%0 Generic %1 noauthororeditor %A Behnen, Philipp %A Kessler, René %A Kruse, Frlix %A Schoenmakers, Jan %A Zerr, Sergej %A Gómez, Jorge Marx %D 2020 %K %T Evidence, Scale, and Patterns of Systematic Inconsistencies in Google Trends data %U https://www.researchgate.net/publication/342014458_White_Paper_Evidence_Scale_and_Patterns_of_Systematic_Inconsistencies_in_Google_Trends_data %X Search analytics and trends data is widely used by media, politicians, economists, and scientists in various decision-making processes. Google offers such data on their Google Trends platform free of charge and assures the representativity of the samples provided. While working with this data, the German big data analytics and consulting company HASE & IGEL detected contradictions in results provided by Google Trends, which cast a doubt on this purported representativity. For example, when searching the term “Kurzarbeit” (“short-time work”) multiple times (e.g. once at 8 pm, once at 9 pm) for the same timespan (e.g. 1st quarter 2020), the values diverged considerably, producing contradictory trendlines. To analyze the patterns and scale of these deviations, we formed a team of researchers from HASE & IGEL, the Very Large Business Applications department of the University of Oldenburg, and the L3S Research Center (University of Hannover). In this paper, we demonstrate that the inconsistencies in Google Trends Data and the resulting contradictions are systematic and particularly large when analyzing timespans of eights months or less. In such cases, the representativity claimed by Google frequently can be disproved, and analyses based on such data should not be used in the decision-making process. While Google states that small deviations may occur for keywords with low search volumes, we found that this point falls short: while there is a significant correlation between search volume and the consistency of Google Trends data, this only accounts for half of the divergences at best and does not apply to all keywords equally. It is obvious that there are additional factors for those deviations that can only be explained by Google itself. When working with Google Trends data, users must be aware of the marked risks associated with the inconsistencies in the samples drawn by Google, even more so as our analysis shows that the index provided in Google Trends only allows very limited conclusions about the actual search volume.

@preprint{noauthororeditor, abstract = {Search analytics and trends data is widely used by media, politicians, economists, and scientists in various decision-making processes. Google offers such data on their Google Trends platform free of charge and assures the representativity of the samples provided. While working with this data, the German big data analytics and consulting company HASE & IGEL detected contradictions in results provided by Google Trends, which cast a doubt on this purported representativity. For example, when searching the term “Kurzarbeit” (“short-time work”) multiple times (e.g. once at 8 pm, once at 9 pm) for the same timespan (e.g. 1st quarter 2020), the values diverged considerably, producing contradictory trendlines. To analyze the patterns and scale of these deviations, we formed a team of researchers from HASE & IGEL, the Very Large Business Applications department of the University of Oldenburg, and the L3S Research Center (University of Hannover). In this paper, we demonstrate that the inconsistencies in Google Trends Data and the resulting contradictions are systematic and particularly large when analyzing timespans of eights months or less. In such cases, the representativity claimed by Google frequently can be disproved, and analyses based on such data should not be used in the decision-making process. While Google states that small deviations may occur for keywords with low search volumes, we found that this point falls short: while there is a significant correlation between search volume and the consistency of Google Trends data, this only accounts for half of the divergences at best and does not apply to all keywords equally. It is obvious that there are additional factors for those deviations that can only be explained by Google itself. When working with Google Trends data, users must be aware of the marked risks associated with the inconsistencies in the samples drawn by Google, even more so as our analysis shows that the index provided in Google Trends only allows very limited conclusions about the actual search volume.}, added-at = {2023-12-12T19:53:45.000+0100}, author = {Behnen, Philipp and Kessler, René and Kruse, Frlix and Schoenmakers, Jan and Zerr, Sergej and Gómez, Jorge Marx}, biburl = {https://www.bibsonomy.org/bibtex/2c2dbcab84a0b590daa71e78f53766e2d/admin}, day = 08, interhash = {278941c72f1aa138dcccdcd260f0b69a}, intrahash = {c2dbcab84a0b590daa71e78f53766e2d}, keywords = {}, month = {06}, timestamp = {2023-12-12T19:53:45.000+0100}, title = {Evidence, Scale, and Patterns of Systematic Inconsistencies in Google Trends data}, url = {https://www.researchgate.net/publication/342014458_White_Paper_Evidence_Scale_and_Patterns_of_Systematic_Inconsistencies_in_Google_Trends_data}, year = 2020 }