Section 4 Data selection

Overall, 8,406 discussion threads with 1,164,303 posts were retrieved. Note that only one single post in a whole thread had to mention the search term for all posts to be collected, so many posts were about different topics unrelated to vaccination. We therefore first narrowed down the material to posts which either a) included the term “impf”, or b) followed directly to a post with the term “impf”, or c) were published in a thread whose initial post included the term “impf”. Second, we excluded posts about animal vaccination based on a keyword list (hund[e], katze[n], haustier[e], pferd[e], pony, ponies, tierarzt). Third, we included only posts published since 2012. Finally, we removed all posts with less than 20 words, because very short texts do not contain enough information for our analysis. Note that the majority of all ineligible posts were from the community Babycenter, where the discussions within one thread consisted of more posts and were much more topically diverse. Table 4.1 summarizes the number of eligible posts after each data selection step. The rows which start with “…” indicate the number of posts to which all criteria above also applied.

Table 4.1: Number of threads and eligible posts

	Urbia	Rund ums Baby	Netmoms	Babycenter	Sum
Total number of threads	2,441	3,581	489	1,895	8,406
Total number of posts	53,295	48,549	4,415	1,058,044	1,164,303
vaccine-related	31,385	31,546	4,415	66,682	134,028
… without animal vaccination	30,767	30,394	4,360	66,583	132,104
… since 2012	27,593	30,385	2,271	66,532	126,781
… with at least 20 words	21,936	23,683	1,740	51,229	98,588