The two online sources I used were SuperAds Database, a database archiving all Super Bowl video advertisements for 2019-2021, and cnet.com for a list of 2022 Super Bowl video advertisements. As of today, February 15, 2022, the 2022 Super Bowl was two days ago (Feb. 13). Likely due to this recency, most of the advertisements from the 2022 Super Bowl were not available on SuperAds Database, so I used cnet.com to find them. In SuperAds Database, I began by filtering the pages of advertisements by year. SuperAds Database sorts its advertisements in order of popularity amongst fans. For example, there were 78 ranked advertisements for the year 2020 (Each page sorted by year in which they played during the Super Bowl contained a different number of advertisements (See Table 2)). It is important to note that the popularity of advertisements amongst fans does not influence my research in any way. Within the scope of my project, the ranking of advertisements was arbitrary.
Moving forward with the same example, I used a random number generator (numbergenerator.org) to select fifteen numbers at random from 1 to 78. The website would then give me fifteen numbers in this range, and I would use the corresponding advertisement in my database. For example, one of the numbers I received for 2020 was 61, which corresponded to an advertisement by YouTube named “What Will You Learn.” The advertisements on cnet.com were not ranked in any obvious order, so I gave all 44 of the advertisements listed a number based on the order they were presented in. I then used the same random number method to select my data points for the 2022 Super Bowl.
After finding and logging the randomized advertisements, I reduced the amount of data I would use in my project. Fifteen advertisements for four Super Bowls brought my total amount of advertisements to sixty. At the suggestion of my SOC 400 professor, Professor Hays, I reduced the amount of included advertisements to 32 as this will likely make my project more achievable within the given timespan for its completion. Therefore, I will use eight advertisements shown in each Super Bowl, trimming each dataset by seven advertisements. The advertisements I removed from the datasets correlated to their random number placement. For example, my fifteen random numbers for the 2019 data set were [23, 53, 39, 38, 28, 41, 56, 4, 20, 55, 6, 13, 58, 48, and 36]. I then cut the last seven random numbers, bringing my updated dataset numbers to [23, 53, 39, 38, 28, 41, 56, and 4].One last important note about my datasets is that they exclude advertisements for media such as television shows or films, commonly referred to as trailers. If a random number correlated with a trailer, I would have the random number generator generate a new random number to replace it (see Table 2 for replacements). The analysis of trailers is beyond the scope of my research of advertisements, and therefore they are omitted from my data.