Picture an invented cannabis company post. It contains a mountain photograph, a close-up of a package and a caption announcing a product launch. The last line hints that the product might help with sleep. How many pieces of marketing are there? The person scrolling sees one post. A researcher studying photography has two pictures to examine. A researcher studying health claims might code a single sentence and ignore the mountains entirely.
Each choice produces a defensible count. Trouble starts when an analysis changes units halfway through, or when a headline turns the number of messages a company published into the number of people who saw them.
Five companies, one platform, one year
Marthe Ongenaert and Tom Decorte assembled 3,755 original posts from five prominent North American cannabis companies. The posts appeared in 2022 on the platform then called Twitter. The authors combined descriptive counts with close reading and identified conventional business communications alongside messages presenting cannabis in relation to health conditions. A corporate account can discuss a quarterly report one day and make a product announcement the next. Calling every entry a product advertisement would remove a distinction the researchers wanted to investigate.
The phrase “original posts” does considerable work here. A company composing a message and a thousand accounts passing it along represent different parts of the communication process. An archive of original posts describes what those companies chose to say. To study circulation, a researcher would also need evidence about reposts, timing and audiences. A collection from five companies cannot stand in for every independent retailer, personal account or platform. A Washington State study of free print publications, for example, started with advertisements that a reader could encounter by turning a page. It sampled a different piece of the market.
A photograph can change the coding decision
Consider the mountain photograph again. A landscape on its own makes no explicit claim about sleep. Alongside a package and a caption about winding down, it might contribute to a broader impression of relaxation. A study of explicit medical statements could reasonably leave the image uncoded. A study of lifestyle imagery might give it a prominent place. The question determines what details the coding rules have to preserve.
Researchers examining recurring visual conventions in cannabis promotion approached this problem by paying attention to images as communication in their own right. A useful test for a coding scheme is to give the same sample to two readers working independently. If one sees every landscape as a wellness claim and the other requires an explicit caption, their counts will diverge for a predictable reason. Writing down the rule lets another reader understand the disagreement.
One post can belong in several columns
Suppose ten invented posts include six package photographs, four price announcements and three claims about well-being. Those totals can comfortably exceed ten when a post qualifies for several categories. They would mean something different if the team forced each post into a single category according to its principal message. A figure such as “three health claims” also needs a denominator. It might mean that three posts mentioned health or that one post repeated the same claim three times.
Attribution matters just as much as arithmetic. A study of health effects in marijuana news coverage dealt with material that could mention potential benefits and risks. A reporter quoting a disputed claim and a seller endorsing it may use identical words. A search that simply counts appearances of the words can locate both passages, but it cannot supply their speakers or their stance.
There is a final reader missing from the imaginary mountain post: the person to whom it appeared. An account’s publication record supplies evidence about the message available to that reader. Whether they noticed the caption, believed it or changed a purchase would require a different observation. Counting posts carefully is valuable precisely because it leaves a clear record of what was said, who said it and how often. Those are substantial findings even before anyone asks what the audience did next.