View on GitHub


How the Scientific Establishment Generates Bad Science, and What We Can Do About It

Mutual Admiration Society

The term mutual admiration society is attributed to Thoreau, and was popularized by the 1956 Broadway musical Happy Hunting. It refers to a group of people who mutually respect and promote each other. In the Web 2.0 era, this term introduced for this is groupthink.

Your group will explore how academia is structured to support groupthink, and approaches to fix it. Diversity in academia are based on the protected classes (race, ethnicity, gender, sexual orientation, etc.), but there has been very little progress in terms of intellectual diversity and respect for different points of view.

Many members of our communities have never really interacted with half of the nation. A paper found that at Dartmouth, Princeton, Yale, Penn and Brown, more students came from the top 1 percent of the income scale than from the entire bottom 60 percent. The two highest-growth religious groups – Evangelical Christianity and resurgent Islam – have virtually no representation in elite academia. Fewer than 1% of professors at elite universities are born-again Christians. Furthermore, the faculty hiring excludes anyone without large numbers of publications (e.g. top individuals from industry, government, or otherwise, no matter how good), and most disciplines hire into a very narrow set of sub-fields (for example, computer science operates on a bucket system, where there are no jobs for individuals outside of pre-established sub-fields). Within Harvard’s student body, there was an 80/6 split in favor of Clinton over Trump. Within Harvard’s faculty, anecdotally, the split was close to 100/0.

Hence, despite the focus on diversity, the diversity of ideas and ideologies remains very low in elite academia. Academic fields often become increasingly narrow and ideological with time.


Skim this popular article on spanking.

Now, skim the original paper. This paper is not an example of a particularly bad paper (something we can all find). It is typical of the practice in the academy.

As a side note, most scientific results are published in proprietary journals, hidden from public scrutiny, and only available to self-selected peers. This particular paper could not even be accessed through MIT Libraries. Hiding research undermines our credibility as scientists as well. But I brought a copy, if you’d like to borrow one.

In the academy, established professors:

Research communities are, in effect, self-perpetuating. They are extremely self-perpetuating at the tenure level, where one upset letter writer can tank a tenure case (if a vocal Trump supporter were up for tenure at Harvard, consider the odds that at least one letter writers would dislike them, whether due to explicit or implicit bias). This effect leads to poor faculty selection (it is generally impossible to find jobs in new fields, and old fields stick around well past their usefulness), but we will focus on in this group, we will explore a different aspect: politics and culture.

In hard sciences and engineering, this problem generally just leads to important problems being ignored (in computer science, for example, anything which falls outside of the buckets). In the humanities and social sciences, such communities generally have underlying ideologies, assumptions, and taboos which cannot be questioned.

A paper which supports a conservative point of view will, at the very least, be given much more scrutiny in the peer review process, and an academic with a conservative point of view will, likewise, at the very least, have much more scrutiny when applying for academic positions. Given the competitiveness of the process, the extreme splits we see are unsurprising.

In the case of this study, the community reached the (reasonable, but unproven) consensus that spanking harms children. The study was run with the stated goal of validating this belief: “there is a need for definitive conclusions about the potential consequences of spanking for children,” and the press coverage reflected that goal.

My conclusion from the study is – unsurprisingly – that parents who follow guidelines from organizations such as the American Academy of Pediatrics have better results. Following such guidelines is a good parenting practice, but individual guidelines are hardly scientifically proven to the level shown in the press coverage. Spanking kids is probably a bad idea, but it has nowhere near the support alleged in the University of Texas popular article.

Your problem is to consider:

Think about the key players, incentives, and what one would realistically need to do to shift them.