phenomenon studied?’ A lot of the thinking then happens while writing memos. Figure 5.5 shows that noticing, collecting and thinking go hand in hand and back and forth. It’s not a straight path through your data. You will rework your codes and the code system several times; you may come to the realization that you need more data even in the second analysis phase. If you query your data, you may still be adapting codings. With every cycle you go through, the changes will be smaller and you will understand your data better, until you are satisfied with the insights you have gained. Overall, it’s a rewarding process. Let’s now start with the first step of the cycle. • Read the article by Powdthavee (D1 in your project if you did not change the order), which provides the backdrop. • Start to read Belkin’s parenting blog discussion (D3). When you start noticing things, begin to collect them by assigning them a code. Do not think too much about a perfect code label, just write whatever comes to mind and continue. You can always rename the codes later. The main research question that you can keep in mind when reading the blog data is: ‘How do blog discussants perceive the reported relationship between children Figure 5.6 Starting the coding process: noticing and collecting things
110 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI and happiness? Are there differences between parents and non-parents or between males and females? What other aspects come up in discussing the scientific findings?’ Just keep an open mind, notice as many things as you can and collect them via coding. If you feel that it is important to read all the data first and to write down notes on a piece of paper before you create codes in ATLAS.ti, then this is a suitable way to proceed. If, however, after reading a few of the blog comments, you already have some ideas for codes, then go straight ahead and start coding in ATLAS.ti. Do whatever feels most natural to you. • Continue with this exercise for at least 45–60 minutes. You may create up to 20 to 30 codes within this time or only six or seven, depending on your style of coding. • If you have questions on how to structure your list of codes and how to organize them, don’t worry about it now. I will discuss it afterwards. I want you to keep your mind open and notice as many things as possible without sorting them into lower- and higher-order levels too soon. Remember, this is only your first step of coding. Discussion of the coding exercise How did it go? Was it fun? How did you cope with the technical aspects of coding in ATLAS.ti: creating codes, modifying quotations, renaming codes, etc.? Do write down any thoughts and questions that arose during this first part of the journey in your research diary memo before you read on. I cannot ask you to show me your list of codes, but I have some lists created by four other coders who have already done this exercise in a face-to-face workshop (Table 5.1). Table 5.1 Ideas collected by four coders Coder 1 Coder 2 Coder 3 Coder 4 Reasons for a baby Consequences of early parenthood No regrets Financial difficulties Experience of unhappiness Reasons for having children More than I was before Negative impact on career Definition of happiness Reasons for not having children Richer life Marital satisfaction goes down Study critique Hard work but joy Types of happiness More joy Effects of parenting Responsibility Drop in marital quality Becoming a bit wiser Impact on relationships Biological imperative Children are more about meaning of life than happiness Logical consequence of having a family Third-person perspective Associated experience Less burdened before Motivation Biology Love my partner more Critique of the concept of happiness Questioning the sampling More stress Life is richer
CREATING A CODING SCHEME 111 Coder 1 Coder 2 Coder 3 Coder 4 Scared Desire a change Changed feelings about own parents Improving the world Fulfillment Planned/unplanned children Less time Happiness is unrelated to having children New aspects of love You can see from Table 5.1 that coder 1 came up with six codes, coders 2 and 3 with nine codes and coder 4 with 19 codes. If you take a closer look at these codes, you may already get an idea of why the number of codes ranges between six and 19. Coder 1 has coded similarly to the way I explained in the puzzle analogy, collecting data segments under a common term, like all reasons for having a baby. Bazeley and Richards (2000: 54) refer to this form of coding as ‘broad-brush coding’. Coder 4 has named the single pieces of the puzzle (labeled ‘concept coding’ by Bazeley and Richards). It does not matter whether you start with only a few broader codes or with a larger list of codes that are close to the data. What matters is that you know how to continue and what to do with your initial ideas. Neither coder 1 nor coder 4 should continue to code all their data in this way; coder 1 needs to develop subcategories and coder 4 needs to code on a more aggregated level. Coders 2 and 3 are likely to do both. Since the codes are not hard printed in the data, you can approach coding in a flexible manner, as it bests suits your methodology and personal style. Below, and more specifically in Skills trainings 5.2 and 5.3, you will learn how to proceed with coding and with developing your coding schema, depending on your starting point. How to add more structure to your exploration To generate enough ideas for developing a coding system, an hour or less is not enough. You need to plan more time for it. There are no set rules, but I can offer a few guidelines. If you conduct an interview study, select three to five interviews depending on their length. If you analyze newspaper data, you will need about 20 to 30 articles. Begin by coding the data as instructed above: keep it very open, just collecting some ideas. At first, you will generate lots of new codes; in time, you will reuse more and more of the codes that you already have, and you won’t need to create new ones. You will have reached your first saturation point. In technical terms, you will drag and drop existing codes from the Code Manager or navigation panel onto the data segments. Looking at the data landscape, you have now roughly described the various elements.
112 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI Figure 5.7 Describing the data landscape As soon as you reach this point – that is, when you no longer add new codes and only use drag-and-drop coding – it is time to review your coding system. If you do it at a much later stage it will need more work, because then you will have to go through all the documents again to apply newly developed subcategories and recheck all other codings. This is an appropriate time to export your project, so you can use it later when writing your report. This preserves the original coding and allows you to compare it later with more advanced versions of your project. In this way you can, for example, describe in your method section how you got from A to B and C in your project. Let’s assume you have taken your first round of coding up to this point. Those coders who naturally develop a mix of descriptive and abstract codes will have around 100 codes, depending on the project. Smaller student projects may hold around 50– 70 codes. The cleaning up and restructuring of a first code list is done within the software. When you do it on paper, you need to apply the changes inside ATLAS.ti in a second step. If you have noticed a lot of things – let’s say you already have 200 or 300 codes – your codes are probably very descriptive. As mentioned, coders of this type are referred to in the literature as splitters (Guest et al., 2012; Bernard and Ryan, 2010). Splitters often find it very difficult to let go of their codes by merging and sorting them, for fear of losing something. I can assure you that this is not going to happen. Instead of tagging each data segment, after
CREATING A CODING SCHEME 113 restructuring, you will have a single code that might hold ten quotations in their original form. This is far better than ten codes that only summarize the original data. The need to push codes from a descriptive to a conceptual, more abstract, level is just as relevant to manual ways of coding. Corbin and Strauss wrote: One of the mistakes beginning analysts make is to fail to differentiate between levels of concepts. They don’t start early in the analytic process differentiating lower-level explanatory concepts from the larger ideas or higher-level concepts that seem to unite them. […] If an analyst does not begin to differentiate at this early stage of analysis, he or she is likely to end up with pages and pages of concepts and no idea how they fit together. (2008: 165) This also applies to computer-aided analysis, although it is no problem for the computer to manage 5,000 or more codes. Instead of being conducive to your analysis, a high number of codes prevents further analysis. Creating too many codes is one of the dangers you will encounter on your journey through the qualitative data landscape. Based on the codes shown in Table 5.1, I would like to show how an initial list of codes can be sorted and structured. As the codes were generated by four coders in four different projects, this also serves as an example of how you can develop a code list when working in teams. I merged all four projects (see Chapter 9 for more details on how to merge projects) to create one project that holds the codes of all four coders. The code list then had 43 codes. Next, I went through the list and merged all codes that have a different label but appear to have the same meaning, like ‘reasons for a baby’ and ‘reasons for having children’. Having cleaned up the list in this way, I moved on, adding more conceptual structure to the list. Codes like ‘scared’ or ‘biology’ do not tell us much apart from the fact that the name or word is likely to be mentioned in the coded data segment. These codes must be either renamed or subsumed under other codes. When looking at the context, ‘scared’ is mentioned as a reason for not having children, and ‘biology/biological imperative’, on the other hand, is mentioned as a reason for having children. ‘Becoming a bit wiser’, ‘fulfillment’, ‘more joy’, ‘being more than I was before’ are all mentioned as positive effects of having children. Developing a category for reasons for having children/reasons for not having children and positive as well as negative effects of having children seems to be a promising idea. This way, I can sort all these aspects into different subcategories. How can this be accomplished in ATLAS.ti? The code list in ATLAS.ti is linear and by default sorted in alphabetical order. Therefore, you need to tweak the code labels to structure the list. I usually add prefixes to the name followed by an underscore or a colon. It is important that you separate the prefix that names the main category from the subcategory name. This way, all subcategories are automatically sorted under the main category name. For the main category name, I use capital letters. This is a habit I developed when using ATLAS.ti version 5, when code names could not be colored. In version 6, coloring became possible, but I still like the capital letters for the main category name to distinguish between types of codes. Sometimes you will need to play around with the prefixes so that the codes are in the order that you want them to be, with the main category name on top. In the example below, I did not want the full category name as the prefix, because otherwise the code name would have become very long.
114 QUALITATIVE DATA ANALYSIS WITH ATLAS.TI EFFECTS POS Effects pos: becoming a bit wiser Effects pos: fulfillment Effects pos: more joy Effects pos: richer life The categories as proposed here are provisional, as they are based on very little coding. With more coding, they are likely to change and develop further. I like Saldaña’s idea of first-cycle and second-cycle coding (Saldaña, 2013: 8). The idea of the cycle fits the nature of the N–C–T model, where you have seen that qualitative analysis is cyclical rather than linear. First-cycle coding, according to Saldaña (2009: 45), refers to those processes that happen during the initial coding. In N–C–T analysis, these are the ideas you notice and collect when you begin the coding process as shown in Table 5.1. Second-cycle coding is the next step. Its main goal is ‘to develop a sense of categorical, thematic, conceptual, and/or theoretical organization from your array of first cycle codes’ (Saldaña, 2009: 149). The process of second-cycle coding entails the following activities: classifying, prioritizing, integrating, synthesizing, abstracting and conceptualizing. In other words, this means sorting and ordering codes. The goal in computer-aided analysis is to create a code system. In my experience, you will find that a third and fourth cycle of coding, perhaps more, will be necessary. The aim of this process is to develop a structured code list based on a subsample of your data. Once you have developed a first structure, you can apply the codes to the remaining data. You will likely continue to make changes to the code list and refine the structure the more you code. But this is OK. Figure 5.8 First steps in sorting and structuring a code list
CREATING A CODING SCHEME 115 Describing coding as a process that consists of at least two stages is common to several authors (see, e.g., Bazeley, 2007; Bazeley and Richards; 2000; Charmaz, 2006; Fielding and Lee, 1998; Kuckartz, 1995; Richards, 2009; Richards and Morse, 2013; Saillard, 2011; Silver and Lewins, 2014). Richards (2009), for example, refers to it as a catalog of codes. As advantages of a well-sorted catalog she mentions speed, reliability and efficiency. Further, she provides some tips for building a good code system. The problem, as I see repeatedly in my everyday work, is the translation of this process into mouse clicks and the technicalities of it in a software environment. Even if users know the technical aspects of coding, on the one hand, and read the useful tips, on the other, they often find it difficult to apply these skills. It is not so difficult, but neither is it self-explanatory. Therefore, you can practice it in Skills trainings 5.3 and 5.4. Figure 5.8 shows an excerpt of how the codes from the four coders were sorted and reordered. There are a few codes that could not be classified into a category, since it was too early to decide. The code ‘study critique’, for instance, may turn into a category of its own at some point. If there is a third-person perspective, there is likely to be a first-person perspective as well if you read on. If you look at the subcategories of ‘effects pos’ and ‘effects neg’, you may notice that they are still very close to the data. Reviewing more data material is likely to allow for formulating more abstract labels. What you can also see in Figure 5.8 is that I make use of code groups. As you will see in Skills training 5.2, code groups are useful as filters. When I began to sort the codes, I realized that some codes referred to positive and others to negative aspects of parenting. To collect all those codes, I created the two code groups ‘effects positive’ and ‘effects negative’ and dragged all codes that fitted into the two groups. With a click on a group, the list of codes is filtered. This makes it much easier to add prefixes. All codes that cannot yet be assigned to a category will not receive a color, and if they are not already at the top of the list, I add an asterisk (*) in front of the code name as a little trick to force them to the top of the list due to the alphabetical ordering. This way, it is easier to spot where you still have codes that do not belong anywhere. Another way of sorting a code list is with the help of numbers and special characters. The sort order is as follows: special characters, numbers and letters. Sorting by numbers allows you to add another dimension in addition to alphabetical sorting. The code list in Figure 5.9 describes a merger and acquisition process and questions about organizational fit come before evaluating the final results. As the label ‘final results’ begins with an F, it would come before the category ‘organizational fit’. In addition to the number, letters were inserted as reference to the major theme, like J for ‘justice’ and C for ‘controllability’. The letters a, b and c were used when new subcategories had to be inserted into existing Figure 5.9 Sorting a code list by numbers and letters