Lucy,
This message intends to clarify the issues you raise about confidentiality and anonymity in the Non-Instructional Faculty (NIF) Mail and Telephone Surveys.
I asked Steve Johnson to reply to David Frank's forward of your message because I had to teach before I could make a prompt reply, because you and he know each other, and because he spoke with you previously on the same subject. But Steve inadvertently conflated the mail and telephone survey issues in his reply to your issues about the NIF *mail* survey. Since I, not Steve, have had the managerial role in these surveys, these things can get confused. Thus, I will clarify. The "OSRL Interviewer Training Manual" has a Glossary with the following definitions:
The front page of the mail questionnaire says: "Feel free to skip any questions you find personal or sensitive. However, this survey is *anonymous*; no one will ever be able to connect your name or your department with your answers." The back page includes the following: "Your comments will be transcribed anonymously and taken into account."
These statements are all true. OSRL stands by these statements to such an extent that Steve and I have advised our staff that if a court of law subpoenaed any *confidential* survey data OSRL collected, we would go to jail if need be to protect our research subjects. OSRL has conducted numerous sensitive surveys (e.g., concerning sex, drugs, and lawbreaking) that were confidential, not anonymous. We make sure clients know in advance that they will never receive respondent-specific data. We have turned down the few studies for which clients wanted such data. (BTW - such policies and procedures are typical in academic survey research labs and in all but the sleaziest private market research firms.)
The following will give you an idea of OSRL's efforts to prevent respondent identification: when Banner Tables of cross-tabulated data result in cell sizes of 1 or 2, we delete those columns from our reports if there was a chance an unknown individual could be identified. (This has occurred in data for small geographic regions when race/ethnic minority status was cross-tabulated with other demographic data.)
You claim that the items in the NIF mail questionnaire's section D could identify you. Because OSRL *only* presents data in aggregate form, however, this could not occur. A less conscientious organization might allow some devious soul to design extremely fine cross-tabulations that report cell sizes of one, but, as discussed above, OSRL does not. I invite you to peruse results of the Instructional Faculty (IF) Mail Survey on OSRL's WWW site to get an idea of what I mean by "aggregate" results: the report is at http://darkwing.uoregon.edu/~osrl/facatt00/facattmail/facmail.html and associated banner tables at http://darkwing.uoregon.edu/~osrl/facatt00/facattmail/tblIndex.htm You will find some cell sizes of 1, but not cross-classified in such a way that anyone is identifiable.
If you are concerned about the ability of coders to identify you (i.e., persons who transfer the questionnaire data to the computer), I would be happy to share with you a copy of the "Pledge of Confidentiality" and "Statement of Professional Ethics" that all OSRL employees must sign. You may also wish to examine OSRL's "Respondent Bill of Rights" (see http://darkwing.uoregon.edu/~osrl/r_rights.html). Coders see so many thousands of surveys in such rapid fire, that they cannot do their job and sustain an interest in respondent identity. (How many urologists or gynecologists can regard the genitalia of a random opposite-sex person as sexually interesting after seeing some many thousands? OSRL coders and interviewers have similar professional exposure and experiences with regard to surveys.)
If you are concerned about OSRL researchers being able to take one specific person's interview or questionnaire data and bring up UO's Banner System with all its personnel data, well - none of the researchers have access to it! Years ago, I accidentally tried to sign on, thinking it was something else, and was summarily warned and booted off!
The demographic questions in Section D of the mail questionnaire and at the end of the telephone survey are intended to answer such questions as the following. How does job satisfaction vary with years of service -- are OAs employed at UO for 30 years more dissatisfied than those who have been here 2 years? Do women have different sources of dissatisfaction than men? Do parents of minor children have lower compensation satisfaction than those without children? Do minorities (race, ethnic, sexual, etc.) feel differentially treated than non-minorities? Do OAs in the higher-paying units on campus have higher satisfaction than those in the lower-paying units? If OSRL separated demographic data from substantive data, as you suggest, OSRL could NOT answer these questions. Moreover, OSRL's demographically-based analysis of the IF surveys were precisely what several faculty praised (see http://darkwing.uoregon.edu/~osrl/facatt00/UOFARE.htm).
While you may know "a number of people who either will not return the survey, or who will self-edit their responses...", the people you know may not be representative. So far the response rate to the NIF mail survey is about the same as the IF, and the response rate to the NIF phone survey is significantly higher.
As for self-editing, PLEASE do so. That's what we ask you to do in the survey instructions ("feel free to skip any question..."). This is the best way to assuage your confidentiality concerns and simultaneously provide UO with the information it needs to improve OAs' and ORs' jobs. A couple other OAs and ORs have contacted OSRL about these issues and we gave them the same advice - follow the survey instructions. (It should be clear by now that we have no way of knowing if they actually did so.)
ALL surveys are potentially biased -in more ways than you mention and in ways that are virtually immeasurable. Most surveys also are superficial, not getting at the depth that, say, in-depth interviews or focus groups can (but those offer even less privacy protection). For such reasons, I highly recommend that survey results be taken with more than a grain of salt - they offer indicators, not answers. Fortunately, UO has persons on campus with decades of experience in both survey science and craft. Only such experience allows us to anticipate and try to correct surveys' many potential flaws.
In conclusion, the NIF data collection instruments are not "seriously flawed." Invective aside, you raise many potential survey pitfalls that OSRL anticipates routinely. However, if such accusations are circulated to potential respondents while a survey is in progress, they could seriously undermine its response rate and thereby prevent the institution from gathering the information it needs to improve employee morale and compensation. (For low response rates typically result in biased samples). Luckily, the NIF telephone survey response rate is already robust and the mail survey is about what we expected.
Sincerely,
Patty
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Patricia A. Gwartney, Ph.D. Professor
Founding Director Department of Sociology Oregon Survey Research Laboratory
University of Oregon University of Oregon Eugene OR 97403-1291 Eugene
OR 97403-2545
telephone: 541-346-5007 WWW: fax: 541-346-5026 pattygg@oregon.uoregon.edu
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<