The Evidence Loop

• Removing paperwork friction — not persuasion — raised college enrolment by 8 percentage points; information alone produced far weaker effects (Bettinger et al., QJE, 2012).
• Personalising a court-fine text message with the debtor's name lifted average payments by 189% versus no reminder, and added a further 34% on top of a generic reminder (Haynes et al., 2013).
• Redesigning residential energy bills to show neighbours' usage cut electricity consumption by 2.0% across 600,000 households — equivalent to an 11–20% price rise, at near-zero cost (Allcott, Journal of Public Economics, 2011).
• Asking voters to state a specific voting plan — where, when, how — raised turnout by 9.1 percentage points in single-voter households (N = 287,228; Nickerson & Rogers, Psychological Science, 2010).
• Over 200 institutions worldwide now apply behavioral insights to public policy; in September 2025, Dubai's MBRSG launched the Dubai Behavioral Insights Lab to test interventions locally rather than import findings untested (OECD; MBRSG, 2025).

The discipline most behavioral economics content leaves out

Ask a room of business leaders what behavioral economics means in practice and they'll name a bias — loss aversion, anchoring, social proof. Almost none will name the discipline that separates a genuine behavioral insights unit from a slide deck with a bias glossary: the habit of running a real randomised controlled trial before trusting that an intervention works here, on this population, in this channel.

"The bias is real. The famous study is real. Whether it transfers to your population, your channel, your specific behavior — that part was never tested. It was assumed."

That gap — between knowing a bias exists and knowing it will produce a measurable effect in your context — is what The Evidence Loop is written to close.

Four trials. Four EAST levers. Zero obvious results.

The paper's evidence base is four independently verified randomised controlled trials, each demonstrating one lever from the EAST framework (Easy, Attractive, Social, Timely), each run at a scale that rules out coincidence, and each producing a result that was genuinely non-obvious in advance.

• Easy — Financial aid (Bettinger et al., QJE, 2012): Helping families complete a complex aid application using data already on hand raised college enrolment 8 percentage points. Information alone — telling families what they were eligible for — produced far weaker effects. The friction, not the awareness gap, was the binding constraint.
• Attractive — Court fines (Haynes et al., 2013): Any text reminder to pay an outstanding fine lifted average payments 145%. Adding the debtor's name pushed that to 189%. Personalisation alone — the same information, addressed to a name rather than a category — did measurable, separable work.
• Social — Home energy (Allcott, Journal of Public Economics, 2011): Showing households how their electricity use compared to similar nearby homes produced a 2.0% average reduction across 600,000 households — equivalent to an 11–20% price increase, achieved without touching the price. The effect concentrated among the heaviest users (−6.3%) and barely moved those already below the norm (−0.3%).
• Timely — Voter turnout (Nickerson & Rogers, Psychological Science, 2010): Asking people to articulate a specific voting plan — what time, where from, what they'd be doing beforehand — raised turnout 9.1 points among single-voter households in a field experiment of 287,228 people. The prompt contained no new information about the election; only the specificity and timing changed.

The Renascence point of view

The listicle version of behavioral economics treats each bias as a transferable fact. These four trials say otherwise. None of the research teams assumed their intervention would work — they ran the trial precisely because the outcome wasn't obvious. The FAFSA researchers didn't know information alone would underperform so dramatically. The energy researchers didn't know the effect would concentrate so heavily among the highest-usage households.

"Most behavioral economics content sells the bias as the insight. The actual insight is that even the people who study these biases for a living don't trust their own predictions enough to skip the trial."

The Renascence position: any organisation borrowing a mechanism from this paper should treat it as a hypothesis worth testing in its own context — not a proven fact ready to deploy.

The Evidence Loop framework

The paper's second half introduces the Evidence Loop: a five-stage cycle — Hypothesis, Design, Test, Decide, Scale or Discard — with Test as the hinge where every one of the four case studies actually lives. A companion EAST Lever Map shows which lever to match to which type of friction, and where to run a first test.

The toolkit — four instruments you can use this week

• Tool 01 · EAST Opportunity Scan: A five-minute diagnostic scoring where each lever is unused, unexamined, or deliberately designed and tested.
• Tool 02 · Hypothesis & Test Card: A one-page protocol for designing a real test — behavior, lever, control, variant, random assignment method, single metric, and a pre-committed decision rule.
• Tool 03 · Minimum-Viable-RCT Rigor Checklist: A sanity check for whether a planned test is a genuine randomised trial or a rollout wearing a lab coat.
• Tool 04 · Test Results Ledger: A running record that turns individual tests into institutional memory — including the losses, which is where the real evidence base gets built.

Who needs this paper

CX leaders, behavioral insights practitioners, policy teams, and transformation leads who have borrowed a behavioral mechanism from a case study and want to know whether it actually works for their population — and how to build the testing discipline that answers that question reliably, at scale.

The regional signal

The Gulf is building this capability directly rather than importing conclusions. In September 2025, Dubai's Mohammed Bin Rashid School of Government launched the Dubai Behavioral Insights Lab — an applied platform designed to test interventions in real government settings. Its first initiative, a structured Nudgeathon with GDRFA Dubai in November 2025, applied behavioral-science methodology to residency and identity services. No published results yet; the honest signal is that the infrastructure to test, not just to borrow, is being built.