Findings › Mixed evidence

16 biases whose cited evidence is mixed

These are the biases stored with evidence rating “mixed”. The sentence on each row is the critique, replication note, or evidence sentence already on that bias. The link is a source stored with it.

Last updated 2026-11-04 · 16 of 96 biases · How we rate · /api/biases.json

What mixed means here

On the bias directory, mixed means credible evidence both ways, such as a large or pre-registered failed replication, a meta-analysis near zero or heavily affected by publication bias, or a live dispute that it’s an artefact. The critique is cited on the bias page. A row appears here only when the stored rating is mixed.

Each bias and the note already stored for it

Choice overload (simplification)
Meta-analysis: mean effect close to zero (Scheibehenne et al., 2010).
Mixed evidence · Source: doi.org/10.1086/651235
Cognitive dissonance
A 2024 multi-lab replication of induced compliance failed (Vaidis et al.).
Mixed evidence · Source: doi.org/10.1177/25152459231213375
No teardown in the library yet.
Compromise effect
Rated Mixed for consistency with our batch 5 treatment, until a meta-analysis is checked.
Mixed evidence · Cited on the bias page: doi.org/10.1086/209205
Decoy effect
Typically absent when products are experienced or shown perceptually (Frederick et al., 2014).
Mixed evidence · Source: doi.org/10.1509/jmr.12.0061
- Adobe’s regular Photoshop price sits above the Photography plan
- Surfshark prices Starter the same as the bigger One plan
Dunning–Kruger effect
Disputed as partly a statistical artefact (Nuhfer et al.; Gignac & Zajenkowski 2020).
Mixed evidence · Source: doi.org/10.5038/1936-4660.9.1.4 · doi.org/10.5038/1936-4660.10.1.4 · doi.org/10.1016/j.intell.2020.101449
No teardown in the library yet.
Fundamental attribution error
The evidence is mixed (provisional).
Mixed evidence · Provisional · Cited on the bias page: doi.org/10.1016/S0065-2601(08)60357-3 · doi.org/10.1037/0033-2909.125.1.47
No teardown in the library yet.
Identifiable victim effect
Maier et al. (2024) ran a pre-registered replication of studies in Small et al. (2007) and found no support for the effect, after a reanalysis suggested publication bias.
Mixed evidence · Source: doi.org/10.1525/collabra.90203
- ActionAid sells community funding through one child, and shows Harriet
Identity labelling
Gerber et al. (2016) ran a large-scale field experiment and found the noun wording had no effect on turnout.
Mixed evidence · Source: doi.org/10.1073/pnas.1513727113
- Amnesty asks you to become a member from £1 a month
Information gap (curiosity)
In subject lines and headlines it can lift attention, but the evidence is mixed and it backfires when the payoff disappoints.
Mixed evidence · Cited on the bias page: doi.org/10.1037/0033-2909.116.1.75
- Sakuto opens a question in the subject line and answers it in the first line
Licensing effect
Blanken et al. (2015) meta-analysed 91 studies (7,397 participants) and estimated d = 0.31, with larger effects in published than unpublished studies.
Mixed evidence · Source: doi.org/10.1177/0146167215572134
No teardown in the library yet.
Loss aversion
Gal & Rucker (2018) argue many effects attributed to loss aversion have other explanations.
Mixed evidence · Source: doi.org/10.1002/jcpy.1047
Mental accounting
Field evidence on how reward labels are treated is thinner than the original theory.
Mixed evidence · Cited on the bias page: doi.org/10.1287/mksc.4.3.199 · doi.org/10.1002/(SICI)1099-0771(199909)12:3%3C183::AID-BDM318%3E3.0.CO;2-F
Overjustification effect
Meta-analyses disagree (Deci et al. 1999 vs Cameron & Pierce 1994).
Mixed evidence · Source: doi.org/10.1037/0033-2909.125.6.627 · doi.org/10.3102/00346543064003363
No teardown in the library yet.
Scarcity
Scarcity effect sizes vary widely across studies. [Unverified: we haven’t yet checked this against a meta-analysis.]
Mixed evidence · Cited on the bias page: doi.org/10.1037/0022-3514.32.5.906
- Global Work & Travel frames its Early Bird as a way to “skip” a price increase
- lastminute.com runs a ticking “Ends in” countdown over its “Only here” deals
Sunk cost effect
Real, but its size depends on context (Roth et al.; Friedman et al. 2007).
Mixed evidence · Source: doi.org/10.1007/s40685-014-0014-8 · doi.org/10.1007/s10683-006-9134-0
- Duolingo sends you back to units you’ve already completed
Zero-price effect
Evidence comes from a small number of mostly lab studies.
Mixed evidence · Cited on the bias page: doi.org/10.1287/mksc.1060.0254

What this list leaves out

Biases rated strong or limited stay off this page, including ones whose page mentions a single caveat.
Where the library marks a note unverified, that mark is kept in the sentence.
Counts of how brands use a tactic are on Findings. The rating rules are on how we rate.

5 new nudges, every Friday

Five real product and marketing moments, one thing to try from each, and the psychology behind it. Weekly issues start soon. Your email is stored with our email provider (Resend) and used only for this newsletter.