Discrimination through Optimization

Classified as probably not a substantial claim.

Fairness is not necessarily composable: a collection of classifiers that are fair in isolation do not necessarily result in a fair outcome when used as part of a larger system.

“Furthermore, recent work has demonstrated that fairness is not necessarily composable, i.e., for several notions of fairness such as individual fairness [24], a collection of classifiers that are fair in isolation do not necessarily result in a fair outcome when they are used as part of a larger system [25].”

• cites [25] 25. Cynthia Dwork and Christina Ilvento. 2018. Fairness Under Composition. In 10th Innovations in Theoretical Computer Science Conference (ITCS 2019) (Leibniz International Proceedings in Informatics (LIPIcs)), Vol. 124. 33:1-33:20. https: //doi.org/10.4230/LIPIcs.ITCS.2019.33 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Most platforms have a review process to prevent abuse or violations of their platforms' advertising policies.

“Most platforms have a review process to prevent abuse or violations of their platforms' advertising policies[8,77].”

• cites [8] 8. About Twitter Ads approval [n.d.]. About Twitter Ads Approval. https://business.twitter.com/en/help/ads-policies/ introduction-to-twitter-ads/about-twitter-ads-approval.html.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [77] 77. Giridhari Venkatadri, Piotr Sapiezyński, Elissa M. Redmiles, Alan Mislove, Oana Goga, Michelle Mazurek, and Krishna P. Gummadi. 2019. Auditing Offline Data Brokers Via Facebook's Advertising Platform. In International World Wide Web Conference. San Francisco, California, USA. sourceCould not verify

  The available Abstract discusses auditing data brokers through Facebook advertising, but does not address ad-review processes. Searches for review processes, ad approval, and advertising-policy violations found no passage supporting the claim that most platforms have such a process; the accessible copy appears to contain mainly the abstract, references, and page metadata. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Google, Twitter, and Pinterest all provide similar features, called Customer Match, Tailored Audiences, and Customer Lists, respectively.

“Google, Twitter, and Pinterest all provide similar features; these are called Customer Match[7], Tailored Audiences, and Customer Lists[61], respectively.”

• cites [7] 7. About Customer Match [n.d.]. About Customer Match. https://support.google.com/adwords/answer/6379332?hl=en. sourceCould not verify

  The Hiring Websites section (page not provided) discusses resume-search services, not advertising audience features. Searches of our full-text copy for Customer Lists, Tailored Audiences, Customer Match, and Pinterest found no passage addressing the attributed feature names. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [61] 61. New Marketing API Requirements for all Advertising Campaigns [n.d.]. New Marketing Api Re- quirements For All Advertising Campaigns. https://developers.facebook.com/blog/post/2019/08/15/ new-marketing-api-requirements-for-all-advertising-campaigns/.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Google and Pinterest offer similar features: on Google it is called Similar Audiences, and on Pinterest it is called Actalike Audiences.

“Google and Pinterest offer similar features: on Google it is called Similar Audiences[40], and on Pinterest it is called Actalike Audiences[63].”

• cites [40] 40. Muhammad Ali, et al. sourceNot checked — no full text of the source
• cites [63] 63. Javier Parra-Arnau, Jagdish Prasad Achara, and Claude Castelluccia. 2017. Myadchoices: Bringing Transparency And Control To Online Advertising. ACM Transactions on the Web 11 (2017). sourceCould not verify

  The citation to this paper is attached to the Google portion of the sentence, not the Pinterest portion. The Results (p. 19) discuss Google's ad-platform tracking, and User-Targeting Objectives (p. 4) discusses interest-based targeting, but neither identifies a Google feature called “Similar Audiences.” Searches for that name and for Google lookalike-audience targeting found no supporting passage in our copy. (LLM, conf=0.70)

  Experimental — accuracy not yet measured

May combine two claims.

Ads referring to cultural content stereotypically of most interest to Black users (e.g., hip-hop) can deliver to over 85% Black users, and those referring to content stereotypically of interest to white users (e.g., country music) can deliver to over 80% white users, even when targeted identically by the advertiser.

“Similarly, ads referring to cultural content stereotypically of most interest to Black users (e.g., hip-hop) can deliver to over 85% Black users, and those referring to content stereotypically of interest to white users (e.g., country music) can deliver to over 80% white users, even when targeted identically by the advertiser.”

May combine two claims.

In the most extreme cases, ads for jobs in the lumber industry reach an audience that is 72% white and 90% male, ads for cashier positions in supermarkets reach an 85% female audience, and ads for positions in taxi companies reach a 75% Black audience, despite identical targeted audiences.

“In the most extreme cases, our ads for jobs in the lumber industry reach an audience that is 72% white and 90% male, our ads for cashier positions in supermarkets reach an 85% female audience, and our ads for positions in taxi companies reach a 75% Black audience, even though the targeted audience specified by us as an advertiser is identical for all three.”

May combine two claims.

The fraction of ads delivered outside predefined DMAs remains fairly consistent across experiments regardless of what is advertised, introducing the same amount of noise across measurements.

“This fraction remains fairly consistent across our experiments regardless of what we advertise, thus introducing the same amount of noise across our measurements.”

The quote only partly states this claim.

Internal competition does not appear to influence the results.

“However, as we show with ads targeting all U.S. users, internal competition does not appear to influence the results.”

May combine two claims.

Ads targeting the same audience but with a creative stereotypically of most interest to men (e.g., bodybuilding) can deliver to over 80% men, and those with a creative stereotypically of most interest to women (e.g., cosmetics) can deliver to over 90% women.

“ads targeting the same audience but that include a creative that would stereotypically be of the most interest to men (e.g., bodybuilding) can deliver to over 80% men, and those that include a creative that would stereotypically be of the most interest to women (e.g., cosmetics) can deliver to over 90% women.”

May combine two claims.

Both the advertiser's budget and the content of the ad each significantly contribute to the skew of Facebook's ad delivery.

“We find that both the advertiser's budget and the content of the ad each significantly contribute to the skew of Facebook's ad delivery.”

May combine two claims.

The voter record data we obtained has 7,560,885 individuals, with 93% reporting their race as either Black or White.

“The data we obtained has 7,560,885 individuals, with 93% reporting their race as either Black or White.”

• AssumptionRests on an assumption: Users' self-reported race in voter records is accurate.

May combine two claims.

We do not observe a significant change in users' behavior to explain the swap; click through rates for both ads by men and women over time show no significant change.

“Interestingly, we do not observe a significant change in users' behavior to explain this swap: the bottom graph plots the click through rates (CTRs) for both ads by men and women over time.”

May combine two claims.

The ad with an image of bodybuilding but cosmetics text delivers closest to 50:50 across genders, but the ad with an image of cosmetics but bodybuilding text does not.

“Notably, we observe that the ad with an image of bodybuilding but cosmetics text delivers closest to 50:50 across genders, but the ad with the image of cosmetics but bodybuilding text does not.”

May combine two claims.

Higher daily budget leads to a smaller fraction of men in the audience, with Pearson's correlation ρ = -0.88, p < 10^-5 for all U.S. users and ρ = -0.73, p < 10^-3 for custom audiences.

“In both cases, we observe that changes in ad delivery due to differences in budget are indeed happening: the higher the daily budget, the smaller the fraction of men in the audience, with the Pearson's correlation of ρ = -0.88,p val < 10 -5 for all U.S. users and ρ = -0.73,p val < 10 -3 for the custom audiences.”

• Also stated in: § 5 CONCLUDING DISCUSSION · boosted

  “• Replicated and affirmed prior research suggesting that market and pricing dynamics can create conditions that lead to differential outcomes, by showing that the lower the daily budget for an ad, the fewer women it is delivered to.”

• AssumptionRests on an assumption: The experimental controls (same time, budget, targeting) are sufficient to rule out other explanations for the observed delivery skew.

May combine two claims.

We observe significant ad delivery skew along racial lines in the delivery of our housing ads, with certain ads delivering to over 72% Black users while others delivering to as little as 51% Black users.

“We observe significant ad delivery skew along racial lines in the delivery of our ads, with certain ads delivering to an audience of over 72% Black users (comparable to the baseline results) while others delivering to an audience of as little as 51% Black users.”

• AssumptionRests on an assumption: The experimental controls (same time, budget, targeting) are sufficient to rule out other explanations for the observed delivery skew.
• AssumptionRests on an assumption: The housing ads' implied cost through wording and image is perceived as intended.

May combine two claims.

The bodybuilding ad was delivered to over 75% men on average, while the cosmetics ad was delivered to over 90% women on average.

“In particular, the bodybuilding ad ended up being delivered to over 75% men on average, while the cosmetics ad ended up being delivered to over 90% women on average.”

May combine two claims.

In the all U.S. users experiment, the male visible and invisible images deliver to approximately 60% men, while the female visible and invisible images deliver to approximately 45% men.

“In Figure 6B we observe that the ad delivery differences are, indeed, even greater: the male visible and invisible images deliver to approximately 60% men, while the female visible and invisible images deliver to approximately 45% men.”

May combine two claims.

Neutral ads are seen by a relatively balanced, 45% white audience, while the audiences receiving the country and hip-hop ads are 80% and 13% white, respectively.

“Neutral ads are seen by a relatively balanced, 45% white audience, while the audiences receiving the country and hip-hop ads are 80% and 13% white, respectively.”

• AssumptionRests on an assumption: The entertainment ads' stereotypical associations (country music = white, hip-hop = Black) are valid.

May combine two claims.

Our five ads for lumber industry positions deliver to over 90% men and over 70% white users in aggregate, while our five ads for janitors deliver to over 65% women and over 75% black users in aggregate.

“We can immediately observe drastic differences in ad delivery across our ads along both racial and gender lines: our five ads for positions in the lumber industry deliver to over 90% men and to over 70% white users in aggregate, while our five ads for janitors deliver to over 65% women and over 75% black users in aggregate.”

• Also stated in: § 5 CONCLUDING DISCUSSION · boosted

  “• Confirmed that skewed delivery can take place on real-world ads for housing and employment opportunities by running a series of employment ads and housing ads with the same targeting parameters and bidding strategy.”

• AssumptionRests on an assumption: The experimental controls (same time, budget, targeting) are sufficient to rule out other explanations for the observed delivery skew.

Probably not a substantial claim: may not pass S4 (based).

The images with 98% opacity are not discernible to the human eye.

“As the reader can see, these images are not discernible to the human eye.”

May combine two claims.

The base bodybuilding ad delivers to 48% men, while the base cosmetics ad delivers to 40% men.

“First, we can observe an ad delivery difference due to the destination URL itself; the base bodybuilding ad delivers to 48% men, while the base cosmetics ad delivers to 40% men.”

May combine two claims.

The fraction of men in the delivery of male ads is significantly higher than in female-centered and neutral ads, and higher in neutral ads than in female-centered ads.

“As shown in Figure 6A, we verify that the fraction of men in the delivery of the male ads is significantly higher than in female-centered and neutral ads, as well as higher in neutral ads than in female-centered ads.”

May combine two claims.

Ad delivery is skewed, with ads with stereotypically masculine images delivering to over 43% men and ads with feminine images delivering to 39% men in the custom audience experiment, and 58% and 44% respectively in the all U.S. users experiment.

“We can observe that ad delivery is, in fact, skewed, with the ads with stereotypically masculine images delivering to over 43% men and the ads with feminine images delivering to 39% men in the experiment targeting custom audiences as well as 58% and 44% respectively in the experiment targeting all U.S. users.”

May combine two claims.

We observe dramatic differences in ad delivery even though the bidding strategy is the same for all ads and each pair of ads target the same gender-agnostic audience.

“We observe dramatic differences in ad delivery, even though the bidding strategy is the same for all ads, and each pair of ads target the same gender-agnostic audience.”

May combine two claims.

Introducing the image into the ad dramatically changes delivery, returning to over 75% male for bodybuilding and over 90% female for cosmetics.

“However, once we introduce the image into the ad, the delivery changes dramatically, returning to the level of skewed delivery discussed above (over 75% male for bodybuilding, and over 90% female for cosmetics).”

May combine two claims.

The masculine invisible ads appear indistinguishable in gender breakdown from masculine visible ads, and feminine invisible ads appear indistinguishable from feminine visible ads.

“Interestingly, we also observe that the masculine invisible ads appear to be indistinguishable in the gender breakdown of their delivery from the masculine visible ads, and the feminine invisible ads appear to be indistinguishable in their delivery from the feminine visible ads.”

May combine two claims.

The images with the alpha channel are almost invisible to humans, but are still delivered in a skewed manner.

“The images with the alpha channel are almost invisible to humans, but are still delivered in a skewed manner.”

Classified as probably not a substantial claim. The quote only partly states this claim.

All of our ads for lumberjacks deliver to an audience of primarily white and male users.

“we observe that all of our ads for lumberjacks deliver to an audience of primarily white and male users”

May combine two claims.

When running experiments measuring race and targeting specific DMAs, about 10% of ads are delivered to audiences outside the predefined DMAs, making it impossible to infer their race.

“However, when running our experiments measuring race (and targeting specific DMAs), we observe that a fraction (∼10%) of our ads are delivered to audiences outside of our predefined DMAs, thus making it impossible for us to infer their race.”

• AssumptionThe authors state this as an assumption: DMAs are a valid proxy for race.

Classified as probably not a substantial claim.

In experiments measuring racial skew, all ads target the same target audience.

“Note that in experiments that involve measuring racial skew all ads target the same target audience.”

Probably not a substantial claim: may not pass S3 (load-bearing).

We calculate 99% confidence intervals using the method recommended by Agresti and Coull.

“We calculate the lower and upper limits of the 99% confidence interval around this fraction using the method recommended by Agresti and Coull [9], defined in Equation 1:”

• cites Agresti and Coull [9] 9. Alan Agresti and Brent A Coull. 1998. Approximate Is Better Than "exact" For Interval Estimation Of Binomial Proportions. The American Statistician 52, 2 (1998), 119-126. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Facebook provides a relevance score between 1 and 10, calculated based on expected positive and negative feedback from the target audience.

“Facebook provides advertisers with an overview of how well-matched it believes an ad is with the target audience using a metric called relevance score, which ranges between 1 and 10.Facebook says [68]: Relevance score is calculated based on the positive and negative feedback we expect an ad to receive from its target audience.”

• cites [68] 68. Piotr Sapiezynski, Wesley Zeng, Ronald E. Robertson, Alan Mislove, and Christo Wilson. 2019. Quantifying The Impact Of User Attention On Fair Group Representation In Ranked Lists. In Workshop on Fairness, Accountability, Transparency, Ethics, and Society on the Web. San Francisco, California, USA. sourceNot checked — no full text of the source

May combine two claims.

Lambrecht et al. found that STEM education ads were consistently delivered more to men than to women, even though there are more female users on Facebook and they are more likely to click on ads and generate conversions.

“For example, Lambrecht et al. ran a series of ads for STEM education and found they were consistently delivered more to men than to women, even though there are more female users on Facebook, and they are known to be more likely to click on ads and generate conversions [52].”

• cites [52] 52. Juhi Kulshrestha, Motahhare Eslami, Johnnatan Messias, Muhammad Bilal Zafar, Saptarshi Ghosh, Krishna Gummadi, and Karrie Karahalios. 2017. Quantifying Search Bias: Investigating Sources Of Bias For Political Searches In Social Media. In ACM Conference on Computer Supported Cooperative Work. Portland, Oregon, USA.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The study ensures that each racial group (white and Black) from a set of DMAs has a matching number of records of the other group in the other set of DMAs.

“We ensure that each racial group (white and Black) from a set of DMAs has a matching number of records of the other group in the other set of DMAs.”

Classified as probably not a substantial claim.

Certain users on advertising platforms are more valuable to advertisers than others.

“For example, it is well-known that certain users on advertising platforms are more valuable to advertisers than others [48,55,65].”

• cites [48] 48. Matthew Kay, Cynthia Matuszek, and Sean A. Munson. 2015. Unequal Representation And Gender Stereotypes In Image Search Results For Occupations. In Annual Conference of the ACM Special Interest Group on Computer Human Interaction. sourceNot checked — no full text of the source
• cites [55] 55. Mathias Lecuyer, Riley Spahn, Yannis Spiliopolous, Augustin Chaintreau, Roxana Geambasu, and Daniel Hsu. 2015. Sunlight: Fine-grained Targeting Detection At Scale With Statistical Confidence. In ACM Conference on Computer and Communications Security. sourceNot checked — no full text of the source
• cites [65] 65. Ronald E. Robertson, Shan Jiang, Kenneth Joseph, Lisa Friedland, David Lazer, and Christo Wilson. 2018. Auditing Partisan Audience Bias Within Google Search. In Annual Conference of the ACM Special Interest Group on Computer Human Interaction. sourceNot checked — no full text of the source

May combine two claims.

Facebook attempts to identify users within an advertiser's selected audience who are most likely to result in an optimization event and shows the ad preferentially to those users.

“Thus, it is clear that Facebook attempts to identify the users within an advertiser's selected audience who they believe would find the ad most useful (i.e., those who are most likely to result in an optimization event) and shows the ad preferentially to those users.”

• cites [4] 4. About Ad Delivery [n.d.]. About Ad Delivery. https://www.facebook.com/business/help/1000688343301256.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

According to Facebook's documentation, 'uncategorized' gender represents users who did not list their gender.

“According to Facebook's documentation this represents the users who did not not list their gender.”

• cites Facebook's documentationNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

We carefully control for time-of-day effects by running all ads at the same time.

“We carefully control for any time-of-day effects that might be present due to different user demographics using Facebook at different times of the day: for any given experiment, we run all ads at the same time to ensure that any such effects are experienced equally by all ads.”

Classified as probably not a substantial claim. May combine two claims.

Facebook's ad delivery system is designed to show the right content to the right people, and a high relevance score is seen by the system as a positive signal.

“Facebook goes on to say [68]: Put simply, the higher an ad's relevance score, the less it will cost to be delivered.This is because our ad delivery system is designed to show the right content to the right people, and a high relevance score is seen by the system as a positive signal.”

• cites [68] 68. Piotr Sapiezynski, Wesley Zeng, Ronald E. Robertson, Alan Mislove, and Christo Wilson. 2019. Quantifying The Impact Of User Attention On Fair Group Representation In Ranked Lists. In Workshop on Fairness, Accountability, Transparency, Ethics, and Society on the Web. San Francisco, California, USA. sourceNot checked — no full text of the source

May combine two claims.

Facebook likely automatically classifies ad images, and this classification can skew delivery from the beginning of the ad's run.

“Fourth, we find that the ad image is likely automatically classified by Facebook, and that this classification can skew delivery from the beginning of the ad's run.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The study uses random PII-based custom audiences with mutually exclusive U.S. Facebook users to prevent the study's own ads from competing against each other.

“To this end, we use random PII-based custom audiences, where we randomly select U.S. Facebook users to be included in mutually-exclusive audiences.By doing so, we can ensure that our ads are only competing against each other in the cases where we wish them to.”

Classified as probably not a substantial claim. May combine two claims. The quote only partly states this claim.

We run each of the bodybuilding and cosmetics ads at the same time and with the same bidding strategy and budget.

“We run each of the ads at the same time and with the same bidding strategy and budget.”

Classified as probably not a substantial claim.

We can use this methodology to test whether there is an automatic image classifier present by examining whether running different transparent white ads results in different delivery.

“Thus, we can use this methodology to test whether there is an automatic image classifier present by examining whether running different transparent white ads results in different delivery.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The stronger effect when targeting all U.S. users may be due to the additional freedom the ad delivery system has when choosing who to deliver to, as this is a significantly larger audience.

“The stronger effect we see when targeting all U.S. users may be due to the additional freedom that the ad delivery system has when choosing who to deliver to, as this is a significantly larger audience.”

Classified as probably not a substantial claim.

To eliminate market effects in following experiments, we ensure all runs of a given experiment use the same bidding strategy and budget limit.

“To eliminate the impact that market effects can have on delivery in our following experiments, we ensure that all runs of a given experiment use the same bidding strategy and budget limit.”

Classified as probably not a substantial claim.

Prior work by Lambrecht et al. observed that their ads were predominantly shown to men even though women had consistently higher click through rates.

“In particular, they observed that their ads were predominantly shown to men even though women had consistently higher click through rates (CTRs).”

• cites Lambrecht et al. [52]Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

In budget experiments, we keep ad creative and targeted audience constant, only changing the bidding strategy to give Facebook different daily limits, so any ad delivery differences can be attributed to budget alone.

“When running these experiments, we keep the ad creative and targeted audience constant, only changing the bidding strategy to give Facebook different daily limits (thus, any ad delivery differences can be attributed to the budget alone).”

• AssumptionRests on an assumption: The experimental controls (same time, budget, targeting) are sufficient to rule out other explanations for the observed delivery skew.

Classified as probably not a substantial claim. May combine two claims.

We run the employment ads for 24 hours, using the objective of Traffic, all targeting the same audience with the same bidding strategy.

“We run these ads for 24 hours, using the objective of Traffic, all targeting the same audience with the same bidding strategy.”

May combine two claims.

The only difference between these employment ads are the ad creative and destination link; we used the same bidding strategy and target audience, and ran all ads at the same time.

“Recall that the only difference between these ads are the ad creative and destination link; we (the advertiser) used the same bidding strategy and target audience, and ran all ads at the same time.”

Classified as probably not a substantial claim. May combine two claims.

We run the budget experiment twice, once targeting random phone number custom audiences and once targeting all U.S. users, to verify the effect is not a function of the particular target audience and persists with non-custom audiences.

“Finally, we run the experiment twice, once targeting our random phone number custom audiences, and once targeting all users located in U.S.; we do so to verify that any effect we see is not a function of our particular target audience, and that it persists also when non-custom audiences are targeted.”

Classified as probably not a substantial claim. May combine two claims.

For all 25 ads, we hold the ad headline, text, and destination link constant, run them all at the same time, and use the same bidding strategy and target custom audiences in a way that each user is potentially exposed to up to three ads.

“For all 25 of these ads, we hold the ad headline, text, and destination link constant, run them all at the same time, and use the same bidding strategy and target custom audiences in a way that each user is potentially exposed to up to three ads (one masculine image, one feminine image, and one blank image).”

Classified as probably not a substantial claim. May combine two claims. The quote only partly states this claim.

All housing ads ran for 12 hours, using the objective of Traffic, all targeting the same North Carolina audiences and using the same bidding strategy.

“All of the ads ran for 12 hours, using the objective of Traffic, all targeting the same North Carolina audiences and using the same bidding strategy.”

Probably not a substantial claim: may not pass S1 (asserted). May combine two claims.

Lambrecht et al. hypothesized that higher CTRs led to women being more expensive to advertise to, meaning they were more likely to lose auctions for women compared to men.

“They then hypothesized that the higher CTRs led to women being more expensive to advertise to, meaning they were more likely to lose auctions for women when compared to auctions for men.”

• cites Lambrecht et al. [52]Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The difference in ad delivery of our invisible male and female images is statistically significant, despite humans not being able to perceive any differences in these ads.

“Thus, we can conclude that the difference in ad delivery of our invisible male and female images is statistically significant, despite the fact that humans would not be able to perceive any differences in these ads.”

Classified as probably not a substantial claim. May combine two claims.

We do not explicitly target either ad based on gender; the only targeting restrictions are 18+ year old users in the U.S.

“Note that we do not explicitly target either ad based on gender; the only targeting restrictions we stipulate are 18+ year old users in the U.S.”

Classified as probably not a substantial claim.

Discrimination in online housing ads has recently been a source of concern.

“Finally, we create a suite of ads that advertise a variety of housing opportunities, as discrimination in online housing ads has recently been a source of concern [32].”

• cites [32] 32. Facebook Engages in Housing Discrimination With Its Ad Practices, U.S. Says [n.d.]. Facebook Engages In Housing Discrimination With Its Ad Practices, U.S. Says. https://www.nytimes.com/2019/03/28/us/politics/ facebook-housing-discrimination.html.Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based).

The skew in delivery of employment and housing ads is potentially discriminatory in a legal sense.

“the skew in delivery of employment and housing ads is potentially discriminatory in a legal sense.”

May combine two claims.

These effects can be invisible to, and might even create liability for, Facebook's advertisers.

“These effects can be invisible to, and might even create liability for, Facebook's advertisers.”

Classified as probably not a substantial claim.

The study used the Bonferroni correction to address the problem of making multiple comparisons.

“we use the Bonferroni correction [14], a statistical technique used to address the problem of making multiple comparisons.”

• cites [14] 14. Robert J. Cabin and Randall J. Mitchell. 2000. To Bonferroni Or Not To Bonferroni: When And How Are The Questions. Bulletin of the Ecological Society of America 81, 3 (2000), 246-248. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The skew observed in delivery of ads for cosmetics or bodybuilding is unlikely to have legal implications.

“the skew we observe in the delivery of ads for cosmetics or bodybuilding might be interpreted as reinforcing gender stereotypes but is unlikely to have legal implications.”

May combine two claims.

Facebook banned the use of certain attributes for housing ads, but many other un-banned mechanisms exist for advertisers that achieve the same outcome.

“Facebook banned the use of certain attributes for housing ads, but many other, un-banned, mechanisms exist for advertisers that achieve the same outcome[69].”

• cites [69] 69. Showing Relevance Scores for Ads on Facebook [n.d.]. Showing Relevance Scores For Ads On Facebook. https: //www.facebook.com/business/news/relevance-score.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Facebook does not report the number of these users who match as the study uses multiple PII fields in the upload file.

“Unfortunately, Facebook does not report the number of these users who match as we use multiple PII fields in the upload file[73].”

• cites [73] 73. Upturn Amicus Brief in Onuoha v. Facebook [n.d.]. Upturn Amicus Brief In Onuoha V. Facebook. https://www. courtlistener.com/recap/gov.uscourts.cand.304918/gov.uscourts.cand.304918.76.1.pdf.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Using incongruent images skews the delivery to a lesser degree, indicating that the image is not the only element of the ad that drives the skew in delivery.

“Using incongruent images skews the delivery to a lesser degree, indicating that the image is not the only element of the ad that drives the skew in delivery.”

May combine two claims.

Ad delivery is conducted and controlled by the advertising platform itself, while ad targeting is facilitated by the platform but nominally controlled by advertisers.

“While ad targeting is facilitated by an advertising platform-but nominally controlled by advertisers-ad delivery is conducted and controlled by the advertising platform itself.”

May combine two claims.

Even absent unlawful behavior by advertisers, Facebook's own, independent actions during the delivery phase are crucial to determining how, when, and to whom ads are shown, and might produce unlawful outcomes.

“Even absent unlawful behavior by advertisers, our research demonstrates that Facebook's own, independent actions during the delivery phase are crucial to determining how, when, and to whom ads are shown, and might produce unlawful outcomes.”

May combine two claims.

The fact that ads are not delivered to an evenly split audience does not by itself indicate skew, because there might be differences in matching rates per ethnicity or different temporal usage patterns.

“solely the fact that ads are not delivered to an evenly split audience does not indicate skew, as there might be differences in matching rates (what fraction of registered voters are active Facebook users) per ethnicity, or the groups could have different temporal usage patterns.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Facebook agreed as part of a lawsuit settlement to ban age, gender, and certain kinds of location targeting, as well as some related attributes, for housing, employment, or credit ads.

“Facebook agreed as part of a lawsuit settlement stemming from these issues to go further by banning age, gender, and certain kinds of location targeting-as well as some related attributes-for housing, employment, or credit ads[22].”

• cites [22] 22. Doing More to Protect Against Discrimination in Housing, Employment and Credit Advertising [n.d.]. Doing More To Protect Against Discrimination In Housing, Employment And Credit Advertising. https://newsroom.fb.com/news/ 2019/03/protecting-against-discrimination-in-ads/.Not checked yet

  Not checked in the free run.

May combine two claims.

The paper claims to observe skew in delivery only when two or more ads are run at the same time, target the same audience, and are delivered with different proportions to white and Black users.

“Only when we run two or more ads at the same time, targeting the same audience, and these ads are delivered with different proportions to white and Black users, do we claim we observe skew in delivery.”