We designed an experimental survey to study people's attitudes towards human enhancement technologies. Our aim is to understand the public's perception of these technologies and how information and specific features of the technology might influence these perceptions. We will administer the survey to a representative sample of US residents. Recruitment will occur through a professional survey company, Respondi/Bilendi.
Intervention
We administer a randomized survey experiment on attitudes toward human enhancement technologies (HETs). After a short, standardized description of HETs, each respondent is randomly assigned to evaluate one vignette that varies along pre-specified dimensions:
Benefit domain: cosmetic, health, or productivity.
Mechanism & inheritability: genetic (non-inheritable), genetic (inheritable), or non-genetic implant/device.
Purpose: restorative (return to typical levels) or enhancing (beyond typical levels).
Risk severity (probability held constant): 1% risk of mild, temporary side effects vs. 1% risk of severe, long-lasting side effects.
Distribution of productivity gains (subset of non-genetic productivity vignettes): general; complementary to college-educated work; complementary to high-school-or-less work.
Two additional randomized elements affect information salience: (i) an expectations module (shown to 50% before the vignette) and (ii) the ethics module placement (before vs. after main outcomes). The survey concludes with socio-demographics and economic-preference measures. Fieldwork is conducted by Bilendi/Respondi.
Experimental design
Overview. We run a preregistered, online survey experiment on attitudes toward human enhancement technologies (HETs). After a short standardized description of HETs, each respondent evaluates one randomly assigned vignette that varies along pre-specified dimensions. The study is fielded to a U.S. adult sample (target N≈5,000) constructed to be representative by gender, age, race/ethnicity, education, and geography, recruited by Bilendi/Respondi. A short Prolific pilot precedes fielding.
Vignette randomization. We use a factorial design that varies: (i) benefit domain (cosmetic; health; productivity), (ii) mechanism and inheritability (genetic non-inheritable; genetic inheritable; non-genetic implant/device), (iii) purpose (restorative vs. enhancing), and (iv) risk severity holding probability fixed (1% risk of mild, temporary side effects vs. 1% risk of severe, long-lasting side effects). For a subset of non-genetic productivity vignettes, we also vary the distribution of gains (general; complementary to college; complementary to high-school-or-less skills).
Additional randomized elements. Before the vignette, 50% of respondents are randomized to report expectations about societal consequences of HETs; the ethics module (global ethical-concern slider and checklist) is randomized to appear either before or after the main outcomes to test priming.
Design details
Randomization method
Randomization is implemented in Qualtrics. We use Qualtrics’ built-in Randomizer at the respondent level to assign treatments with equal probability. Independent randomization modules assign: (i) the vignette cell (benefit domain × mechanism/heritability × purpose × risk severity), (ii) the expectations prompt (shown before the vignette vs. not), and (iii) the placement of the ethics module (before vs. after outcomes). Randomization is executed by Qualtrics’ server-side RNG upon page load; assignment flags are stored in embedded data. Sampling quotas for representativeness (gender, age, race/ethnicity, education, geography) are handled by the panel provider and are separate from treatment assignment.
Randomization unit
Individual
Clusters
5,000 respondents
Observations
5,000
Arms
Planned equal allocation via Qualtrics.
Randomization assigns respondents equally across 40 vignette cells:
- Cosmetic: 3 mechanisms (genetic non-inheritable / genetic inheritable / non-genetic) × 2 purposes (restorative / enhancing) × 2 risks (1% mild / 1% severe) = 12 cells.
- Health: same structure = 12 cells.
- Productivity: Genetic non-inheritable (restorative/enhancing) × 2 risks = 4 cells. Genetic inheritable (restorative/enhancing) × 2 risks = 4 cells. Non-genetic “general” (restorative/enhancing) × 2 risks = 4 cells. Non-genetic education-specific, enhancing only (college-complementary; HS-or-less–complementary) × 2 risks = 4 cells. Total productivity = 16 cells.
Target sample: N ≈ 5,000 U.S. adults, yielding ≈125 respondents per vignette arm (5,000 / 40).
Orthogonal cross-randomizations:
- Expectations asked pre-vignette vs. not: ~2,500 vs. ~2,500.
- Ethics module before vs. after outcomes: ~2,500 vs. ~2,500.
Final realized counts will be reported ex post.
Primary outcomes
Primary adoption outcomes
Adopt (Yes/No): whether the respondent would adopt the technology if available today.
Willingness to pay (WTP): open-ended one-time USD amount among those answering “Yes.”
Reservation benefit among non-adopters (domain-specific):
Health vignettes: minimum healthy life-years required (0–10; plus “would not adopt regardless of benefit,” analyzed as a separate indicator).
Productivity vignettes: minimum annual salary increase (%) required (0–100; plus “would not adopt regardless of benefit”).
Primary policy/ethics outcomes
4. Desired regulatory stringency (0–10 slider): 0=no regulation, 10=prohibition.
5. Overall ethical concern (0–10 slider): 0=no concern, 10=significant concern.
Secondary outcomes
Attitudinal and framing outcomes
- Expectations randomization indicator: whether respondents were randomized to state expectations about societal consequences of HETs prior to the scenario (vs. not).
- Selection of specific ethical concerns (multi-select checklist: inequality, “playing God,” human nature/identity, autonomy of future generations, eugenic drift, unintended consequences, military/commercial misuse, privacy/surveillance, technological dependence, other, none).
Text-as-data outcomes
- Coded themes and sentiments from the open-ended justification (bag-of-words and dictionary/embedding methods; share mentioning specified themes).
Heterogeneity variables (not outcomes, but pre-specified moderators)
- Trust in institutions/actors; domain-specific risk tolerance; general technology orientation / early-adopter propensity; moral/morality battery; demographics (age, gender, education, income, parenthood, religion/attendance, urbanicity, health).
Cited by
No citation data available for this registration yet.