--- title: "Reaching_the_Voters_Supplementary_Material" author: "Jasmin Fitzpatrick and Felix-Christopher von Nostitz" format: html code-fold: true embed-resources: true editor: visual --- #'R version 4.4.1 (2024-06-14 ucrt) -- "Race for Your Life" #Copyright (C) 2024 The R Foundation for Statistical Computing' # Data Retrieval: Export via transparency library: https://adstransparency.google.com/political?region=DE&topic=political # original export date: 22/10/2021, 19:36 CET #working file: google-political-ads-creative-stats.csv #additional information: https://support.google.com/adspolicy/answer/6014595?hl=en#zippy= ```{r} #Prep of source data # Package readr version 2.1.5 documented via https://readr.tidyverse.org/ library(readr) ####the source is real real.data <- read.csv("google-political-ads-creative-stats.csv", sep=',') View(real.data) ``` ```{r} ####creating working file data.prep <- real.data ``` ```{r} ####adjusting observation period data.prep$Date_Range_Start <- gsub("-", "", data.prep$Date_Range_Start) data.prep$Date_Range_End <- gsub("-", "", data.prep$Date_Range_End) data.prep01 <- subset(data.prep, Date_Range_Start > 20210726 & Date_Range_End < 20210927) View(data.prep01) ####adjusting geo area to D for Germany data.prep02 <- subset(data.prep01, Regions != "US") View(data.prep02) data.DE<- subset(data.prep02, Regions == "DE, EU") View(data.DE) #5728 obs ``` ```{r} ####cleaning from individual, non-party advertisers ##### this step was conducted in several single steps, ##### the documentation below includes info on the individual or non-party advertiser #####which is provided as a justification for exclusion from the data set ######Sebastian Münzenmaier, AfD politician data.clean <- subset(data.DE, Advertiser_Name != "Sebastian Münzenmaier") #5699 obs ######Urban Media GmbH, publisher data.clean <- subset(data.clean, Advertiser_Name != "Urban Media GmbH") #5695 obs ######Sabine Buder, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Sabine Buder") #5687 obs ######Jorinde Schulz, LEFT politician data.clean <- subset(data.clean, Advertiser_Name != "Jorinde Schulz") #5684 obs ######Julius Kröger, (Jusos Aachen?) data.clean <- subset(data.clean, Advertiser_Name != "Julius Kröger") #5683 obs ######Mareen Theil, spokesperson for MP Heilmann, CDU data.clean <- subset(data.clean, Advertiser_Name != "Mareen Theil") #5680 obs ######Lucas Brandes, Intern for MP Frank Oelsterhelweg, CDU data.clean <- subset(data.clean, Advertiser_Name != "Lucas Brandes") #5663 obs ######Campact e.V., NGO supporting democracy and social justice mainly through petitions data.clean <- subset(data.clean, Advertiser_Name != "Campact e.V.") #5629 obs ######Karsten Scherschanski, FDP politician data.clean <- subset(data.clean, Advertiser_Name != "Karsten Scherschanski") #5619 obs ######Verlag der Tagesspiegel GmbH, media data.clean <- subset(data.clean, Advertiser_Name != "Verlag der Tagesspiegel GmbH") #5565 obs ######Katharina Pötter, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Katharina Pötter") #5563 obs ######Peter Stein, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Peter Stein") #5550 obs ######Verena Ute Hubertz, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Verena Ute Hubertz") #5539 obs ######Zweites Deutsches Fernsehen, media data.clean <- subset(data.clean, Advertiser_Name != "Zweites Deutsches Fernsehen") #5534 obs ######Dr. Franka Kretschmer, independent candidate for Magdeburg data.clean <- subset(data.clean, Advertiser_Name != "Dr. Franka Kretschmer") #5533 obs ######Florian Fuchs, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Florian Fuchs") #5531 obs ######Katharina Bartsch, GREEN politician data.clean <- subset(data.clean, Advertiser_Name != "Katharina Bartsch") #5529 obs ######Timo Schisanowski, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Timo Schisanowski") #5458 obs ######Daniel Föst, FDP politician data.clean <- subset(data.clean, Advertiser_Name != "Daniel Föst") #5453 obs ######Jan-Philipp Knoop, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Jan-Philipp Knoop") #5451 obs ######Jens MATHIAS STEIN, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Jens MATHIAS STEIN") #5427 obs ######Volker Brauer, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Volker Brauer") #5423 obs ######Hans-Böckler-Stiftung, political foundation data.clean <- subset(data.clean, Advertiser_Name != "Hans-Böckler-Stiftung") #5422 obs ######Philipp da Cunha, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Philipp da Cunha") #5421 obs ######DEMOCRACY Deutschland e.V., NGO supporting democracy data.clean <- subset(data.clean, Advertiser_Name != "DEMOCRACY Deutschland e.V.") #5401 obs ######Dr. Gerold Eberhard Papsch, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Dr. Gerold Eberhard Papsch") #5399 obs ######Thore Stein, AfD politician data.clean <- subset(data.clean, Advertiser_Name != "Thore Stein") #5396 obs ######Kai Whittaker, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Kai Whittaker") #5386 obs ######Die BürgerNahen e.V., local candidates for municipal election data.clean <- subset(data.clean, Advertiser_Name != "Die BürgerNahen e.V.") #5385 obs ######AfD-Fraktion im Sächs. Landtag, Party in parliament data.clean <- subset(data.clean, Advertiser_Name != "AfD-Fraktion im Sächs. Landtag") #5384 obs ######Ulrika Schöllner, independent candidate for Leipzig data.clean <- subset(data.clean, Advertiser_Name != "Ulrika Schöllner") #5383 obs ######FDP-Fraktion im Deutschen Bundestag, Party in parliament data.clean <- subset(data.clean, Advertiser_Name != "FDP-Fraktion im Deutschen Bundestag") #5380 obs ######Isabell Charlotte Steidel, GREEN politician data.clean <- subset(data.clean, Advertiser_Name != "Isabell Charlotte Steidel") #5373 obs ######Niklas Teßmann, independent candidate data.clean <- subset(data.clean, Advertiser_Name != "Niklas Teßmann") #5371 obs ######Sören Pellmann, LEFt politician data.clean <- subset(data.clean, Advertiser_Name != "Sören Pellmann") #5369 obs ######David Hess, ? data.clean <- subset(data.clean, Advertiser_Name != "David Hess") #5344 obs ######Verlag Lensing-Wolff GmbH & Co. KG, media data.clean <- subset(data.clean, Advertiser_Name != "Verlag Lensing-Wolff GmbH & Co. KG") #5324 obs ######SPD-Fraktion im Bundestag, party in parliament data.clean <- subset(data.clean, Advertiser_Name != "SPD-Fraktion im Bundestag") #5323 obs ######Ann-Veruschka Jurisch, fdp politician data.clean <- subset(data.clean, Advertiser_Name != "Ann-Veruschka Jurisch") #5318 obs ######Kai Koeser, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Kai Koeser") #5306 obs ######Henrik Werner, GREEN politician data.clean <- subset(data.clean, Advertiser_Name != "Henrik Werner") #5305 obs ######Cansel Kiziltepe, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Cansel Kiziltepe") #5303 obs ######Marcus Carsten Fuchs, independent candidate Sachsen data.clean <- subset(data.clean, Advertiser_Name != "Marcus Carsten Fuchs") #5295 obs ######Florian Westerwalbesloh, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Florian Westerwalbesloh") #5294 obs ######The ONE Campaign gGmbH, campaign agency data.clean <- subset(data.clean, Advertiser_Name != "The ONE Campaign gGmbH") #5293 obs ######campaigners Network GmbH, campaign agency data.clean <- subset(data.clean, Advertiser_Name != "campaigners Network GmbH") #5291 obs ######Neue Wege für Berlin e.V., NGO supporting change in Berlin data.clean <- subset(data.clean, Advertiser_Name != "Neue Wege für Berlin e.V.") #5283 obs ######Andrew Ullmann, FDP politician data.clean <- subset(data.clean, Advertiser_Name != "Andrew Ullmann") #5273 obs ######Volker Mayer-Lay, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Volker Mayer-Lay") #5272 obs ######Christina Daniela Stumpp, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Christina Daniela Stumpp") #5269 obs ######Dirk Stamer, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Dirk Stamer") #5267 obs ######Roman Wick (?) data.clean <- subset(data.clean, Advertiser_Name != "Roman Wick") #5264 obs ######Joachim Horst Krämer, FDP politician data.clean <- subset(data.clean, Advertiser_Name != "Joachim Horst Krämer") #5261 obs ######Saskia Ludwig, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Saskia Ludwig") #5248 obs ######Julia Söhne, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Julia Söhne") #5239 obs ######Thomas Bareiß, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Thomas Bareiß") #5236 obs ######Franziska Hoppermann, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Franziska Hoppermann") #5234 obs ######Konrad-Adenauer-Stiftung e.V., political foundation data.clean <- subset(data.clean, Advertiser_Name != "Konrad-Adenauer-Stiftung e.V.") #5229 obs ######Dr. Jens Zimmermann, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Dr. Jens Zimmermann") #5227 obs ######DIE FAMILIENUNTERNEHMER e. V., Lobbying for family owned businesses data.clean <- subset(data.clean, Advertiser_Name != "DIE FAMILIENUNTERNEHMER e. V.") #5218 obs ######Zentrum Liberale Moderne, Think Tank data.clean <- subset(data.clean, Advertiser_Name != "Zentrum Liberale Moderne") #5217 obs ######Marcus Finselberger, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Marcus Finselberger") #5214 obs ######Alexander Engelhard, CSU politician data.clean <- subset(data.clean, Advertiser_Name != "Alexander Engelhard") #5213 obs ######Petra Nicolaisen, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Petra Nicolaisen") #5197 obs ######Ronald Härtlein, Republik in Bewegung politician data.clean <- subset(data.clean, Advertiser_Name != "Ronald Härtlein") #5190 obs ######Thomas Knott, Mittelstandsinitiative (?) data.clean <- subset(data.clean, Advertiser_Name != "Thomas Knott") #5180 obs ######Oliver Sieke, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Oliver Sieke") #5166 obs ######Matthias Katsch, SPD politician data.clean <- subset(data.clean, Advertiser_Name != "Matthias Katsch") #5161 obs ######Matthias Hauer, CDU politician data.clean <- subset(data.clean, Advertiser_Name != "Matthias Hauer") #5159 obs ######Susanne Ferschl, LEFT politician data.clean <- subset(data.clean, Advertiser_Name != "Susanne Ferschl") #5158 obs ######Forum für politischen und gesellschaftlichen Informationsaustausch, Think Tank data.clean <- subset(data.clean, Advertiser_Name != "Forum für politischen und gesellschaftlichen Informationsaustausch") #5155 obs ######Jens Teutrine, FDP politician data.clean <- subset(data.clean, Advertiser_Name != "Jens Teutrine") #5148 obs ######AfD Fraktion i.d. Hbg. Bürgers, Parliamentary group data.clean <- subset(data.clean, Advertiser_Name != "AfD Fraktion i.d. Hbg. Bürgers") #5131 obs ``` ```{r} ####data.clean is maintained as milestone View(data.clean ) #5131 obs write.csv(data.clean, "data_clean.csv") ``` ```{r} ###### #### second access option #library(readr) #df <- read.csv("data_clean.csv", sep=',') df <- data.clean View(df) ``` ```{r} ################## # Package openxlsx version 4.2.5.2 documented via https://github.com/ycphs/openxlsx library(openxlsx) # Package dplyr version 1.1.4 documented via https://dplyr.tidyverse.org/ library(dplyr) ``` ```{r} ########### ####stepwise pooling each party #Recode Parties afd <- c( "AfD Landesverband Berlin", "Alternative für Deutschland", "Alternative für Deutschland (AfD)", "AfD Baden-Württemberg") b90 <- c( "BÜNDNIS 90 / DIE GRÜNEN Charlottenburg Wilmersdorf", "Bündnis 90/ Die Grünen Mecklenburg-Vorpommern", "Bündnis 90/Die Grünen Berlin", "BÜNDNIS 90/DIE GRÜNEN Kreisverband Aachen", "BÜNDNIS 90/DIE GRÜNEN Kreisverband Schweinfurt", "Bündnis 90/Die Grünen KV Stuttgart", "Bündnis 90/Die Grünen Ortsverband Gütersloh Bündnis 90/Die Grünen Gütersloh", "BÜNDNIS 90/DIE GRÜNEN Ortsverband Willich", "Bündnis 90/Die Grünen/GAL Münster", "Bündnis90/ Die Grünen", "Bündnis90/Die Grünen Kreisverband Bodensee") buc <- "Bündnis C - Christen für Deutschland" cdu <- c( "CDU Christlich Demokratische Union", "CDU Landesverband Mecklenburg/Vorpommern", "CDU Deutschlands", "Christlich Demokratische Union (CDU) im Wahlkreis 208 Neustadt - Speyer", "CDU-Kreisverband Pinneberg", "CDU Landesverband Berlin", "CDU Kreisverband Kassel-Land", "CDU Kreisverband Cuxhaven", "CDU Stadtverband Schwalmstadt", "CDU-Landesverband Baden-Württemberg", "CDU Kreisverband Görlitz", "Christlich Demokratische Union (CDU) Landesverband Thüringen", "CDU Landesverband Rheinland-Pfalz", "CDU Kreisverband Main-Tauber", "CDU Kreisverband Bremen-Nord", "CDU Kreisverband Mainz", "CDU Schwalm-Eder" ) csu <- c( "Christlich-Soziale Union Bayern eingetragener Verein", "CSU" ) fdp <- c( "Freie Demokratische Partei (FDP), Landesverband Bayern eingetragener Verein", "FDP Kreisverband Ravensburg", "Freie Demokratische Partei", "FDP Freiburg", "Freie Demokratische Partei (FDP)", "FDP Landesverband Saarland", "FDP Kreisverband Bielefeld", "Junge Liberale e.V.", "FDP Kreisverband Bochum", "FDP Kreisverband Speyer", "FDP Oberallgäu", "FDP Heidelberg", "FDP Kreisverband Heidekreis", "FDP Bayern Kreisverband München-Land Ortsverband (OV) Grünwald", "FDP Nürnberg", "FDP Kreisverband Regensburg-Stadt" ) fw <- c( "FREIE WÄHLER Niedersachsen", "Landesvereinigung FREIE WÄHLER Bayern e.V." ) humanisten <- "Partei der Humanisten" linke <- c( "DIE LINKE. Mecklenburg-Vorpommern", "DIE LINKE", "DIE LINKE. Kreisverband Hildesheim", "DIE LINKE Landesverband Berlin", "Die Linke Hessen" ) odp <- "Ökologisch-Demokratische Partei (ÖDP)" piraten <- c( "Piratenpartei Deutschland", "Piratenpartei Deutschland Landesverband Hessen" ) sgp <- "Sozialistische Gleichheitspartei (SGP)" spd <- c( "Sozialdemokratische Partei Deutschlands (SPD) SPD Ortsverein Neu Wulmstorf", "Sozialdemokratische Partei Deutschlands SPD", "Sozialdemokratische Partei Deutschlands Unterbezirk Bonn", "Sozialdemokratische Partei Deutschlands, Landesverband Berlin", "SPD-Kreisverband Rastatt/Baden-Baden", "SPD-Ortsverein Uelzen", "SPD-UB Landkreis Harburg", "SPD-Unterbezirk Aachen-Stadt", "SPD-Unterbezirk Coesfeld", "SPD-Unterbezirk Main-Taunus", "SPD Baden-Württemberg", "SPD Eimsbüttel", "SPD Kreisverband Gütersloh", "SPD Kreisverband Warendorf", "SPD Landesorganisation Hamburg", "SPD Mecklenburg Vorpommern", "SPD Ortsverein Augustfehn", "SPD Ortsverein Löningen", "SPD Ortsverein Uchte", "SPD Osnabrück", "SPD Rhein-Neckar", "SPD Schwalm-Eder" ) tierschutz <- "PARTEI MENSCH UMWELT TIERSCHUTZ (Tierschutzpartei)" volt <- "Volt Deutschland" ``` ```{r} df <- df |> mutate(advertiser = case_when( Advertiser_Name %in% spd ~ 1, Advertiser_Name %in% cdu ~ 2, Advertiser_Name %in% csu ~ 3, Advertiser_Name %in% b90 ~ 4, Advertiser_Name %in% afd ~ 5, Advertiser_Name %in% fdp ~ 6, Advertiser_Name %in% linke ~ 7, Advertiser_Name %in% fw ~ 8, Advertiser_Name %in% odp ~ 9, Advertiser_Name %in% tierschutz ~ 10, Advertiser_Name %in% piraten ~ 11, Advertiser_Name %in% humanisten ~ 12, Advertiser_Name %in% buc ~ 13, Advertiser_Name %in% volt ~ 14, Advertiser_Name %in% sgp ~ 15, TRUE ~ NA )) table(df$advertiser) ``` # Parties | Code | Party | |------|--------------------------------------| | 1 | SPD | | 2 | CDU | | 3 | CSU | | 4 | Die Grünen | | 5 | AfD | | 6 | FDP | | 7 | Die Linke | | 8 | Freie Wähler | | 9 | ÖDP | | 10 | Tierschutzpartei | | 11 | Piraten | | 12 | Humanisten | | 13 | Bündnis C - Christen für Deutschland | | 14 | Volt | | 15 | Sozialistische Gleichheitspartei | ```{r} ############ #Recode Ad type library(dplyr) df <- df |> mutate(Ad_Type_num = as.numeric(factor(Ad_Type))) table(df$Ad_Type_num) detach("package:dplyr", unload = TRUE) ``` # Ad Type | Code | Meaning | |------|---------| | 1 | Image | | 2 | Text | | 3 | Video | #Figure 1 in the Text ```{r} #Package ggplot; version: 3.5.1; full documentation: https://ggplot2.tidyverse.org/ library(ggplot2) # Create a data frame df.plot <- data.frame( category = c("SPD", "CDU", "CSU", "Greens", "AfD", "FDP", "Left", "FW", "ÖDP", "Tierschutz", "Pirates", "Humanists", "BUC", "Volt", "SGP"), values = c(374, 301, 35, 185, 2494, 1257, 76, 17, 38, 30, 9, 3, 8, 299, 3) ) #Prepare category for plotting (factor maintains order) df.plot$category <- factor(df.plot$category, levels = df.plot$category) # Create the bar plot ggplot(df.plot, aes(x = category, y = values)) + geom_bar(stat = "identity", fill = "darkslategray") + labs(title = "Parties advertising via Google services", x = "Party", y = "N of Ads") + theme_minimal() + theme(axis.text.x = element_text(angle = 45, hjust = 1)) detach("package:ggplot2", unload = TRUE) ``` #Table 1 in the text ```{r} library(dplyr) df <- df |> mutate(advertiser_reduced = case_when( Advertiser_Name %in% spd ~ 1, Advertiser_Name %in% cdu ~ 2, Advertiser_Name %in% csu ~ 3, Advertiser_Name %in% b90 ~ 4, Advertiser_Name %in% afd ~ 5, Advertiser_Name %in% fdp ~ 6, Advertiser_Name %in% linke ~ 7, Advertiser_Name %in% volt ~ 14, TRUE ~ NA ), advertiser_label = factor(case_when( advertiser_reduced == 1 ~ "SPD", advertiser_reduced == 2 ~ "CDU", advertiser_reduced == 3 ~ "CSU", advertiser_reduced == 4 ~ "Greens", advertiser_reduced == 5 ~ "AfD", advertiser_reduced == 6 ~ "FDP", advertiser_reduced == 7 ~ "Left", advertiser_reduced == 14 ~ "Volt", TRUE ~ "Other" ), levels = c("SPD", "CDU", "CSU", "Greens", "AfD", "FDP", "Left", "Volt", "Other"))) table(df$advertiser_label) table01 <- table(df$Ad_Type_num, df$advertiser_label) print(table01) ``` # Parties | Code | Party | |------|--------------------------------------| | 1 | SPD | | 2 | CDU | | 3 | CSU | | 4 | Die Grünen | | 5 | AfD | | 6 | FDP | | 7 | Die Linke | | 14 | Volt | #Figure 2 in the text ```{r} table(df$advertiser) library(ggplot2) #Package reshape 2; version: 1.4.4, full documentation: https://github.com/hadley/reshape library(reshape2) # Create a data frame matrix1 <- matrix(c(108, 42, 18, 83, 2191, 493, 21, 122, 105, 48, 5, 81, 4, 433, 19, 31, 161, 211, 12, 21, 299, 331, 36, 146), ncol = 8, byrow = TRUE) rownames(matrix1) <- c("Image", "Text", "Video") colnames (matrix1) <- c("SPD (n=374)", "CDU (n=301)", "CSU (n=35)", "Greens (n=185)", "AfD (n=2494)", "FDP (n=1257)", "Left (n=76)", "Volt (n=299)") # calculating percentages per column prop_data <- prop.table(matrix1, margin =2) print(prop_data) # DataFrame for ggplot2 df.plot01 <- as.data.frame(prop_data) df.plot01$category <-rownames(df.plot01) df_melted <- melt(df.plot01, id.vars = "category") colors <- c("Image" = "darkslategray", "Text" = "darksalmon", "Video" = "deepskyblue4") # creating bar plot ggplot2 ggplot(df_melted, aes(x = variable, y = value, fill = category)) + geom_bar(stat = "identity", position = "dodge") + labs(title = "Share of Media Types by Party", x = "Party", y = "%", fill = "Media Type") + scale_fill_manual(values = colors) + theme(axis.text.x = element_text(angle = 45, hjust = 1)) ``` # Party Level or unit as advertiser ```{r} ########################## ####stepwise pooling #Recode Party Level fed_party <- c( "CDU Christlich Demokratische Union", "Volt Deutschland", "CDU Deutschlands", "Freie Demokratische Partei", "Alternative für Deutschland", "Freie Demokratische Partei (FDP)", "Sozialdemokratische Partei Deutschlands SPD", "DIE LINKE", "Alternative für Deutschland (AfD)", "Ökologisch-Demokratische Partei (ÖDP)", "Bündnis90/ Die Grünen", "Sozialistische Gleichheitspartei (SGP)", "PARTEI MENSCH UMWELT TIERSCHUTZ (Tierschutzpartei)", "Piratenpartei Deutschland", "Ökologisch Demokratische Partei", "Bündnis C - Christen für Deutschland", "Partei der Humanisten" ) state_party <- c( "AfD Landesverband Berlin", "AfD Baden-Württemberg", "Bündnis 90/ Die Grünen Mecklenburg-Vorpommern", "Bündnis 90/Die Grünen Berlin", "CDU Landesverband Mecklenburg/Vorpommern", "CDU Landesverband Berlin", "CDU Landesverband Rheinland-Pfalz", "Christlich-Soziale Union Bayern eingetragener Verein", "CSU", "Freie Demokratische Partei (FDP), Landesverband Bayern eingetragener Verein", "FDP Landesverband Saarland", "FREIE WÄHLER Niedersachsen", "Landesvereinigung FREIE WÄHLER Bayern e.V.", "DIE LINKE Landesverband Berlin", "Die Linke Hessen", "DIE LINKE. Mecklenburg-Vorpommern", "Piratenpartei Deutschland Landesverband Hessen", "SPD Baden-Württemberg", "SPD Landesorganisation Hamburg", "SPD Mecklenburg Vorpommern", "Sozialdemokratische Partei Deutschlands, Landesverband Berlin" ) local_party <- c( "BÜNDNIS 90 / DIE GRÜNEN Charlottenburg Wilmersdorf", "BÜNDNIS 90/DIE GRÜNEN Kreisverband Aachen", "BÜNDNIS 90/DIE GRÜNEN Kreisverband Schweinfurt", "Bündnis 90/Die Grünen KV Stuttgart", "Bündnis 90/Die Grünen Ortsverband Gütersloh Bündnis 90/Die Grünen Gütersloh", "BÜNDNIS 90/DIE GRÜNEN Ortsverband Willich", "Bündnis 90/Die Grünen/GAL Münster", "Bündnis90/Die Grünen Kreisverband Bodensee", "Christlich Demokratische Union (CDU) im Wahlkreis 208 Neustadt - Speyer", "CDU-Kreisverband Pinneberg", "CDU Kreisverband Kassel-Land", "CDU Kreisverband Cuxhaven", "CDU Stadtverband Schwalmstadt", "CDU Kreisverband Görlitz", "CDU Kreisverband Main-Tauber", "CDU Kreisverband Bremen-Nord", "CDU Kreisverband Mainz", "CDU Schwalm-Eder", "FDP Kreisverband Ravensburg", "FDP Freiburg", "FDP Kreisverband Bielefeld", "FDP Kreisverband Bochum", "FDP Kreisverband Speyer", "FDP Oberallgäu", "FDP Heidelberg", "FDP Kreisverband Heidekreis", "FDP Nürnberg", "FDP Kreisverband Regensburg-Stadt", "DIE LINKE. Kreisverband Hildesheim", "Sozialdemokratische Partei Deutschlands (SPD) SPD Ortsverein Neu Wulmstorf", "Sozialdemokratische Partei Deutschlands Unterbezirk Bonn", "SPD-Kreisverband Rastatt/Baden-Baden", "SPD-Ortsverein Uelzen", "SPD-UB Landkreis Harburg", "SPD-Unterbezirk Aachen-Stadt", "SPD-Unterbezirk Coesfeld", "SPD-Unterbezirk Main-Taunus", "SPD Eimsbüttel", "SPD Kreisverband Gütersloh", "SPD Kreisverband Warendorf", "SPD Ortsverein Augustfehn", "SPD Ortsverein Löningen", "SPD Ortsverein Uchte", "SPD Osnabrück", "SPD Rhein-Neckar", "SPD Schwalm-Eder" ) youth <- c( "Junge Liberale e.V." ) library(dplyr) ####creation of new variable df <- df |> mutate(party_level = case_when( Advertiser_Name %in% fed_party ~ 1, Advertiser_Name %in% state_party ~ 2, Advertiser_Name %in% local_party ~ 3, Advertiser_Name %in% youth ~ 4, TRUE ~ NA )) table(df$party_level) prop.table(table(df$party_level)) ``` # Party Level | Code | Party | |------|--------------------------------------| | 1 | Federal level | | 2 | State level | | 3 | Local Level | | 4 | Youth Organisation | ```{r} library(ggplot2) # Create a data frame df.plot1 <- data.frame( category = c("Federal Party", "State Party", "Local Party", "Youth Organisation"), values = c(84.8, 4.8, 10.1, 0.2) ) # Create the bar plot ggplot(df.plot1, aes(x = category, y = values)) + geom_bar(stat = "identity") + labs(title = "Advertising Level or Unit", x = "Level or Unit", y = "%")+ theme_minimal() + theme(axis.text.x = element_text(angle = 45, hjust = 1)) detach("package:ggplot2", unload = TRUE) ``` # Gender Targeting ```{r} ########################## #Recode Gender Target df <- df |> mutate(target_gender = as.numeric(factor(Gender_Targeting))) table(df$target_gender) table(df$target_gender, df$Gender_Targeting) # 1= 'male,female, unknown", 2= 'Not targeted' # "Male, Female, Unknown gender" includes all genders. Therefore, this is not targeted. ``` # Gender Codes | Code | Party | |------|--------------------------------------| | 1 | Male, Female, Unknown gender | | 2 | Not targeted | # Age targeting ```{r} #Age table(df$Age_Targeting) #Explanation of Findings: 4953 ads were not focused on a certain group, another 157 were explicitly not targeted; 5 focused on the younger demographic (18-24, 25-34, unknown age) and 16 targeted the age groups 35 and older (+ unknown age) table(df$Age_Targeting, df$advertiser_label) #the FDP was the only party targeting for different age groups ``` # Findings on geotargeting (included areas) ```{r} ######################## #####pooling the different target areas was conducted by a series of steps ### first step: generating a new variable based on the geo-targeting (included areas) ##Recode Geo Target df <- df |> mutate(target_geo = as.numeric(factor(Geo_Targeting_Included))) table(df$target_geo) #write.csv(df, "df_geo.csv") ``` ```{r} #####second step: the original information included numeric and character codes in the same cell, which required looking up the respective zip-codes (web research; ChatGPT) #####and grouping them on the state level and adding the codes already indicating a town, district or state #####other codes: 161 for Germany; df$target_BL[df$target_geo %in% c(115, 79, 80, 81, 84, 85, 114, 137, 150, 151, 152, 153, 159, 160, 189, 190)] <- 1001 #Baden-Württemberg df$target_BL[df$target_geo %in% c(119, 82, 83, 86, 99, 100, 101, 117, 120, 129, 178, 184, 188, 192, 193, 194, 195, 201, 202)] <- 1002 #Bayern df$target_BL[df$target_geo %in% c(122, 5, 6, 7, 8, 9, 10, 11, 12, 13, 123, 146)] <- 1003 #Berlin df$target_BL[df$target_geo %in% c(135, 25)] <- 1004 #Brandenburg df$target_BL[df$target_geo %in% c(138, 142, 143)] <- 1005 #Bremen df$target_BL[df$target_geo %in% c(171, 37, 38, 39, 40, 41, 42, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 154, 172)] <- 1006 #Hamburg df$target_BL[df$target_geo %in% c(174, 69, 77, 106, 107, 112, 158, 180, 181)] <- 1007 #Hessen df$target_BL[df$target_geo %in% c(183, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 197)] <- 1008 #Mecklenburg-Vorpommern df$target_BL[df$target_geo %in% c(43, 60, 61, 62, 63, 64, 65, 70, 74, 97, 98, 113, 144, 145, 147, 167, 173, 175, 186, 200, 203, 204, 205)] <- 1009 #Niedersachsen df$target_BL[df$target_geo %in% c(66, 67, 68, 71, 72, 73, 75, 76, 88, 89, 90, 91, 111, 130, 131, 132, 134, 157, 168, 169, 170, 185, 206)] <- 1010 #Nordrhein-Westfalen df$target_BL[df$target_geo %in% c(92, 93, 94, 96, 102, 103, 104, 105, 108, 109, 110, 121, 133, 148, 177, 179, 196)] <- 1011 #Rheinland-Pfalz df$target_BL[df$target_geo %in% c(198, 78, 182)] <- 1012 #Saarland df$target_BL[df$target_geo %in% c(199, 1, 2, 3, 95, 128, 149, 166)] <- 1013 #Sachsen df$target_BL[df$target_geo %in% c()] <- 1014 #Sachen-Anhalt df$target_BL[df$target_geo %in% c(59, 118, 155)] <- 1015 #Schleswig-Holstein df$target_BL[df$target_geo %in% c(156, 165)] <- 1016 #Thüringen df$target_BL[df$target_geo %in% c(161, 87, 125, 126, 127, 136)] <- 1000 #Germany df$target_BL[df$target_geo %in% c(4, 124, 164)] <- 999 #Germany_East df$target_BL[df$target_geo %in% c(163, 176)] <- 998 #Germany-West df$target_BL[df$target_geo %in% c(116)] <- 997 #Germany-East_West_mix df$target_BL[df$target_geo %in% c(14, 15, 16, 17, 18, 19, 20, 22, 21, 23, 24, 139, 140, 141, 162, 191)] <- 900 #unspezifisch df$target_BL[df$target_geo %in% c(187)] <- 9999 #not targeted table(df$target_BL) ``` # Geo_Targeting_Inclusive | Code | Party | |------|--------------------------------------| | 900 | unspecific | | 997 | Germany-East_West_mix | | 998 | Germany-West_mix | | 999 | Germany-East_mix | | 1000 | Germany | | 1001 | Baden-Württemberg | | 1002 | Bayern | | 1003 | Berlin | | 1004 | Brandenburg | | 1005 | Bremen | | 1006 | Hamburg | | 1007 | Hessen | | 1008 | Mecklenburg-Vorpommern | | 1009 | Niedersachsen | | 1010 | Nordrhein-Westfalen | | 1011 | Rheinland-Pfalz | | 1012 | Saarland | | 1013 | Sachsen | | 1014 | Sachsen-Anhalt | | 1015 | Schleswig-Holstein | | 1016 | Thüringen | | 9999 | not targeted | ```{r} ############# ####Package labelled, version: 2.13.0; full documentation: https://larmarange.github.io/labelled/ library(labelled) ####Package expss, version: 0.11.6; full documentation: https://gdemin.github.io/expss/ library(expss) val_lab(df$target_BL) <- make_labels("900 unspezifisch 997 West_East_mix 998 West_mix 999 East_mix 1000 Deutschland 1001 Baden-Württemberg 1002 Bayern 1003 Berlin 1004 Brandenburg 1005 Bremen 1006 Hamburg 1007 Hessen 1008 Mecklenburg-Vorpommern 1009 Niedersachsen 1010 Nordrhein-Westfalen 1011 Rheinland-Pfalz 1012 Saarland 1013 Sachsen 1014 Sachen-Anhalt 1015 Schleswig-Holstein 1016 Thüringen 9999 not_targeted") table(df$target_BL) ``` ```{r} # Which Party tageted which State? table(df$target_BL, df$advertiser_label) #Explanation of major findings: CSU only runs in Bavaria; VOLT and FDP targeted cities (unspecific for States); AfD competed for certain districts in Hamburg; Left defended two out of four direct mandates in Berlin (Gysi and Lötzsche) and one in Leipzig, Sachsen (Pellmann); ``` # Findings on geotargeting (excluded areas) ```{r} ##Recode Geo Target df <- df |> mutate(target_geo_ex = as.numeric(factor(Geo_Targeting_Excluded))) table(df$target_geo_ex) library(dplyr) df$target_BL_ex[df$target_geo_ex %in% c(20)] <- 100 # not targeted df$target_BL_ex[df$target_geo_ex %in% c(1, 3, 19)] <- 101 # unspezifisch Land df$target_BL_ex[df$target_geo_ex %in% c(13, 14, 15, 16)] <- 102 # unspezifisch Stadt df$target_BL_ex[df$target_geo_ex %in% c(2)] <- 103 #Berlin Prenzl, Mitte, Friedrichshain, Kreuzberg df$target_BL_ex[df$target_geo_ex %in% c(4)] <- 104 # Stolzenau Niedersachsen df$target_BL_ex[df$target_geo_ex %in% c(5)] <- 105 # Petershagen (NRW), Stolzenau (NieS) df$target_BL_ex[df$target_geo_ex %in% c(6,8)] <- 106 # Austria, Switzerland df$target_BL_ex[df$target_geo_ex %in% c(7, 9)] <- 107 # Austria, Italy, Baden-Württemberg, Argenbrühl df$target_BL_ex[df$target_geo_ex %in% c(10)] <- 108 # Austria, Switzerland, Baden-Württemberg, Hessen, Sachsen, Thüringen df$target_BL_ex[df$target_geo_ex %in% c(11)] <- 109 # Baden-Württemberg df$target_BL_ex[df$target_geo_ex %in% c(12)] <- 110 # Bayern df$target_BL_ex[df$target_geo_ex %in% c(17)] <- 111 # Konstanz df$target_BL_ex[df$target_geo_ex %in% c(18)] <- 112 # Mannheim, Frankenthal, Ludwigshafen df$target_BL_ex[df$target_geo_ex %in% c(21)] <- 113 # Saarland df$target_BL_ex[df$target_geo_ex %in% c(22)] <- 114 # Schleswig-Holstein, Saarland library(labelled) library (expss) val_lab(df$target_BL_ex) <- make_labels("100 not targeted 101 unspezifisch Land 102 unspezifisch Stadt 103 Berlin Prenzl, Mitte, Friedrichshain, Kreuzberg 104 Stolzenau Niedersachsen 105 Petershagen (NRW), Stolzenau (NieS) 106 Austria, Switzerland 107 Austria, Italy, Baden-Württemberg, Argenbrühl 108 Austria, Switzerland, Baden-Württemberg, Hessen, Sachsen, Thüringen 109 Baden-Württemberg 110 Bayern 111 Konstanz 112 Mannheim, Frankenthal, Ludwigshafen 113 Saarland 114 Schleswig-Holstein, Saarland") detach("package:expss", unload = TRUE) detach("package:labelled", unload = TRUE) table(df$target_BL_ex) ``` # Geo_Targeting_Excluded | Code | Party | |------|---------------------------------------------------------------------| | 100 | not targeted | | 101 | unspecific State | | 102 | unspecific City | | 103 | Berlin Prenzl, Mitte, Friedrichshain, Kreuzberg | | 104 | Stolzenau Niedersachsen | | 105 | Petershagen (NRW), Stolzenau (NieS) | | 106 | Austria, Switzerland | | 107 | Austria, Italy, Baden-Württemberg, Argenbrühl | | 108 | Austria, Switzerland, Baden-Württemberg, Hessen, Sachsen, Thüringen | | 109 | Baden-Württemberg | | 110 | Bayern | | 111 | Konstanz | | 112 | Mannheim, Frankenthal, Ludwigshafen | | 113 | Saarland | | 114 | Schleswig-Holstein, Saarland | ```{r} # Which party excluded which areas? table(df$target_BL_ex, df$advertiser_label) #explanation of major findings: CDU: Does not compete in Bavaria; CSU: Bavaria borders Switzerland and Austria; Greens were not able to run in Saarland dur to a formal mistake when submitting the list to the electoral commission; AfD very week in These Berlin districts) ``` # Figure 3 in the text ```{r} table(df$Date_Range_Start) df$Date_Range_Start1 <- as.Date(df$Date_Range_Start, format = "%Y%m%d") table(df$Date_Range_Start1) library(ggplot2) # Convert date to Date format and create a data frame dfplot2 <- data.frame( date = as.Date(c("2021-07-27", "2021-07-28", "2021-07-30", "2021-08-01", "2021-08-02", "2021-08-03", "2021-08-04", "2021-08-05", "2021-08-06", "2021-08-07", "2021-08-08", "2021-08-09", "2021-08-10", "2021-08-11", "2021-08-12", "2021-08-13", "2021-08-14", "2021-08-15", "2021-08-16", "2021-08-17", "2021-08-18", "2021-08-19", "2021-08-20", "2021-08-21", "2021-08-22", "2021-08-23", "2021-08-24", "2021-08-25", "2021-08-26", "2021-08-27", "2021-08-28", "2021-08-29", "2021-08-30", "2021-08-31", "2021-09-01", "2021-09-02", "2021-09-03", "2021-09-04", "2021-09-05", "2021-09-06", "2021-09-07", "2021-09-08", "2021-09-09", "2021-09-10", "2021-09-11", "2021-09-12", "2021-09-13", "2021-09-14", "2021-09-15", "2021-09-16", "2021-09-17", "2021-09-18", "2021-09-19", "2021-09-20", "2021-09-21", "2021-09-22", "2021-09-23", "2021-09-24", "2021-09-25")), values = c(22, 1, 1, 25, 53, 23, 11, 8, 37, 16, 3, 49, 6, 52, 11, 27, 12, 42, 56, 30, 10, 303, 117, 128, 40, 120, 47, 303, 311, 31, 14, 15, 88, 89, 914, 209, 246, 34, 27, 31, 85, 37, 38, 278, 72, 42, 72, 49, 308, 105, 36, 22, 35, 158, 57, 72, 22, 49, 32) ) dfplot2$date <- as.Date(dfplot2$date, format = "%Y-%m-%d") min <- as.Date(as.factor("2021-07-27")) max <- NA # Plot Line plot for new ads across observation period ggplot(data = dfplot2, aes(x = date, y = values)) + geom_line(color = "darkslategray", linewidth = 1) + scale_x_date(date_labels = "%Y-%m-%d", date_breaks = "2 days", limits = c(min, max)) + theme(axis.text.x = element_text(angle = 45, hjust = 1)) + labs(x = "Date", y = "N New Ads", title = "New Ads per Day across observation period") ``` # Figure 4 in the text ```{r} library(dplyr) df <- df |> mutate(advertiser_reduced = case_when( Advertiser_Name %in% spd ~ 1, Advertiser_Name %in% cdu ~ 2, Advertiser_Name %in% csu ~ 3, Advertiser_Name %in% b90 ~ 4, Advertiser_Name %in% afd ~ 5, Advertiser_Name %in% fdp ~ 6, Advertiser_Name %in% linke ~ 7, Advertiser_Name %in% volt ~ 14, TRUE ~ NA ), advertiser_label = factor(case_when( advertiser_reduced == 1 ~ "SPD", advertiser_reduced == 2 ~ "CDU", advertiser_reduced == 3 ~ "CSU", advertiser_reduced == 4 ~ "Greens", advertiser_reduced == 5 ~ "AfD", advertiser_reduced == 6 ~ "FDP", advertiser_reduced == 7 ~ "Left", advertiser_reduced == 14 ~ "Volt", TRUE ~ "Other" ), levels = c("SPD", "CDU", "CSU", "Greens", "AfD", "FDP", "Left", "Volt", "Other"))) table(df$advertiser_label) table_data1 <- table(df$Num_of_Days, df$advertiser_label) print(table_data1) #conversion into data frame df_table1 <- as.data.frame(table_data1) colnames(df_table1) <- c("Num_of_Days", "Advertiser", "Frequency") df_table1$Num_of_Days <- as.numeric(as.character(df_table1$Num_of_Days)) df_table1 <- df_table1 %>% group_by(Num_of_Days) %>% mutate(Share = Frequency / sum(Frequency)) library(ggplot2) ggplot(df_table1, aes(x = Num_of_Days, y = Share, fill = Advertiser)) + geom_area(alpha = 0.6, linewidth = 0.5, colour = "white") + labs(x = "Number of Days", y = "Share", title = "Number of Days Ads were displayed per Party") + scale_x_continuous(breaks = seq(1, 55, by = 5), limits = c(1, 55)) + scale_fill_manual(values = c("SPD" = "#FF0000", "CDU" = "#000000", "CSU" = "#666666", "Greens" = "#009900", "AfD" = "#3366FF", "FDP" = "#FFFF00", "Left" = "#990066", "Volt" = "#663399", "Other" = "white"), labels = c("SPD", "CDU", "CSU", "Greens", "AfD", "FDP", "Left", "Volt", "Other")) + theme_minimal() + theme(axis.text.x = element_text(angle = 45, hjust = 1)) ``` # Figure 5 in the text ```{r} df$Date_Range_Startc <- as.character(df$Date_Range_Start) # Convert to Date format df$Date_Range_Start2 <- as.Date(df$Date_Range_Startc, format = "%Y%m%d") # Create table table_data2 <- table(df$Date_Range_Start2, df$advertiser_label) # Print the table print(table_data2) # Summarize the data to get the share per day for each advertiser #Package tidyr: version 1.3.1 full documentation:https://tidyr.tidyverse.org/ library(tidyr) df_summary <- df %>% group_by(Date_Range_Start2, advertiser_label) %>% summarise(Count = n()) %>% mutate(Share1 = Count / sum(Count)) %>% ungroup() %>% complete(Date_Range_Start2, advertiser_label) %>% mutate(Share1 = ifelse(is.na(Share1), 0, Share1)) # Create the proportional stacked area plot ggplot(df_summary, aes(x = Date_Range_Start2, y = Share1, fill = advertiser_label)) + geom_area(alpha = 0.6, linewidth = 0.5, colour = "white", position= "fill") + labs(x = "Date", y = "Share", title = "Proportional Share of Ads Displayed per Party Over Time") + scale_x_date(date_labels = "%Y-%m-%d", date_breaks = "5 days") + scale_fill_manual(values = c("SPD" = "#FF0000", "CDU" = "#000000", "CSU" = "#666666", "Greens" = "#009900", "AfD" = "#3366FF", "FDP" = "#FFFF00", "Left" = "#990066", "Volt" = "#663399", "Other" = "white"), labels = c("SPD", "CDU", "CSU", "Greens", "AfD", "FDP", "Left", "Volt", "Other")) + theme_minimal() + theme(axis.text.x = element_text(angle = 45, hjust = 1)) ``` ```{r} write.csv(df, "reaching_the_voter_complete.csv") ```