2015,
> la i
Below are my decision and comments about this application. i
seseeeecs '
After much deliberation, [am affaid that I can not approve this application, The concerns that
Thave about the research can not be restfied and I do not feel comforsable approving a
project that includes aspects that are worrisome tome.
Lappreciate the efforts that the researchers went to address my first set of comments,
However, [ am not convinced by their argument about consent and participants being made
aware of their right to withdraw from a research project at anytime. I am sympathetic to the
general argument that some data are available for research purposes without consent having
been given by the people in the dataset. For example, government date, schools data, and so
forth. However, the key difference from that sort of publicly available data and this project is
that there is a gatekeeper who reviews a proposal/request for the data and does the
anonymising the data, That gatekeeper plays an incredibly important role in safeguarding and
securing the data. The application and response have not clearly identified such @ gatekeeper.
‘And, the response seems to put the onus of the gatekeeping on the research team, There are
limits to the extent that one can use data for research purposes without direct consent, For
example, if one wants to compare data from one government agency with another, then more
strict procedures are required. And, if one wants to compaze data from a research
intervention/study to well as existing data, then one needs to get consent directly from those
involved to do that,
My concems here is that consent has been thought to be granted in a way that falls far below
the ethical expectations of the university. Clear information sheets and consent forms that
allow the participant to agree to specific aspects of research, 9s well as clear guidance about
participants’ right to withdraw are an important aspect of the consent process not followed for
the participants and no opportunity relsted to declining consent is given to the Facebook
‘friends’. [ worry about this because I do hot see olear safeguarding for vulnerable groups
(eg, children, people with disabilities, ets.), This concer js magnified because the
information given i this study could involve criminal behaviour (e.g., underage adults liking
alcohol pictures or illegal drug references). :
‘The researchers have made @ case thet Facebook allows its users to set their privecy settings
to avoid being part of the study, The researchers seem to beclaiming that understanding these
settings is clear to all users. It is aot, Even as well-informed, long-time user of Facebook, I
find the setting confusing and I know thet many othersdo as well. Facebook’s privacy policy
is not sufficient protection to address my concerns
1 I c's Sse th sven
approved an application. [ have taken a long time fo return-my decision and comments
because I do not take this decision dightly. T/have read the paper published last year in the
Proceedings of the National Academy of Soiences (Kranier, Guillory, Hancock, 2014), that
really kicked off discussions about using Racebook data forresearch. In response to that
controversy, the editor of that journal expressed a goncernzelated to Facebook's consent
procedures. The editor points to:Cornell University’s ethical review of the project. Comell
confirmed that it did not review the projeet because the researchers were ‘not directlyResponse to the appeal from] provided on the request of
the Chair.
{ can't make the meeting but | am inclined to support the original decision of the psychology
committee. 4
Members might care to look at
httos:/Avww.a fault Fallacy L.ndt
for some recent work on real US consumer attitudes to the collection of consumer data for sale to,
marketers. One simply cannot assume that commercial databases are kosher; one must look at the
details.
‘Anonymisation can be harder than they seem to think, and data from which medical facts about
identifiable individuals can be inferred must be treated as sensitive.
It's also somewhat ridiculous to claim that Facebook friends aren't people, when they clearly are. It
also not the case that material shared with lends "published" lagree wth
that Fa + consented, nor had any opportunity to withdraw from the study. In
oe lll csiock share kes tects ease; you eh ep
catching what your friends have. If there are bad apps, or malware, or even dodgy marketing offers,
‘they get passed along through friendship networks. An ethical approach to using social network data
hhas to take account of this. You can't consider consent to be contagious.
It's probably Ok to use social networksfor subject reéruitment, as we've been doing that for decades
‘anyway (it's what snowball sampling is about). But once you have persuaded someone to run your
Facebook app, you can really only colléct that subject's data. What his or her friends have disclosed
Is by default disclosed to “friends” only, that is, with an expectation of confidence.
Itis perfectly true that Facebook is rather deceptive on this and creates the appearance of a cosy
and confidential peer group enviroment, as a means of gulling users into disclosing private
information that they then sell to advertisers, but this doesn't make it right for an ethical researcher
to follow their lead
In short, the committee got it broadly right. That at least is my take on the matter.
08 July 2015,