Sunday, 22 December 2024
A role for qualitative methods in researching Twitter data on a popular science article's communication
Written for scholars and students who are interested in using qualitative research methods for research with small data, such as tweets on X.
Myself, Dr Corrie Uys, Dr Pat Harpur and Prof Izak van Zyl's open-access paper, 'A role for qualitative methods in researching Twitter data on a popular science article's communication' identifies several potential qualitative research contributions in analysing small data from microblogging communications:
Qualitative research can provide a rich contextual framing for how micro-practices (such as tweet shares for journal articles...) relate to important social dynamics (... like debates on paradigms within higher-level social strata in the Global Health Science field) plus professionals' related identity work. Also, in-depth explorations of microblogging data following qualitative methods can contribute to the research process by supporting meta-level critiques of missing data, (mis-) categorisations, and flawed automated (and manual) results.
Published in Frontiers in Research Metrics and Analytics journal's special topic, Network Analysis of Social Media Texts, our paper responds to calls from Big Data communication researchers for qualitative analysis of online science conversations to better explore their meaning. We identified a scholarly gap in the Science Communication field regarding the role that qualitative methods might play in researching small data regarding micro-bloggers' article communications. Although social media attention assists with academic article dissemination, qualitative research into related microblogging practices is scant. To support calls for the qualitative analysis of such communications, we provided a practical example:
Mixed methods were applied for better understanding an unorthodox, but popular, article (Diet, Diabetes Status, and Personal Experiences of Individuals with Type 2 diabetes Who Self-Selected and Followed a Low Carbohydrate High Fat diet) and its Twitter users' shares over two years. Big Data studies describe patterns in micro-bloggers' activities from large sets of data. In contrast, this small data set was analysed in NVivo™ by me (a pragmatist), and in MAXQDA™ by Corrie (a statistician). As part of the data preparation and cleaning, a comprehensive view of hyperlink sharing and conversations was developed, which quantitative extraction alone could not support. For example, through neglecting the general publication paths that fall outside listed academic publications, and related formal correspondence (such as academic letters, and sharing via open resources).
My multimodal content analysis found that links related to the article were largely shared by health professionals. Its popularity related to its position as a communication event within a longstanding debate in the Health Sciences. This issue arena sees an emergent Insulin Resistance (IR) paradigm contesting the dominant “cholesterol” model of chronic disease development. Health experts mostly shared this article, and their profiles reflected support for the emergent IR paradigm. We identified that these professionals followed a wider range of deliberation practices, than previously described by quantitative SciComm Twitter studies. Practices ranged from being included as part of a lecture-reading list, to language localisation in translating the article's title from English to Spanish, and study participants mentioning being involved. Contributing under their genuine identities, expert contributors carried the formal norms for civil communication into the scientific Twitter genre. There were no original URL shares from IR critics, suggesting how sharing evidence for an unconventional low-carbohydrate, healthy fats approach might be viewed as undermining orthodox identities. However, critics did respond with pro-social replies, and constructive criticism linked to the article's content, and its methodological limitations.
The statistician's semantic network analysis (SNA) confirmed that terms used by the article's tweeters related strongly to the article's content, and its discussion was pro-social. A few prominent IR individual advocates and organisations shared academic links to the article repeatedly, with its most influential tweeters and sharers being from England and South Africa. In using Atlas.ti and MAXQDA's tools for automated sentiment analysis, the statistician found many instances where sentiment was inaccurately described as negative when it should have been positive. This suggested a methodological limitation of quantitative approaches, such as QDAS, in (i) accurately analysing microblogging data. The SNA also uncovered concerns with (ii) incorrect automated counts for link shares. Concerns i & ii indicate how microblogging statistics may oversimplify complex categories, leading to inaccurate comparisons. In response, close readings of microblogging data present a distinct opportunity for meta-critique. Qualitative research can support critiques of microblogging data sources, as well as its use in QDAS. A lack of support for static Twitter data spreadsheet analysis was concerning.
Meta-inferences were then derived from the two methods' varied claims above. These findings flagged the importance of contextualising a health science article's sharing in relation to tweeters' professional identities and stances on what is healthy. In addition, meta-critiques spotlighted challenges with preparing accurate tweet data, and their analysis via qualitative data analysis software. Such findings suggest the valuable contributions that qualitative research can make to research with microblogging data in science communication.
The manuscript's development history
In 2020, Dr Pat Harpur and I selected an outlier IR scientific publication based on its unusually high Twitter popularity. At that time, the editorial, 'It is time to bust the myth of physical inactivity and obesity: you cannot outrun a bad diet' had been tweeted about over 3,000 times (now nearing 4,000 according to Altmetric!). However analysing this highly popular outlier stalled after its static export in qualitative data analysis software proved unsuitable for efficient coding. The large quantum of tweet data also proved very difficult to analyse. Accordingly, we shifted focus to a popular article that had been shared as an episode of a broader, long-running IR versus cholesterol debate. Even with its relatively small volume of tweets, organising this data for qualitative analysis proved challenging. For example, it was necessary to refine the Python extraction code, while cross-checks of static vs Twitter search results necessitated the capture of “missing” conversations.
We originally developed a multimodal analysis of these tweets, which focused on their relationship to Twitter user's profiles, potentially reflecting a wide range of communication goals. Our manuscript was submitted in 2022 to Science Communication, where Professor Susanna Priest kindly gave in-depth feedback on changing the original manuscript's contribution to a methodological one. We tackled this through developing a rationale for qualitative research with small data in the majorly revised article, which Dr Corrie Uys did a semantic network analysis for, while I revisited the social semiotic analysis.
If you have any questions, comments or concerns about our article, please comment below.
Acknowledgements
P.S. Related research manuscript from the team
Tuesday, 26 September 2023
Noteworthy disparities with four CAQDAS tools: explorations in organising live Twitter (now known as X) data
QDAS tools that support live data extraction are a relatively recent innovation. At the time of our fieldwork, four prominent QDAS provided this: only ATLAS.ti™, NVivo™, MAXQDA™ and QDA Miner™ had Twitter data import functionalities. Little has been written concerning the research implications of differences between their functionalities, and how such disparities might contribute to contrasting analytical opportunities. Consequently, early-stage researchers may experience difficulties in choosing an apt QDAS to extract live data for Twitter academic research.
To ensure that each week's Twitter data extractions could produce much data for potential evaluation, we focused on extracting and organising communiqués from the national electrical company, the Electricity Supply Commission (Eskom). ‘Load-shedding’ is the Pan South African Language Board’s word of the year for 2022 (PanSALB, 2022), due to it most frequent use in credible print, broadcast and online media. Invented as a euphemism by Eskom’s public-relations team, load-shedding describes electricity blackouts. Since 2007, planned rolling blackouts have been used in a rotating schedule for periods ‘where short supply threatens the integrity of the grid’ (McGregor & Nuttall, 2013). In the weeks up to, and during, the researchers’ fieldwork, Eskom, and the different stages of loadshedding strongly trended on Twitter. These tweets reflected the depth of public disapproval, discontent, anger, frustration, and general concern.
QDAS packages commonly serve as tools that researchers can use for four broad activities in the qualitative analysis process (Gilbert, Jackson, & di Gregorio, 2014). These are (a) organising- coding sets, families and hyperlinking; (b) exploring - models, maps, networks, coding and text searches; (c) reflecting - through memoing, annotating and mapping; and (d) integrating qualitative data through memoing with hyperlinks and merging projects (Davidson & di Gregorio, 2011; Di Gregorio, 2010; Lewins & Silver, 2014).
Please comment below if you have any questions or comments regarding our paper?
Thursday, 29 June 2023
Twitter Support must do better for helping celebrity and public victims of a global diet phishing scam!
This post presents the underwhelming example of reporting diet phishing accounts to Twitter Support as a way to spotlight the difficulties of tackling fraud via social media platforms. Hopefully publicly shaming @TwitterSupport will encourage its leaders to help address the global diet phishing scam properly, whilst also providing decent reporting options for celebrities and their representatives:
South African celebrities hijacked in fake diet adverts
A major factor in the “success" of this global scam (it has been running since 2014!) is the poor response from Facebook, Instagram, Twitter and other social media companies to formal requests to close fake accounts and their advertisement campaigns. Their ineffective responses are legally shortsighted: social media companies that repeatedly permit diet phishing ads on their platforms are complicit in a fraud, and possibly in the delict of passing off. For example, in South Africa, the diet phishing scam has undoubtedly harmed the reputation of Prof Tim Noakes and The Noakes Foundation through its fraudulent, direct misrepresentation, of fake products. These have certainly confused the public and @TheNoakesF has lost goodwill from the many victims of the fraud’s misrepresentation!Since Prof Noakes’ identity was first hijacked in 2020, The Noakes Foundation (TNF) and partners (such as Dr Michael Mol and Hello Doctor) have tried many options to stop the scam. For example, TNF developed and publicised content against it via blogposts, such as Keto Extreme Scams Social Media Users Out of Thousands. TNF also produced these videos: Professor Tim Noakes vs. Diet Phishing: Exposing a Global Scam with Fake Celebrity Endorsements, Dr Michael Mol highlighting Diet Scams and Prof Noakes Speaks Out Against The Ongoing Diet Scam. Sadly, The Noakes Foundation’s repeated warnings to the public don’t seem to be making much difference in preventing new victims!
American, Australian, British and Swedish celebs hijacked, too!
In the United States, the diet phishing scam has also stolen the identities of major celebrities. Most are in popular TV franchises: Oprah Winfrey (@Oprah), Dr Mehmet Oz (@DrOz) Dr Phil (@DrPhil), Dolly Parton (@DollyParton), Kelly Clarkson (@kellyclarkson), the Kardashian Family (@kardashianshulu + @KimKardashian), Kelly Osbourne (@KellyOsbourne), Chrissy Teigen (@chrissyteigen), Martha Maccallum (@marthamaccallum), Blake Shelton (@blakeshelton) and #TomSelleck 🥸. It’s a Magnum opus of fraud!Amazing female celebs in the United Kingdom have also seen their identities stolen. Diet phishing scammers have hijacked the IDs of Holly Willis (@hollywills), Amanda Holden (@AmandaHolden), Anne Hegerty (@anne_hegerty) and Dawn French (@Dawn_French). Even the British (@RoyalFamily) has not been immune, with the targeting of Catherine, the Princess of Wales (@KensingtonRoyal) and the Former Queen Elisabeth II, RIP and God Bless. Sadly, Meghan Duchess of Sussex, has been targeted too...
Down Under, well-known Australian personalities, such as its national treasure Maggie Beer (@maggie_beer) and Farmer Wants A Wife host Sam Armytage (@sam_armytage) have had their identities misused for fake #weightloss endorsements. And also Mr Embarrassing Bodies Down Under himself, Dr Brad McKay (@DrBradMcKay).
In Sweden, Dr Andreas Eenfeldt (@DrEenfeldt from @DietDoctor), another leader in the low carbohydrate movement, has been targeted in promotions of fake #keto products. Sadly, the fake ads seem to generate far more attention and action than his or my father's health advice!
N.B. The examples above are not extensive in terms of all victims. We largely know of celebrities in the Anglosphere whose identities were stolen, then featured in English language reports and related search engine results.
Deceptive "Tim Noakes" Twitter accounts market Keto Gummies
Just as the celebrity names stolen for the fake ads change often, so do the product names. A few examples of these fake names are Capsaicin, FigurWeightLossCapsules, Garcinia, Ketovatru and KetoLifePlus. Be warned that new "products" are added every month! One particularly common term used in the scammers' product names is "Keto Gummies". A recent Twitter search for "Tim Noakes keto gummies" suggested many fake accounts in Figure 1 (just the top view!), plus diverse "product" names.Figure 1. Twitter search results for Tim Noakes keto gummies (fake product accounts) (20 June, 2023)
Twitter Support does not think fake accounts are misleading and deceptive?!
These accounts have clearly been setup to fraudulently market "keto gummies" by suggesting an association with "Tim Noakes". So, the logical response for any representative of The Noakes Foundation would seem to be reporting each fake account for violating Twitter’s misleading and deceptive identities policy, right?Fake Twitter accounts, including those below, were reported to Twitter, with support documentation:
@NoakesGumm28693 0327118996 @TimNoakesHoax 0327120384
@TimGummies 0327119602 @NoakesGumm91126 0327119675
@gummies_tim 0327120030 @TimNoakes_ZA 0327119741
@tim_gummies 0327118910 @NoakesSouth 0327118634
@timnoakesketo0 0327119362 @NoakesGumm22663 0327119487
In each case, @TwitterSupport replied that the following accounts are NOT in violation of Twitter’s misleading and deceptive identities policy. This would seem to contradict the obvious evidence that Tim Noakes' name has been hijacked by scammers for misleading victims with a fake product!
This "Tim Noakes keto gummies" Twitter account is not deceptive?!
Figure 3 shows a typical example of a fake account's style. It uses Tim Noakes' name, plus stock photography in marketing a non-existent product. It only tweeted on May the 24th, and is followed by one person. Any knowledgeable complaint reviewer would surely consider this to be a case of a scammer creating a misleading and deceptive account for gaming Twitter's search engine. However, Twitter Support does not agree, nor explain why in its generic correspondence around each scam account.