Summary: Training Effective Altruism (Mehmood et al.)

Training Effective Altruism is a Global Priorities Institute Working Paper by Sultan Mehmood, Shaheen Naseer, and Daniel L. Chen. This post is part of my sequence of GPI Working Paper summaries.


After Pakistani deputy ministers (of which, 70% joined for “job perks and power”) watched a “utilitarian” lecture on empathy and its personal benefits, they…

  • donated 6% more for no benefit when acting as a dictator in a game and 20% more to charity at their own expense in another game (a 12% and 33% increase relative to the placebo group, respectively).

  • were 40% more likely to use “us” instead of “them” on social media and 20% more likely to use “we” instead of “I” (about double the percentages of the placebo group).

  • were 25% more likely to donate blood when urgently requested by a blood bank (an 80% increase relative to the placebo group).

    • They were only willing to give blood when the bank specifically requested their blood type. The authors state this is evidence they acted as effective altruists—they donated only when they believed it would likely be helpful.

  • were 20% more likely to choose a book on empathy (a 50% increase relative to the placebo group).

  • scored 10% higher on policy scenario assessments of teamwork and cooperation (a 20% increase relative to the placebo group).

… and four to six months after treatment, they…

  • were 20% more likely to visit an orphanage instead of a talk by a senior bureaucrat four months after the treatment (an 80% increase relative to the placebo group).

  • were 20% more likely to volunteer at an impoverished school instead of attending a senior bureaucrat’s talk, six months after the treatment.

  • rated emotional intelligence 1 point more important to policymaking (on a 5-point scale) than the placebo group, six months after the treatment.

  • performed 0.6 standard deviations higher in a teamwork workshop six months after the treatment, as scored by former Supreme Court judges, prominent academics, and former senior deputy ministers.

In contrast, after ministers watched a lecture on how empathy is “malleable” (a practicable skill that can be improved), they found no statistically significant effects. Likewise, ministers who watched a mixed lecture combining the utilitarian lecture with the malleability one had no statistically significant effects.

Experimental design


  • At a training academy in Pakistan, the researchers studied deputy ministers, who are high-stakes decision-makers and advise the President, Prime Minister, and Cabinet Ministers.

  • ~1% of applicants for the job are chosen from ~15,000 based on their exam performance, psychological assessments, and interviews.

  • ~70% reported they mainly joined for job perks and power (self-interested reasons).

  • They were randomly assigned into 4 groups, each having ~53 participants.


  • They carefully watched a lecture and wrote a brief 500-word summary.

  • The (effective) utilitarian lecture presented evidence for empathy being self-serving and narratives of benevolent former deputy ministers. The lecture stated its core message:
    “Qualitative and quantitative evidence backs the idea that showing empathy is good for you. It is not just the right thing to do but also the most sensible thing to do for your performance.”

  • The unsuccessful malleability lecture’s main point:
    “Qualitative and quantitative evidence backs the idea that empathy is not fixed but is malleable. It is a skill that can be developed.”

  • The unsuccessful mixed utilitarian-malleability group’s lecture was a mix of the two above, and the placebo group had a lecture on macroeconomics.