class: bottom, left, title-slide .title[ # Who Does(n’t) Target You? ] .subtitle[ ## Mapping the Worldwide Usage of Online Political Microtargeting ] .author[ ###
Fabio Votta
, Simon Kruschinski
(University of Mainz)
, Mads Hove, Tom Dobber, Natali Helberger, and Claes de Vreese
(University of Amsterdam)
] .date[ ###
favstats.github.io/ddc2023 (Slides)
favstats
favstats@fosstodon.org
favstats.bsky.social 13th October 2023 - DDC Manchester Workshop ] --- layout: true <div class="logo"></div> --- class: white, middle, center ### .font80[Is Political Microtargeting Destroying Democracy? Or is it a Myth?] <img src="img/cambridge_analytica.png" width="80%"> --- class: white, middle, center ### How does (online) political microtargeting ### manifest in practice? ![](img/diss.png) --- class: white, middle, center ## **Two** Gaps in the Literature #### 1. No or limited data (on actual targeting strategies) How can we know how microtargeting is used in practice? <br> #### 2. Focus on the Global North Which countries do we study? --- class: middle, center ### 1. No or limited data (on actual targeting strategies) #### How can we know how microtargeting is used in practice? --- ### Three Sources of Data + Ad Libraries + Browser extensions + Interviews --- class: white ### Ad Libraries .pull-left[ + Ad Libraries *(e.g. Fowler et al., 2021; Kruschinski et al., 2022)* + Meta Ad Library, Google Ad Transparency Report, etc. + some improvements to before ] .pull-right[ ![](img/adlib.png) ] --- class: white ### Ad Libraries .pull-left[ + Ad Libraries *(e.g. Fowler et al., 2021; Kruschinski et al., 2022)* + Meta Ad Library, Google Ad Transparency Report, etc. + some improvement to before *however* + inaccurate and lack targeting criteria + *(Edelson et al., 2020; Leerssen et al., 2019)* + In the EU now provide additional info + on Meta: targeting based on age, gender, location (*thank you DSA*) ] .pull-right[ ![](img/adlib.png) ] --- class: white ### Browser extensions .pull-left[ + Browser extensions *(López Ortega, 2022; Silva et al., 2020)* + Rely on the "Why am I seeing this ad" label + ProPublica, Who Targets Me, NYU Ad Observer, etc. + Platform-independent way to retrieve data + enables measurements of effects ] .pull-right[ ![](img/browsers.png) ] --- class: white ### Browser extensions .pull-left[ + Browser extensions *(López Ortega, 2022; Silva et al., 2020)* + Rely on the "Why am I seeing this ad" label + ProPublica, Who Targets Me, NYU Ad Observer, etc. + Platform-independent way to retrieve data *however* + Risk of getting banned by platforms (NYU case 2021) + do not always correspond to the targeting criteria used by advertisers *(Andreou et al., 2018)* + who is more likely to sign-up for these browser applications? + only Desktop browsers ] .pull-right[ ![](img/browsers.png) ] --- class: white ### Interviews .pull-left[ + Interviews *(e.g. Kefford et al. 2023)* + campaign staff or platform employees + in-depth first-hand knowledge and qualitative data ] .pull-right[ ![](https://media.istockphoto.com/id/1350903564/vector/hand-holding-the-microphone-flat-design-vector-illustration-live-news-journalist-interview.jpg?s=612x612&w=0&k=20&c=-PMSMRE0hsxiaLkn7QLyas3rEFeSouVe6A7AB3_4MPM=) ] --- class: white ### Interviews .pull-left[ + Interviews *(e.g. Kefford et al. 2023)* + campaign staff or platform employees + in-depth first-hand knowledge and qualitative data *however* + campaigns might not be willing to share information + who gets to have access to political parties? ] .pull-right[ ![](https://media.istockphoto.com/id/1350903564/vector/hand-holding-the-microphone-flat-design-vector-illustration-live-news-journalist-interview.jpg?s=612x612&w=0&k=20&c=-PMSMRE0hsxiaLkn7QLyas3rEFeSouVe6A7AB3_4MPM=) ] --- class: middle, center #### current data #### **I N A D E Q U A T E** #### for studying the phenomena (or hard to access) --- class: middle, center ### 2. Focus on the Global North #### Which countries do we study? --- class: white ### Political microtargeting - a WEIRD issue? .font30[(Aagaard & Marthedal 2023 + own research)] ![](img/worldmapss.png) --- class: white, middle, center # Recent Developments --- ### Data on Ad Targeting **Meta Ad Targeting Dataset** + September 2022 Meta gave *vetted researchers* access to an "ad targeting dataset" + Includes **actual** targeting criteria of political ads on the **ad-level** + Coverage + All countries in which political ads run + August 3rd, 2020 - today (monthly updates) + Limited to using data on the Meta internal FORT platform --- class: middle, center ### For the first time we can investigate the usage of online political microtargeting at scale and worldwide **Descriptive analysis of political microtargeting** **Laying the foundation for future microtargeting research** --- ## Research Questions + RQ1: Assessing the Scale + **How prevalent is the use of political microtargeting** across the globe **and which targeting and exclusion strategies** do political campaigns employ? -- <!-- How much money do political parties spend on targeted ads vs. ads that are targeted towards entire country? --> <!-- Split up by country, party type and family/ideology. --> + RQ2: Country differences + How do different strategies of targeting and exclusion manifest across **countries**? -- + RQ3: Party differences + How do different strategies of targeting and exclusion manifest across **parties**? <!-- 1. who are the most targeted audiences --> <!-- 2. who are the most excluded audiences --> <!-- 3. how this differs by country/party type and family/ideology --> --- class: white ## Target Sample .leftcol30[ + All **national-level** elections between *August 2020 and December 2022* + We identified: + *132 countries* + *156 elections* + Only focus on **political parties and candidates** ] <!-- **122 *national* elections** in **101 countries** between --> .rightcol70[ ![](img/worldmap_coverage2.png) ] --- class: white ## Methods .leftcol[ + data annotated & enhanced using the *Party Facts database* *(Döring & Regel 2019)* + mainly based on Wikipedia + we hand-coded over 9k advertisers and 900+ parties + 100 top advertisers + 50 random sample per election + Multi-source matching + Wikipedia, Google data, also Who Targets Me + self-tagging of political advertisers + *Political party, candidate, politician, government official* ] .rightcol[ <img src="img/pf.png" width="100%"> ] --- class: white ## Sample Description .leftcol30[ + All **national-level** elections between *August 2020 and December 2022* + Time-frame: 3 months before election day + Final sample: + **95 countries** + **113 elections** + 10 countries had no data (e.g. Russia, Iran) + 27 countries no political advertisers (e.g. Venezuela, Niger) ] <!-- **122 *national* elections** in **101 countries** between --> .rightcol70[ ![](img/worldmap_coverage25.png) ] --- class: white ## Majority of Ad Spend is in the US ![](img/barrr.png) <br> + Median spent per country: **40k-207k** US Dollar + Total spent: **418m-643m** US Dollar --- class: white ## Four Targeting and Exclusion Strategies <center> <img src="img/fourcats.png" width="85%"> </center> --- class: white ## Microtargeting: Combining Criteria .leftcol[ We count targeting and exclusion combinations each time an additional criterion **reduces the audience size** 0 criterias = all ages, all genders, entire country 1 criterion = one interest audience 2 criterias = one interest audiences in specific cities 3 or more = 18-25 year olds, one interest audience in specific cities ] .rightcol[ <img src="img/fourcats.png" width="100%"> ] --- class: middle, center # Results ## RQ1: Global Prevalence and Usage --- class: white #### RQ1: Global Prevalence and Usage .leftcol40[ > *Targeting more common than excluding* + *Targeting:* Median **80.14%** of budgets are spent on ads with **at least one targeting criterion** + *Excluding:* Median **8.7%** of budgets are spent on ads with **at least one exclusion criterion** ] .rightcol60[ <center> <img src="img/targ1.png" width="76%"> </center> ] --- class: white #### RQ1: Global Prevalence and Usage > *Targeting common across geographical regions* <center> <img src="img/regionpay.png" width="80%"> </center> --- class: white #### RQ1: Global Prevalence and Usage > *Most commonly **one** targeting and exclusion criterion used* .leftcol[ + *One Criterion:* Median **38.12%** of budget ] .rightcol[ + *Two Criterias:* Median **22.46%** of budget ] <center> <img src="img/targ3.png" width="100%"> </center> --- class: white #### RQ1: Global Prevalence and Usage .leftcol30[ > *Location most prevalent* + *Location:* Median **46.92%** + *Socio-Demographics:* Median **32.22%** + *Interests & Behavior:* Median **15.20%** + *Custom & Connected Audiences:* Median **16.48%** ] .rightcol70[ <center> <img src="img/targ2.png" width="120%"> </center> ] --- class: middle, center # Results ## RQ2: Country Differences --- class: white #### RQ2: Country Differences + **Political factors** + Regulatory factors + Resource factors --- class: white #### Political factors > Targeting strategies are common regardless of level of democracy ![](img/democ1.png) --- class: white #### Political factors > Targeting strategies are common regardless of level of democracy ![](img/democ2.png) --- class: middle, center # Results ## RQ3: Party Differences --- class: white #### RQ3: Party Differences > Adoption of microtargeting does not differ between political ideology ![](img/ideo1.png) --- class: white #### RQ3: Party Differences > Adoption of microtargeting does not differ between political ideology ![](img/ideo2.png) --- class: white #### How do they use (micro)targeting? - Right-leaning parties ![](img/targettype_right.png) --- class: white #### Detailed Audiences - Right-leaning parties ![](img/detailed_right.png) --- class: white #### How do they use (micro)targeting? - Left-leaning parties ![](img/targettype_left.png) --- class: white #### Detailed Audiences - Left-leaning parties <img src="img/detailed_left.png" width="83%"> --- class: white #### .font[Top Used Detailed Audiences] > "Interest in politics" most commonly targeted <img src="img/topinterests.png" width="72%"> --- #### RQ3: Party Differences + **Which parties use microtargeting?** + No ideological difference between microtargeting focus + Older parties seem more likely use microtargeting strategies + **How do they use microtargeting?** + Parties seem to use microtargeting primarily to reach core audiences + Parties seem to ue microtargeting primarily to reach politically involved audiences --- ## Conclusions + Online political **targeting** is **common around the world** **However:** + Most commonly using a single targeting criterion + Most commonly using location and socio-demographics (e.g. age) + political microtargeting is used **by both left and right** -- **So is political microtargeting overhyped?** -- + Its usage in less-democratic countries should make us pause -- + parties **target** their respective **core audiences** + i.e. concerns about a **fragmented public sphere*** not unfounded *(Van der Meer, 2020)* + parties often target people already **interested in politics** + i.e. concerns about **"political redlining"** not unfounded *(Howard, 2005)* --- ## Conclusions + The **Ad Targeting Dataset** released by Meta is a **step in the right direction** for ad transparency + **however**, increases **dependency on Meta, should be seen critical** + dataset is **not easily accessible** for researchers, and not at all by civil society + Stepping stone for **future research** + Content! Targeting *and* tailoring --- class: middle, center # Thank you for listening <div style="text-align: center;"> <div style="display: inline-block; text-align: left;"> <p align="left"><svg viewBox="0 0 512 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:lightgrey;" xmlns="http://www.w3.org/2000/svg"> <path d="M326.612 185.391c59.747 59.809 58.927 155.698.36 214.59-.11.12-.24.25-.36.37l-67.2 67.2c-59.27 59.27-155.699 59.262-214.96 0-59.27-59.26-59.27-155.7 0-214.96l37.106-37.106c9.84-9.84 26.786-3.3 27.294 10.606.648 17.722 3.826 35.527 9.69 52.721 1.986 5.822.567 12.262-3.783 16.612l-13.087 13.087c-28.026 28.026-28.905 73.66-1.155 101.96 28.024 28.579 74.086 28.749 102.325.51l67.2-67.19c28.191-28.191 28.073-73.757 0-101.83-3.701-3.694-7.429-6.564-10.341-8.569a16.037 16.037 0 0 1-6.947-12.606c-.396-10.567 3.348-21.456 11.698-29.806l21.054-21.055c5.521-5.521 14.182-6.199 20.584-1.731a152.482 152.482 0 0 1 20.522 17.197zM467.547 44.449c-59.261-59.262-155.69-59.27-214.96 0l-67.2 67.2c-.12.12-.25.25-.36.37-58.566 58.892-59.387 154.781.36 214.59a152.454 152.454 0 0 0 20.521 17.196c6.402 4.468 15.064 3.789 20.584-1.731l21.054-21.055c8.35-8.35 12.094-19.239 11.698-29.806a16.037 16.037 0 0 0-6.947-12.606c-2.912-2.005-6.64-4.875-10.341-8.569-28.073-28.073-28.191-73.639 0-101.83l67.2-67.19c28.239-28.239 74.3-28.069 102.325.51 27.75 28.3 26.872 73.934-1.155 101.96l-13.087 13.087c-4.35 4.35-5.769 10.79-3.783 16.612 5.864 17.194 9.042 34.999 9.69 52.721.509 13.906 17.454 20.446 27.294 10.606l37.106-37.106c59.271-59.259 59.271-155.699.001-214.959z"></path></svg> favstats.github.io/ddc2023 (Slides) <br> <svg viewBox="0 0 512 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:blue;" xmlns="http://www.w3.org/2000/svg"> <path d="M459.37 151.716c.325 4.548.325 9.097.325 13.645 0 138.72-105.583 298.558-298.558 298.558-59.452 0-114.68-17.219-161.137-47.106 8.447.974 16.568 1.299 25.34 1.299 49.055 0 94.213-16.568 130.274-44.832-46.132-.975-84.792-31.188-98.112-72.772 6.498.974 12.995 1.624 19.818 1.624 9.421 0 18.843-1.3 27.614-3.573-48.081-9.747-84.143-51.98-84.143-102.985v-1.299c13.969 7.797 30.214 12.67 47.431 13.319-28.264-18.843-46.781-51.005-46.781-87.391 0-19.492 5.197-37.36 14.294-52.954 51.655 63.675 129.3 105.258 216.365 109.807-1.624-7.797-2.599-15.918-2.599-24.04 0-57.828 46.782-104.934 104.934-104.934 30.213 0 57.502 12.67 76.67 33.137 23.715-4.548 46.456-13.32 66.599-25.34-7.798 24.366-24.366 44.833-46.132 57.827 21.117-2.273 41.584-8.122 60.426-16.243-14.292 20.791-32.161 39.308-52.628 54.253z"></path></svg> favstats <br> <svg viewBox="0 0 448 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:#615ff7;" xmlns="http://www.w3.org/2000/svg"> <path d="M433 179.11c0-97.2-63.71-125.7-63.71-125.7-62.52-28.7-228.56-28.4-290.48 0 0 0-63.72 28.5-63.72 125.7 0 115.7-6.6 259.4 105.63 289.1 40.51 10.7 75.32 13 103.33 11.4 50.81-2.8 79.32-18.1 79.32-18.1l-1.7-36.9s-36.31 11.4-77.12 10.1c-40.41-1.4-83-4.4-89.63-54a102.54 102.54 0 0 1-.9-13.9c85.63 20.9 158.65 9.1 178.75 6.7 56.12-6.7 105-41.3 111.23-72.9 9.8-49.8 9-121.5 9-121.5zm-75.12 125.2h-46.63v-114.2c0-49.7-64-51.6-64 6.9v62.5h-46.33V197c0-58.5-64-56.6-64-6.9v114.2H90.19c0-122.1-5.2-147.9 18.41-175 25.9-28.9 79.82-30.8 103.83 6.1l11.6 19.5 11.6-19.5c24.11-37.1 78.12-34.8 103.83-6.1 23.71 27.3 18.4 53 18.4 175z"></path></svg> favstats@fosstodon.org <br> <svg viewBox="0 0 448 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:#0085ff;" xmlns="http://www.w3.org/2000/svg"> <path d="M400 32H48C21.5 32 0 53.5 0 80v352c0 26.5 21.5 48 48 48h352c26.5 0 48-21.5 48-48V80c0-26.5-21.5-48-48-48z"></path></svg> favstats.bsky.social</p> </div> </div> --- class: middle, center ## APPENDIX --- class: white #### RQ2: Country Differences + Political factors + Regulatory factors + Resource and Reach factors --- class: white #### Political factors > Targeting strategies are common regardless of level of democracy ![](img/democ1.png) --- class: white #### Political factors > Targeting strategies are common regardless of level of democracy ![](img/democ2.png) --- class: white #### Political factors > Proportional systems seem more likely to use microtargeting strategies than majority systems ![](img/system.png) --- class: white #### Regulatory factors > Restrictions on traditional media spending seem to encourage spending on (micro-)targeting ![](img/limits.png) --- class: white #### Regulatory factors > Countries with stricter data protection laws exhibit more spending on microtargeting ![](img/protecc.png) --- class: white #### Resource and Reach factors > GDP per capita correlates more positively with spending on microtargeting ![](img/gdp1.png) --- class: white #### Resource and Reach factors > GDP per capita correlates more positively with spending on (micro-)targeting ![](img/gdp2.png) --- class: white #### Resource and Reach factors > The more users on the platform as a share of population there is more spending on (micro-)targeting ![](img/ress1.png) --- class: white #### Resource and Reach factors > The more users on the platform as a share of population there is more spending on (micro-)targeting ![](img/ress2.png) --- #### RQ2: Country Differences + **Political factors** + Targeting strategies are common regardless of level of democracy + Proportional systems seem more likely to use microtargeting strategies than majority systems + **Regulatory factors** + Restrictions on traditional media spending seem to encourage spending on (micro-)targeting + The stricter data protection laws the more spending on microtargeting + **Resource and Reach factors** + GDP per capita correlates more positively with spending on (micro-)targeting + The more users on the platform as a share of population there is more spending on (micro-)targeting --- class: middle, center # Results ## RQ3: Party Differences --- class: white #### Which parties use microtargeting? > Adoption of microtargeting does not differ between political ideology ![](img/ideo1.png) --- class: white #### Which parties use microtargeting? > Adoption of microtargeting does not differ between political ideology ![](img/ideo2.png) --- class: white #### Which parties use microtargeting? > Older parties seem more likely use microtargeting strategies ![](img/age1.png) --- class: white #### Which parties use microtargeting? > Older parties seem more likely use microtargeting strategies ![](img/age2.png) --- class: white ### European Parliament takes Stance on Political Advertising Regulation .pull-left[ + Draft legislation on the transparency and targeting of political advertising + Article 12, §1c: > Targeting and ad delivery techniques referred to in this paragraph shall **not combine more than four categories of personal data**, including the location of the data subject. So how much ad spend in EU elections would fall under this proposal? #### 36.90% or 5.3 million Euros ] .pull-right[ ![](img/spend_on_targeting_five.png) ] --- class: white #### RQ2: Who are political parties including and excluding on Meta? (party-level) .pull-left[ ![](img/spend_per_targeting_overall_0.jpg) ] -- .pull-right[ ![](img/spend_per_exclusion_overall.png) ] --- class: white #### RQ2: Who are political parties including and excluding on Meta? (party-level) ![](img/spend_per_age.png) --- ### Data on Ad Targeting I + September 2022: Meta adds "Audience" tab to public Meta Ad Library + Coverage: All countries in which political ads run + Targeting criteria: Age, Gender, Location, Language, "Interests" and "Behaviour", Custom and Lookalike Audiences + Downside: only available for last 7, 30, and 90 days windows + No possibility to download data (however: I've written an R package `metargetr` to retrieve the data) ![](img/targeting2.png) --- #### RQ1 How big is the phenomena of Political Microtargeting? ### First Issue: Meta Ad Targeting Dataset is Messy *How do we identify political actors advertising on Meta?* 1. Hand-labeled data: Netherlands, Denmark, and Germany 2. Self-identified labels as "political" party/candidate etc. - scrape this from the Facebook Ad Library (have done this for 180k+ advertisers) ![](img/tags.png) --- ### Netherlands ![](img/NL) --- ### Germany ![](img/DE) --- ### Denmark ![](img/DK) --- ### Second Issue: Which party do advertisers belong to? 1. Matching using party names and abbreviations -> can be messy! + this may include also searching for content of ads 2. (Automated) Wikipedia search (especially for candidates) 3. Let ChatGPT do it? ;') --- ## Validation of Automated Matching 1. Make sure we are not missing any influential spenders + Retrieve 50 Top Spenders for each election + Retrieve 50 Top Spenders who were automatically not matched + hand-label them 2. From the rest, draw a random sample of 100 matched and non-matched advertisers + hand-label them + make sure we also have smaller advertisers cause they count as well 20 Elections X 200 Hand-labelling = 4000 coding task --- ## Questions Does anyone have ideas about how to 1. maximize coverage (more countries) 2. keep data quality high (other validation methods) One other idea: classify with machine learning model that is out there? I know Google can identify names and logos from images.. but can it also identify political party from names? --- ## Conclusion End the paper with a research agenda: what should we research in the future? **Thanks for listening!** ![](https://media3.giphy.com/media/0WHHnRNffwKTo4BvTb/giphy.gif) --- class: middle, center ## Some Rest --- <br> + February 2021 Meta gives *vetted researchers* access to an "ad targeting dataset" + **Actual** targeting criteria used in Social Issue, Electoral, and Political (SIEP) ads on the **ad-level** + through the FORT platform (**F**acebook **O**pen **R**esearch and **T**ransparency) + Timeframe: August 3rd - November 1st , 2020 + Coverage: Only the United States + Difficult to get access --- ### Previous studies suggest advertisers on Meta primarily target own supporters **Fowler et al. 2021** conclude in their comprehensive study on Meta ad targeting in the 2018 US midterm election > "all point[s] toward the use of social media ads for **mobilization of existing supporters** as opposed to persuasion of marginal voters" <!-- based on three factors: --> <!-- + reduced negativity --> <!-- + lower issue content --> <!-- + increased partisanship --> <!-- compared to TV ads --> **Ridout et al. 2021:** > More than 50% of ads are **acquisition, fundraising, and mobilization** ads in selected Senate races on Meta. **Stuckelberger and Koedam 2022:** > Our analysis across five countries [...] suggests that **coalition maintenance** is the dominant party strategy for demographic groups. --- class: white ### Evidence that parties target supporters (WTM data) <img src="img/wtm.png" width="95%" /> --- class: middle, center ## Theories for who gets targeted & ## who gets excluded --- ### Theories for who gets targeted & who gets excluded #### Coalition Maintenance vs. Expansion strategies *Based on Rohrschneider 2002; Panagopoulos and Wielhouwer 2008; Stuckelberger and Koedam 2022* <br> + **Coalition maintenance strategy** (*"mobilizing"*) + Reach out to people who are past and present supporters to strengthen coalition + **Coalition expansion strategy** (*"chasing"*) + Target people who are potential voters or yet undecided to expand coalition --- ### Theories for who gets targeted & who gets excluded #### **Consideration sets** *(Based on Oscarsson and Rosema 2019)* + Originally a marketing theory on consumer behaviour: + set of products that consumer would be willing to buy based on certain criteria + for example, price, brand, etc. + In electoral context: + set of political parties citizen would vote for based on certain criteria + e.g. ideological alignment, chances to govern, be in parliament, etc. <br> + **In targeting context (from perspective of campaigns):** + set of audiences that a party chooses to advertise to + e.g. based on **issue ownership**, past vote, ideological alignment, etc. + implies: audiences outside of consideration sets should be excluded --- class: white ### A theory for who gets targeted & who gets excluded ![](img/theory1.png) --- class: white ### A theory for who gets targeted & who gets excluded ![](img/theory2.png) --- class: white ### A theory for who gets targeted & who gets excluded ![](img/theory3.png) --- class: white ### A theory for who gets targeted & who gets excluded ![](img/theory4.png) --- class: white ### A theory for who gets targeted & who gets excluded ![](img/theory5.png) --- ## Hypotheses > Parties target audiences within consideration set > Parties exclude audiences outside their consideration set Maybe: > Smaller parties are more likely to engage in coalition expansion rather than coalition maintenance in an effort to gain votes from bigger competitors --- ## Requirements 1. Decide which elections should be studied + Suggestion: Netherlands 2021, Germany 2021, Canada 2021, Italy 2022, Sweden 2022, Denmark 2023 (?) + other elections: Lithuania 2020, Portugal 2020, Liechtenstein 2021, Bulgaria 2021, Cyprus 2021, Norway 2021, Iceland 2021, Czech Republic 2021, Hungary 2022, Slovenia 2022 2. Coding of targeting criteria into + **in consideration set** + Supporters + Potential voters + **outside of consideration set** + People who are not likely to vote for party > requires some expertise in studied countries Question: what about targeting criteria that do not fit in any of these categories? --- ## Who targets who and why but.. also how? + So this research design does not incorporate at all **which messages** parties use to target people. + Is this a major flaw? Reasonable compromise? ### More questions Does the theory idea make sense? Any other suggestions? Problem with the theory: what about political parties that want to **demobilize supporters of the opposing party**? That's not accounted for. However: is this even an issue outside of the US?