Thursday, May 7, 2015

The Rise of University Data Mining and Analysis Oligarchies--From Transparency to Confidentiality Regimes in University Operations and the Issue of Salary Information

(Pix (c) Larry Catá Backer 2015)

I have been considering the management of data as a source of power in the relations between university faculty and administration.  In particular, I have been suggesting the ways that  both the harvesting of data on salaries (what is going to be gathered up  as data) and its presentation (the extraction of meaning from the data), can be a substantially manipulative affair, best controlled by those with the power to generate and analyze data (see here, here and here). 

While my focus has been on the quality of the data and the manipulative potential of its presentation, especially in thew way in which it might be used to both undervalue labor and to justify the allocation of productivity gains away from its producers (effectively resulting in increasing wage-productivity gaps that result in effective wage decreases per unit of productivity). But perhaps all is fair in the battlefield of the wage labor markets in which universities operate. Yet when the field of negotiation on which these contests are played out are no longer level--when they are substantially tilted in favor of those with a disproportionate amount of market power, and deliberately so for what appears an unfair advantage--one wonders about the application of general principles of ethics and fairness play within an institution that might consciously operate in that environment. 

That is perhaps what one might be led to think when one considers the very recent changes in the transparency rules at play at Penn State with respect to salary data--data once freely made available to faculty (along with administrative interpretation, quite persuasive traditionally, thereof).  Where for years salary data for the faculty were made available more or less openly, and discussed openly, suddenly the data disappears. The disappearance is made more troublesome becomes it comes so soon after former President Rodney Erickson, in January 2012, moved publicly to expand transparency specifically by making salary data available.
In an apparent break with his predecessor, Penn State President Rodney Erickson said Thursday he would have "no objection" to releasing university employees' salary information to the public.

It's a move "that many individuals want," he said.

"I think there are a lot of other publicly supported universities, at least in part, that have opened up that salary data," Erickson said in a town-hall forum outside Philadelphia. "And I think that's something that we could live with, too."

Spanier had argued that the release of individual salary specifics could affect Penn State's competitiveness and "might influence certain aspects of attracting and retaining faculty," as Erickson recalled Thursday.
But Erickson, in his remarks, appeared to be less concerned about opening up the information.
"I have indicated that I really have no objection to providing salary information, which is something that many individuals want," he said. " ... We certainly would make contracts available and things of that nature that individuals may want to see that are of particular interest." (Erickson: 'No Objection' to Releasing Penn State Employees' Salary Information, State, Jan. 13, 2012)

But we would be wrong.  Reversing the direction of policy under President Erickson, it now appears that the university will deny access to salary data to its faculty, except in the way that one might, in another age, have had access to information from works on the Index Librorum Prohibitorum

And that, it seems, is precisely what has occurred at Penn State (as in other places; Penn State is not special, though it is a "type" with significant public policy ramifications) when, once modestly open to some transparency in salary data reporting, it is now more vigorously protecting the institutional value of data about itself (and its faculty) --from its faculty. The modern university no longer has the bother of seeking to protect the value and privacy of its data--even or especially against its own stakeholders--on its own.  Rather by forming or joining groups dedicated to the harvesting and protection of data (as well as the generation of institution friendly analysis for inward and outward consumption) to the university as an institution can now respond to its stakeholders who complain about transparency and say--"not my fault; I am bound by the rules of the institution to which I belong."

That this is possible now appears to be a function of the rise of specialized institutions among powerful trade associations the purpose of which is to foster efficiency through collective action.  Among these are the Association of American Universities and its data mining apparatus, housed within the Association of American Universities Data Exchange (AAUDE), which styles itself "a public service organization whose purpose is to improve the quality and usability of information about higher education. Our membership is comprised of AAU institutions that support this purpose and participate in the exchange of data/information to support decision-making at their institution." (AAUDE Welcome).  "Institutional Research Officers at AAU universities are organized through the AAU Data Exchange (AAUDE) to collect and share data on a wide range of university activities. AAU and AAUDE collaborate on a number of issues to inform AAU policy positions and reports." (AAU Constituent Groups).

While its service activities are no doubt to be welcome, and their willingness to pool their data gathering and analytical prowess laudable, it seems that this effort is undertaken for the benefit of a fairly narrow band of stakeholders--the administrators of the universities who band together to form and operate this enterprise and for the purposes of the advancement of the objectives of the AAU trade group.  Consider in this light the emphasis on confidentiality, one of the few items that is not confidential about its operations: 

Confidentiality & Data Sharing Policies by rcarr | Thu, 07/02/2009 - 20:48

AAUDE data are subject to data sharing and confidentiality rules as defined in the documents below. The templates used for data sharing are not subject to these rules; they can be shared with other institutions or organizations.
Data Sharing Guidelines and Confidentiality Rules
Warehouse Signature and Approval Form for AAUDE users
Policies related to Confidentiality, Data Sharing and Data Use for Non-AAUDE Users
Statement on Data Integrity

Item-specific Rules and Examples (members only)

Faculty Salary by CIP
Response-level Surveys

For everyone else--nothing.  Or better, a little less than nothing. An institution is free to distribute its own data, as long as it does not include any comparative data or analysis grounded on information derived from any source other than its own.  That, in some sense is fair.  But where an institution necessarily marbles comparative information and metrics across its own data, the result is the application of the confidentiality rules of the organization even as to its own data. Thus, what appears in theory to be possible, is in reality impossible from a practical perspective. But since the administrations of the member institutions have free access to all the information generated, information synergies are preserved, and advantage leveraged against stakeholders whose ability to access information is substantially hobbled.  

Ethical?  The right thing to do?  These are useful first questions that might guide a conversation among university administrators, faculty and the structures of the collective institutions that universities create that might inadvertently no doubt serve as a shield against transparency. These questions are especially useful in the context of discussion about conditions of markets and market integrity.  In that context, it is interesting to note that the rise of these trade organizations, like the rise of international organizations in the public sphere, can serve as a useful means of aggregating information  and developing common cultures.  Sadly, they can also be used to evade accountability and to deflect obligation.  Both are no doubt possible in the context of salary information.  On the one hand, the aggregation of salary data among AAU institutions is useful (and not just for administrators, though the ethics of limiting data just to them has ethical dimensions as yet unexplored).  On the other, using these organizations as the reason or excuse to limit distribution of data poses certain ethical issues.  It would be odd, indeed, for an institution to deny its members information about those very members because of rules that it helped promulgate in organizations which it might have been inclined to join to effect just that result.  Surely, no  one is suggesting that this is the case; yet optics are important, and institutions that continue to enhance confidentiality rules at the multi-institutional levels, while arguing that but for those rules it would be happy to engage in broader transparency, might invite unfortunate inferences about the ethics of administrators and the institutions on whose behalf they are supposed to labor.  

In a Republic, such as ours, founded on notions of fair and robust markets, in which substantial public and private resources are devoted to the maintenance of level playing fields to produce the sort of competition that enhances efficiency and greater societal welfare, the sort of market inhibiting actions around confidentiality of salary data can do society, and the university, little long term good.  The sorts of discussions about salary, as part of a larger discussion about the budget and financial models of universities, so necessary as universities confront challenges to their productivity and output models, are unnecessarily inhibited by the sorts of institutional obstructions that university administrative organizations have begun building around data.  Moreover, universities, like the rest of our society, has been moving toward more vigorous regimes of transparency.  Habits of using information as commodities--as a weapon to give university "employers" and "edge" over their "employees" is at best troubling and at worst an expression of contempt the value of which will not likely exceed its costs over the long term. 

In the context of salary reporting, at a minimum that commitment to transparency ought to lead university administraitons to

1.  Develop, distribute and adhere to a policy (and top down cultures) that affirm ethical conduct in administrative operation, including avoidance of using administrative devices to detach information from institution or make it more difficult for the institution to be transparent about its own data.

2.  Develop robust mechanisms for compliance with policy, including ,omnitoring, assessment and mitigaitonb.

3.  Commit to access of salary data produced by the university or related to university personnel in a complete and useful manner. That, certainly is in line with the latest confidentiality rules of the AAUDE, and to that end avoid enmeshing university data irretrievably in data otherwise subject to non-disclosure.

4.  Report on efforts at rule making by multi-institutional organs, like AAUDE, and provide the institutional representatives of the faculty a meaningful opportunity to comment.

5.  Be willing to include university data on the salaries of administrators and staff in salary reports.

No comments:

Post a Comment