ATLAS/ti provides the means for structuring, known as the „families-editor" Such an editor is available for the three object-types PD, code and memo. However this term might be misleading when thinking of the traditional term „family". The idea was originally based on another conception: Barney Glaser, the co-inventor of the Grounded Theory methodology (Glaser/Strauss 1967) has in an extension to this approach in 1978, suggested the orientation of conceptual coding at „code-families". It is to be interpreted as grouping together commonly used concepts ("theoretical codes") on a certain analytical level, similar to word-families as in vocabulary-exercises in primary school. An example of "code-families" is the "purpose-method-family", consisting of codes such as: target, purpose, anticipated consequence, result etc. (Glaser 1978: 77).
The formal distinction between families in the traditional conception and Glaser's conception, is that there is a relationship or better affinity defined between the particular persons (or roles). The code-families of Glaser are merely a loose collection of terms that are conceptually distributable on a level and do furthermore not state anything about the internal relation that terms have to each other. ATLAS/ti uses this principle in the three object-areas PD, code and memo. A separate „family-editor" may be used for each object-type (for example through the context-menu of the list).
These editors basically consist of four window panes: the family list-window on the top where all previously created families are referenced. Through the context-menu of the list-window you can also carry out a number of family-related functions (create new, delete, rename,...). Below that, you will find the yellow comment-window already familiar from the special-lists in which you add a comment to each family. In the middle you will also find two split windows, connected by two operator-buttons. As soon as you have founded a new family or activated an existing family, the objects from the object-area that are not a member of the current family, appear in red color within the right pane while the left pane displays the list of objects assigned to that family. Now you may assign objects to the family by selecting objects from the list and choosing one of the two operators (from the right to the left or removing (from the left to the right). Moreover, the entire number of objects may become a member of any number of families. Note that there is no exclusiveness of assignment as with a strict categorical classification.
An example: You will find miscellaneous data-types in the area of primary-texts that at the same time belong to miscellaneous cases and where each of which has its own linking- or separating-characteristics. For example there are interviews with actors of a different sex whereas a few belong to case A but some do belong to case B. At the same time you may possibly have documents, expert-talks and transcripts of observations as well as a researcher's diary containing notes. The family-editor allows you to arrange this variety of material by founding families, such as "Interviews", "Observations", "Files", "male questioned persons", "female questioned persons", "Case A", "Case B" etc. and assigning each PD to all logically related families. Thus an interview with a manager of company A would exist in the families "female questioned persons" as well as in "Interview" and in "Case A" respectively. What is the advantage of this structuring? You gain valuable sort-criteria at the level of primary-texts that may be re-used at any work-situation. Initially it will allow you to filter the PD's to be displayed in the PD-list. When using the "Autocode" feature, you can limit the search to specific data-files that are a member of a particular PD-family. You may also make a "textbase selection" with the Query Tool to pre-select data to be involved in the retrieval-process when conducting a quotation-retrieval (see the section about the Query Tool). Here you can also use PD-families as criteria.
As earlier mentioned, families may be used for memos (to structure the types of memo-texts) as well as to create new codes. A frequently asked question regarding code-families, is the conceptual difference between the structural means of families and the structuring of the code-system through defined relationships. However the answer is simple: These code-to-code relationships (which will be described in detail later) form a semantic network that shall be the core of the theory building (for example to verify ad-hoc-Hypotheses) and may be thoroughly investigated at a later stage. The "families" are by contrast a very efficient sort-tool that may however not be used directly for building a theory. It is therefore a conceptual difference whether to create a lead-subordinate-term relationship as a family (whereas the family represents the lead-term and the assigned code is representing the subordinate term) or to visualize this matter in a network of defined code-to-code relationships.
All lists of objects offered by ATLAS/ti, namely the four lists for PD's, quotations, codes and memos displayed as combo-boxes, provide two functions to increase the transparency of the list-display as well as the sort-criteria and filtering by default. Both are accessible through the pertinent menu or the pertinent context-menu.
The "set sort options" option provides, as expected, a number of choices for the sort-order of the list-items. Thus you can specify whether to sort the order of the code-list alphabetically (this is the default setting for codes), by creation- or usage-date, by the number of quotation-references or by the references to other codes. The currently selected filter is displayed within the right area of the status-bar in the special list-windows.
This will help you to quickly get an overview of the degree of „Groundedness" --- that means how strongly a code is tied to the data or about the theoretical density (the intensity of linking to a semantic network). The foremost purpose of sort-criteria is however to arrange the best order the objects for a particular work-step (when coding selectively in the sense of the GT it might be useful to first display most frequently used codes since they are primarily used in this phase).
Not only can you sort list-contents differently but also you may additionally decide whether to display all or only a certain portion defined by a specific set of criteria. This is known as „filtering". Depending on the type of list you may choose from multiple factors as a criteria which cannot be mentioned in full. A special feature of the code-list is the ability to have codes displayed or not displayed that were created by a particular co-author – an important function for team working in a research-task. The particular filter-criteria are displayed in the middle field of the status-bar of the special list.