In turn, codes were consolidated into high-level themes using data-reduction techniques such as category sorting, in which interview segments are grouped according to content similarity [ 45].
The analysis approach taken to the dataset was both inductive, as some sub-themes were closely linked to the data, and deductive as the high-level themes were driven by theory and the researchers' analytical interest [ 56].
A number of high-level themes emerged, relating who the user is perceived to be, the methods used, the perceived value and barriers to user involvement, and the nature of user contributions.
Initially, a deductive stance was adopted, using the three pre-determined high-level themes of the technology acceptance model (TAM): Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Actual Use (AU).
[ 57] The multi-method work of the POET investigators, also conducted in the inpatient setting, identified four high-level themes that summarize perceptions of CPOE systems: 1) organizational issues, 2) clinical/professional issues surrounding customization, 3) technical/implementation issues, and 4) issues of the organization of information and knowledge.
Grouping was performed by idenfication of direct synonyms, shared terminology and/or evidence of a higher level theme.
"We identify the objects in [a video and] we identify higher level themes," he said.
Higher level themes will be suggested in discussions with the research group.
29 Eventually, broader categories were reduced to a set of higher level themes.
Factors were then compared with each other to identify higher level themes.
