Yes, let's go back to the issue of what can/should internet social media
companies do with respect to (a) manipulating the user experience
without
their consent, (b) analyzing the data from (i) general activities on the
social
media website and/or (ii) responses made (apparently) to the website's
manipulations, and (c) the larger issue of using "data mining" and
"found"
results, that is, statistically analysis of data collected for one
purpose
but used to answer questions brought post hoc to the data. I recognize
that the last point represent a long-standing "research" tradition in
some
areas (e.g., analysis of census data, public health data, etc.) but
social
media data seems to be a different kettle of fish.
Why re-visit the issue? Well, outside of the general question of
whether
such research is ethical and/or useful/valid, there's a new book coming
out
this week (Sept 9, 2014) with the title "Dataclasm" and its author is
the
president of the OkCupid website, Christian Rudder, the guy who hit the
haters of the Facebook research study upside the head by essentially
saying:
"Wake Up! Websites manipulate people ALL THE TIME! We use your
personal data and everyday activities HOWEVER WE WANT and there
is NOTHING YOU CAN DO ABOUT IT! See your Terms of Service."
There is an article on Rudder and "Dataclasm" in the NY Times and
the quote that they use is:
|"Guess what, everybody," he wrote, "if you use the Internet, you're
|the subject of hundreds of experiments at any given time, on every
site."
The article is available here:
http://www.nytimes.com/2014/09/07/technology/okcupids-unblushing-analyst-of-attraction.html?emc=edit_th_20140907&nl=todaysheadlines&nlid=389166
Consider the following quote from the article:
|"I realized I could use the data to examine taboos like race by
|direct introspection," [sic?] Mr. Rudder writes, describing how he
|tapped into aggregated information about OkCupid members
|to examine online interactions between white men and black women.
|"The data was sitting right there on our servers. It was an
irresistible
|social opportunity."
One can raise the following issue in methods and statistics courses:
What valid conclusions, if any, can be reached through such an analysis?
What role does "self-selection" and other factors play in affecting the
interpretation of the results and determining to whom the results apply?
Continuing the quote:
|This kind of limitless data-mining is possible because most sites,
|OkCupid included, have use policies that give them unfettered rights
|to harness users' details for research and product improvement.
|
|"The industry has so much data and the data are so accessible, they
|can do research on anything they want," says Elizabeth Buchanan,
|the director of the Center for Applied Ethics at the University of
Wisconsin-Stout.
|"But just because we can do this with the data, should we?"
|
|Some companies, concerned about a lack of industrywide research
|ethics standards, are instituting formal internal review processes to
vet
|their human studies.
The issue of whether doing such research is ethical without informed
consent, especially examining how various manipulations affect behavior,
is an ongoing one but there is another angle on this research:
Can data collected for one purpose be used to answer questions that the
data was not designed to answer?
Consider the "research problem" in the following quote from the article:
|Take the section in "Dataclysm" on racism among date seekers.
|Mr. Rudder describes data on heterosexual OkCupid members who,
|as part of using the site, looked at photos and profiles of potential
|dates and rated their attractiveness on a scale of one to five. People
|of both genders routinely preferred potential dates of their own race
|or ethnicity.
|
|As a group, for instance, Latino men rated Latinas as 13 percent
|more attractive than the average for the site, while they rated
|African-American women 25 percent less attractive. In fact, Mr. Rudder
|reports, black women on the site receive about 25 percent fewer first
|messages than other women do. For Mr. Rudder, these numbers
|unequivocally tell a story of racism.
|
|"The data we see in this chapter shows racism isn't a problem of
|outliers," he writes. "It is pervasive."
Hmm, it appears that men of an ethnic group rate women
of that ethnic as more attractive than women of other ethnic
groups -- but more results are needed to make such a conclusion.
And what about Asian women? Don't all men find Asian women
SUPER HOT! (I believe that there is the opinion that Asian women
are over-sexualized in Western media, thus making them more
attractive to all men, or so I'm told ;-). What questions are not
asked? Are all results provided or are they "cherry picked" to
provide the results most "relevant" to one's "research" question
(i.e., confirmation bias)? We've seen how well this works for Big
Pharma.
Continuing...
|Interpreting a de facto beauty contest as a national referendum on
|racism, however, seems problematic. Just because some people are
|more attracted to members of their own race doesn't automatically
|make them racist; a whole constellation of cultural, familial, societal
|and biological inclinations can contribute to our notions of the ideal
mate.
Under the assumption that such a result has good external validity and
applies to other social media website, I wonder if any Tipster can
answer
the following question:
Are you an implicit racist if you participate on social media website?
If you respond unconsciously in racist ways (e.g., finding
African-American
women less attractive than European-American women), does this type
of research provide evidence of that racism?
Secondly, if you think you're not a racist, what are doing on social
media? ;-)
For those who are interested, Rudder's book can be obtained on Amazon;
see:
http://www.amazon.com/Dataclysm-When-Think-Ones-Looking-ebook/dp/B00J1IQUX8/ref=sr_1_sc_1?s=books&ie=UTF8&qid=1410096926&sr=1-1-spell&keywords=dataclasm
See the quotes/reviews (e.g., Dan Ariely's blurb).
See you later bigots, ;-) <- For the humor challenged. ;-) <- Well, you
know....
-Mike Palij
New York University
m...@nyu.edu
---
You are currently subscribed to tips as: arch...@jab.org.
To unsubscribe click here:
http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=38259
or send a blank email to
leave-38259-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
