Is the person asking to be admitted to an exam the person who is supposed to be taking the exam? How do you prove this person should have access to exam items?

One of the first steps in admitting a person to sit for an exam is checking their identification (ID). Typically, people present a driver license, state issued ID, military ID, or passport for vendor exams, and student IDs for academic exams. The topic of vetting student IDs requires a volume of its own. This post is focused on the ID types used for vendor testing.

The topic of fake IDs became an issue for me many years ago when I was new to the testing world. A test-taker presented me with a passport from a country I knew by name, could come reasonably close to locating on a world map, but had no idea what their passports should look like. In his passport photo, the test-taker was wearing authentic clothing appropriate to that country’s customs. Very few facial features could be seen. His hair was completely covered. He was clean-shaven. The person standing before me was in jean shorts, a polo, full beard, and flip flops. I had little to determine if the passport was authentic and if the person standing before me was the person who was supposed to test. Fortunately, he was able to provide a matching signature, a secondary form of ID, and it was a low-stakes admissions exam for our university where I could verify his enrollment. But the moment caught me off guard and I promised myself I was going to learn about IDs.

It seems that I am not alone in the struggle to authenticate IDs.

Between Operation Varsity Blues and the sudden need to provide at-home testing options for millions worldwide, test security has been thrust into the spotlight over the past few years. Ringers taking tests for others in both live and virtual settings has been reported for large-scale standardized exams and college classes alike. Recently, Derek Newton in The Cheat Sheet reported two high school cheating scandals (Issue 120), one of which was involved impersonation.

Over the last two years of emergency remote instruction, there has been much debate as to where the lines between privacy and security, user authentication and discrimination, convenience and requirements should be drawn. It was important to keep colleges, universities, and certification programs moving forward to provide people opportunities to finish degrees and receive employment credentials. At the same time, it was important to remember the struggle many people had to find private spaces, renew identifications, and have technology compatible with a variety of platforms. Those of us in the testing profession care deeply about both integrity and students' challenges. While academic integrity is the goal of test security, we must also remember that human beings take the tests.

Authenticating test taker identification surfaced as one of many challenges during this time of emergency remote instruction. Colleges and universities struggled with verifying that the person sitting on the other side of the camera was the student enrolled in the course without causing undue stress to the student. Test centers reviewed ID policies looking for ways to increase verification efforts in remote settings. We have always known we do not catch all impersonators or fake IDs at our testing centers, but we now found ourselves wondering if we are doing a decent job of identifying impersonators in remote settings.

Through a grant from the National College Testing Association, Jarret Dyer (College of DuPage) and I have begun investigating the frequency with which live proctors compared to remote proctors accurately detect fraudulent identifications. Using a set of identifications, we are asking proctors to view each ID and determine whether they believe the ID is valid or fake. For proctors participating in a live setting, they may handle the IDs as they would at their testing site. For proctors participating in a remote setting via webcam, they may ask the researcher to turn, tilt, move the ID closer or farther from the camera, or other requests to help determine authenticity. Each proctor is asked to work independently and draw their own conclusions.

This project focuses solely on the identification card itself. At a test site, proctors can use other metrics to help determine test-taker authenticity. Does the photo on the ID match the person presenting the ID? Do the demographics on the ID match the test registration? If the ID says a person is 5’9” and the person standing before you is 5’2” can the difference be explained? Does the signature on the ID match the signature on the sign-in log? Remote proctoring has its own set of checks and balances such as requiring a photo to be submitted by the test-taker prior to testing for matching purposes and keystroke analysis. We chose to start with one variable for this project, hoping our findings will lead to more projects in the future.

One of our goals is to find ways to improve academic integrity efforts for both vendor testing and classroom testing by establishing best practices for ID checking. Correctly identifying test takers is a critical first step. Gone are the days of poorly crafted fake IDs. Modern fakes are plentiful, cheap, and extremely convincing. Most of the sites offer buy-one-get-one-free just in case one gets confiscated. How detectable are these IDs? Can in-person proctors spot fake IDs more frequently than remote proctors? We eagerly await the outcome. If you are planning to attend the NCTA conference in August, please stop by the research room and participate. If you did, do you think you would be able to detect the fakes?