Ad Delivery Audits by Age Group for Job Advertising

confidence: high · 2 sources

⚠
Working research — not a recommendation.

This page records what the evidence says a policy does, not what anyone should do about it. It has been synthesized from the cited sources but has not been independently reviewed.

What the evidence says

The sources on this page contain no direct audit of ad delivery by age group for job ads. They hold three platform and background statements about age and ad systems, and three findings about age in job recommendations. None of the three findings measures ad delivery by age.

Before 2019, advertisers could exclude older workers through targeting parameters. A cited prior study found that "advertisers excluded older workers from seeing job ads". Its example is financial analyst ads targeted only at users aged 25–36 years. This claim is a citation of earlier work, not the page authors' own measurement, so the evidence for the magnitude is weaker than for the direction. The mechanism is advertiser-chosen audience parameters.

Facebook then changed the targeting controls. In 2019 it disabled explicit targeting on gender, age and zip code for some ad categories, including employment. This removes the advertiser-side mechanism described above. The sources do not measure whether age skews in who sees job ads fell afterward. Facebook also describes a Variance Reduction System (VRS) that monitors age, gender and estimated race in delivery. The quoted description covers housing ads. The VRS aims to make the audience shown an ad resemble the audience the advertiser targeted. A system built to match the targeted audience does not address skew that the targeting itself creates.

A separate audit measured job recommendations to matched young and older resume pairs, with gender as the contrast. It finds "little variation across age levels and cities" in the size of gender differences. The gender gap in posted wages is "almost identical" in the young and old resume pairs. The gender gap in recommended experience requirements is largest for older profiles. These results describe gender gaps across ages. They do not show whether age itself changes which job ads a worker sees.

The evidence on delivery audits by age is thin. No source here reports an age-based delivery disparity for job ads, measured before or after 2019.

Where the evidence splits

No contradictions in the current sources.

Thresholds and dose-response

No threshold established in the current sources.

Open empirical questions

Findings

Facebook's 2019 change removes age as an advertiser targeting option for employment ads

confidence — direction: highconfidence — magnitude: highconsensus: thin. It is a single platform statement, with no outcome data.
mechanism
The platform disabled the targeting controls, so advertisers cannot select an age range for covered ad categories.
effect
Explicit age-based targeting is removed for some ad categories, including employment. No delivery outcome is measured.
conditions / moderators
Applies to "some ad categories". Whether delivery still skews by age is not established.

Advertiser targeting parameters decrease older workers' exposure to job ads

confidence — direction: highconfidence — magnitude: mediumconsensus: thin. It is one cited finding.
mechanism
Advertisers set age brackets, so users outside the bracket never enter the audience.
effect
Older workers were excluded from seeing job ads. In the cited example, financial analyst ads were targeted only at users aged 25–36 years. No exposure rate is given.
conditions / moderators
Describes the targeting era before the 2019 change. The claim comes from cited work, not the authors' own measurement.

Facebook's VRS monitors age in delivery, with no measured effect reported

confidence — direction: highconfidence — magnitude: highconsensus: thin. It is a platform policy description, with no outcome audit.
mechanism
The system compares the age, gender and estimated race distribution of the audience shown an ad with that of the audience the advertiser targeted. It aims to make the two closely resemble each other.
effect
No measured effect on age skew. The source documents the design, not the result.
conditions / moderators
The quoted description concerns housing ads. Coverage of job ads is not established in the claim.

Worker age shows no measured effect on the size of gender gaps in recommended job ads and posted wages

confidence — direction: highconfidence — magnitude: highconsensus: thin. It comes from one audit.
mechanism
Not established. The audit compares recommendations to matched resume pairs.
effect
Little variation in gender differences across age levels. The gender gap in posted wages is "almost identical" in young and old resume pairs.
conditions / moderators
Gender differences are slightly greater in gender-neutral jobs than in male- and female-dominated jobs. They are also greater in middle and high skill level jobs than in entry level jobs. Replication across the four job boards separately is not established.

Older worker profiles show larger gender gaps in recommended experience requirements

confidence — direction: highconfidence — magnitude: mediumconsensus: thin. It comes from one audit, and its quoted statement is qualitative.
mechanism
Not established.
effect
Gender experience gaps are highest for older applicant profiles and in highly ranked jobs. The exact size of the difference between age groups is not given.
conditions / moderators
Consistency across all job boards is not established.