It is relevant to place this discussion of thematic analysis at the head of this section, as the process has been inductive and deductive, that is the themes have both been derived from the data and been imposed upon the data. Ayres describes an inherently non-linear process of theme development and allocation, as new links and relationships emerge during the data gathering and analysis process (in Given, 2008, pp. 867-868). Rather than employing traditional methods of data coding and textual analysis of transcripts, themes have been identified through tagging and commentary of the large quantity of notes, posts and other texts in a more grounded approach. These tags became particularly strongly defined during the analysis process and the reflective writing around particular events.
The relationships between these tags began to coalesce into categories or “motifs” (Kidd & Finlayson, 2009, p. 992) centred on particular traits within the descriptions and criteria of Autistic Spectrum Disorders. For example, an observation of a particular interaction in the studio might be tagged body-language, eye-contact and communication. Given the questions regarding the impact of particular traits, it was logical to categorise data chunks into themes and these themes appear in this document in the Insights section. It is worthwhile noting that the individual words used as tags are strongly reflective of increased familiarity with the research into ASD as my own analysis progressed. In many instances the tagging of individual posts was re-visited as insights formed. In this sense the coding process is reflective of a relatively grounded set of themes developing from the literature and a growing self awareness.
Individual texts were therefore assigned to categories within the Scrivener writing package and later to WordPress. This allowed for presentation of (relatively) raw data in the form of the reflective and evocative writing to sit within the themed categories as explicit exemplars (see Given, 2005, p. 464).
This inductive process was used to identify the traits of ASD that are perceived to have significant impact, and also those whose impact is limited or negligible. For example, by this method, issues of emotional recognition and regulation have been found to be of much greater import than repetitive body or mental activity (tics) although both are present. Many pieces of data had relevance to more than one main theme, and deductive analysis was employed to make the principles involved in individual events, circumstances and reflections more explicit. In many cases this was achieved through the “recommended reading” links placed below the Table of Contents.
The initial theming process was greatly aided by the use of meta-data features in Scrivener (labels, status, notes, keywords and tags), but it was not until the texts were subjected to the automated scrutiny of the YARPP plugin within WordPress that many of the cross-themes emerged. In some cases posts were given particular key words during the writing and editing process, and assigned to a given category, but re-assigned or re-assessed after links were indicated by the YARPP plugin. In addition the on-line texts gained many more internal links and required frequent re-visit and re-assessment. In this way the relational links of the hyperlinked document have been absolutely intrinsic to the thematic analysis and the insights derived.
It is possible that a text based analysis of interviews and narratives using NVivo may well have offered similar insights via different means; but I have found the iterative and continuous processes employed to be both effective and exciting, and maintained a close relationship with the data that may have been difficult with a more technical approach. As the document is designed to have a life beyond assessment, especially through the comments of readers, it is expected that further relationships will come to light through the processes in place.