Computational Genomics Group
  • Home
  • Research
  • Publications
  • Teaching
  • Blog
  • Group Members
  • News
  • Computational Biology Book
  • Data Analysis with R Book
  • CG2 github
  • Fiction

Bias, Prejudice and a simple "litmus test" to detect them

11/23/2018

0 Comments

 
As part of my (quite loaded) teaching schedule I get to give introductory lectures on statistics and experimental planning to undergraduate students (mostly freshmen). One of the concepts I have the greatest difficulty in explaining is that of "bias", which in statistics is the difference between an estimator's value and the actual value of the parameter that is being estimated. As it is understandably hard to explain a complex concept to freshmen, who are increasingly becoming more and more mathematically illiterate, through equally complex concepts such as "estimators" and "expected values", I often have to take less rigorous approaches. One such is to resort to more mundane explanations of the terms. According to wikipedia, bias is "a disproportionate weight in favour of or against a certain person, thing or group against another", a definition that is much easier for students to grasp as it resonates with the more commonplace notion of prejudice. In fact, most of the examples I am using to explain bias, are very "non-statistical" ranging from my old favourite story on how sharks prefer eating men to women to everyday issues such as the reporting of crimes committed by immigrants in mainstream media.

These are, of course, issues not to be taken lightly as biased ways to report, write and discuss about events are becoming ever more frequent. In my brief spell as aspiring reporter (back in the day) I was surprised to realize how easy it is to let prejudice infiltrate your reporting (Noam Chomsky and Edward Hermann, devote a whole chapter of their seminal "Manufacturing Consent" on this topic).  In our times, however, of a general return to conservatism and ever expanding bigotry, what is more striking is the way we fail to perceive prejudice and bias in every day life. It seems as if we are becoming blind to even the most outright bias in expressed opinions and this, of course, is not at all helpful for my students (as both students and citizens).
To this end I have been thinking on more straightforward ways to detect bias and I have recently co
me up with some ideas, after exchanging opinions with friends on facebook (yes, it is possible). The examples I am be posting below, were inspired by three very different topics that came up on my facebook timeline on the same day.

The first was brought to my attention by an old school mate and it had to do with a somewhat famous greek actress defending her decision not to have kids after being repeatedly asked why she wouldn't in various interviews. My friend, herself also married and happily childless, was infuriated by the way the actress (Katerina Lechou) had to defend "a woman's right" to not have kids. What immediatelly stroke me was the fact that we were discussing this as a "woman's right" and went on to ask why are men never asked this question. This is a very straightforward way to realize that the question "why won't you have kids?" is only part of the problem as long as it is only addressed to women. My friend and many other women were rightfully offended by the content of the question but failed to realize it was also greatly prejudiced since it implied that having children is either something to be decided by women alone or something that men should not really care about.

The second story had to do with a British food editor being forced to resign after making an admittedly bad joke about vegans in an email. Even though, I respectfully understand that some groups of people may be more sensitive to comments than others, you will have to agree with me that there is absolutely no chance that William Sitwell would have quitted his job, had he asked in his message that "all meat eaters be killed". Here again, we see how easy it is to spot bias simply by substituting the object of the statement with its conceptual counterpart (here "vegans" become "meat eaters"). This is the essence of what I call the "bias touchstone". Invert the argument and see if it makes sense or not. If it still sounds reasonable then bias is not so likely. If, however, it doesn't, prejudice may be implied. In this case, the prejudice is a positive one, aimed at "protecting" the sensitivity of vegans. It remains a prejudice nonetheless.

The last example is a bit more personal, as it has to do with the negative evaluation of a grant application, which I received yesterday. The sole reviewer of my proposal had made an honest effort to read it and had a few plausible arguments for rejecting it, especially given the very low acceptance rate of the call. What was however alarming was his/her blunt statement regarding our work. In the part of his/her assessment, where rejection was being justified, he/she had no reservation in writing that "This is a purely computational biology project". The way it read, made it sound almost offensive to be proposing a computational work in a Life Sciences panel. Besides the fact that I have been working in Biology and Biomedical institutes for my entire adult life, I could not resist applying the "bias touchstone" to the statement. The same reviewer has surely never used a comment like "this is a purely developmental/molecular/cellular biology project" as a justification for rejection. (Even though I can think of other subdisciplines such as structural or evolutionary biology that may have been targeted in a similar manner). A simple substitution of the agent of bias in the statement quickly reveals the bias itself.

As double standards are increasingly becoming the norm in may forms of public discourse, this very simple idea can be easily extended as a first assessment for any sort of statement. In science, it can also serve as a rough evaluation of the originality of a given finding. Take any sentence like "X is found to interact with/regulate/inhibit Y" and form its negation: "
X is found NOT to interact with/regulate/inhibit Y". If it still sounds plausible, then the finding is quite interesting. In what regards X and Y anything could be going on, but now -thanks to this work- we know that it's X that regulates Y. If, however, the negative statement sounds quite improbable then the original finding is suddenly not so original. In this case, the goal is not to spot bias but to confirm an imbalance between two -initially- equivalent possibilities.

​At this point you may have realized that this mental experiment is silently conducted all the time, especially by editors and reviewers of scientific journals when assessing the possible "impact" of a scientific finding. It forms the basis of a rather special kind of bias, called "confirmation bias". 
​But this is a story for another post.  
0 Comments

    RSS Feed

    It's all about...

    Bioinformatics and computational biology with a focus on chromatin and genome architecture, plus a little bit of football and occasional aspects of  University education.

    Archives

    April 2021
    December 2020
    March 2020
    November 2018
    September 2017
    April 2017
    March 2017
    December 2016
    November 2016
    February 2016
    May 2015
    November 2014
    September 2014
    July 2014
    February 2014
    November 2013
    October 2013

    Categories

    All
    Academic Life
    Bioinformatics
    ChIPSeq
    ChIPSeq Bias
    Cpg Islands
    Data Analysis
    Exons
    Football
    Footballomics
    Gene Regulation
    Genetic Diseases
    Genome Architecture
    Genome Structure
    Inflammation
    Journalism
    Math Illiteracy
    NGS
    Nucleosome Positioning
    Nucleotide Composition
    Nucleotide Skews
    Promoters
    R
    Splicing
    Statistics
    Systems Biology
    Tnf
    Transcriptome
    Variation
    Whole Exome

Powered by Create your own unique website with customizable templates.