Dataset for "Identifying misleading corporate narratives: The application of linguistic and qualitative methods to commercial determinants of health research"

Data collection method:

We developed a mixed-methods protocol blending methods and tools from corpus linguistics (CL) with inductive coding methods. For the “investor” targeted communications, we sampled annual and investor reports and presentations to shareholders, which included any associated slides and scripts. For the “wider audience” targeted communications, we sampled corporate webpages, and corporate YouTube videos. Annual reports, investor reports and presentations to shareholders were accessed and downloaded via PMI’s investor web pages. The sample of corporate webpages was downloaded manually according to a sampling filter developed specifically for this research. Corporate YouTube videos were accessed directly through PMI’s dedicated account. Automatically generated audio transcripts available from YouTube videos on a single playlist, titled “Inside Us” were saved in NVivo and checked for accuracy by IF, to allow for qualitative coding. Annual and investor reports were downloaded from PMI's website. Webpage sampling As there was no site map readily available through PMI's website, we used a freely available online site mapping tool (xml-sitemaps.com) to generate an XML site map for www.pmi.com and sampled webpages according to the underlying logic of the page hierarchy. This sample consisted of pages at the “top” level (for example, www.pmi.com) and any pages nested under “home”, “investor relations”, “our business” or “who are we”. Target pages were captured using NVivo and Greenshot, and archived using WayBack Machine.

Technical details and requirements:

Coded in NVivo release 1.5.1 (940). QSR International. Subject to license agreement. File converted to permit reading in latest version of NVivo - release 1.6.1 (1137). SketchEngine (https://www.sketchengine.eu/) was used for Corpus Analysis. Subject to licence agreement. (UoB licence not active as of 01/09/2022)

Additional information:

Data files in NVivo are sorted according to data type/source.