Finches, BFFs and the Limits of Crowdsourcing



Raise your hand if your mother ever said anything along the lines of, “If all of your friends jumped off of a bridge, would you jump, too?” Fifty years before Facebook, parents worried about their kids being unduly influenced by the behaviors of friends. Can you imagine their anxiety had they foreseen the digital age and its propensity to create social herds? The instantaneous proliferation of opinions and the aggregation of those opinions into virtual weathervanes, spinning and pointing in whichever direction the prevailing wind seems to be blowing, have a profound influence on individual behavior, and on the choices people make.

The manifestation of crowdsourcing in health care is the dependence that patients have on star ratings or other peer opinions regarding the suitability of medical providers. Subjective word-of-mouth reviews outweigh scientifically based objective measures of clinical quality by a factor of two or three to one. When a friend or neighbor recommends a restaurant or complains about the cleanliness of a hotel, it’s reasonable to assume that they’re equipped to make that judgment – they know whether the food tasted good and they know that an unwashed sink is unacceptable in a hotel room. It’s less clear whether the youth league soccer coach has the wherewithal to assess the clinical capabilities of a surgical team or if my brother-in-law is the best judge of the differences between two cancer centers. Nevertheless, consumers place the same credence in star ratings for doctors or hospitals that they do for hot dog stands and tennis shoes. As it turns out, that may not be such a good idea.

If you compare the star ratings for hospitals on Google and Yelp to the objective quality measures published by CMS, an interesting pattern emerges. There’s a reasonably good correlation between a hospital getting high marks on social media with the same hospital scoring well in patient experience measures reported by CMS. That’s not surprising, since consumers by and large base their ratings of medical providers on their perceived experience – the only component of the encounter that most patients can fully understand. A completely different – and unsettling – relationship exists between social media ratings and such measures as readmission rates or pneumonia mortality. In a troubling twist, the hospitals rated highest on Google and Yelp are actually more likely to be among the worst performers than the best with respect to readmissions or mortality rates. Choosing a medical provider based on social media star ratings is more likely to steer a consumer to the worst performers as measured by objective quality criteria.

A 2015 study conducted by researchers at Princeton and the Sorbonne found that animals that engage in collective behavior eventually become overly dependent on social information and display suboptimal responses to changes in the environment. When environmental changes are large but infrequent, highly conformist individuals take over the population and lead to almost complete failure in information processing. When groups use social information, accuracy improves at first, but then steep declines in performance occur as social influence increases.

In a separate study of zebra finches, a subset of birds was trained to release seeds from a feeder by pecking when a red light was illuminated. New, uninitiated birds were then paired with either trained partners or a second naïve partner. In pairings of untrained partners, both birds learned to release the seed when the light was illuminated. Birds that were paired with an already trained partner became dependent on the partner and never developed the ability to release seeds themselves. They merely pecked randomly, never assimilating the behavioral cues themselves.

The impact of social influences on individual behavior has only intensified as opinions are shared more quickly and aggregated more easily. When the underlying source of social cues is reliable – such as restaurant ratings or hotel cleanliness assessments – there’s evidence to suggest that the initial effect may be positive, though studies would suggest caution over time to avoid becoming unresponsive to a changing environment. When the foundation of social influence is faulty – as is the case when crowdsourcing points patients to the worst providers more often than the best – we find ourselves closer to the situation that worried mom all those years ago …”If your friends all jumped off a bridge, would you jump, too?”

About the author and the Vizient Research Institute™. As executive director of the Vizient Research Institute, Tom Robertson and his team have conducted strategic research on clinical enterprise challenges for 20 years. The groundbreaking work at the Vizient Research Institute drives exceptional member value using a systematic, integrated approach. The investigations quickly uncover practical, tested results that lead to measurable improvement in clinical and economic performance.

Related Materials