Abstract
As psychology ethics begins to become more standardized and formalized globally (e.g., CitationGauthier, 2007) there are still educational, political, and psychological areas that require significant discussion. For example, test security has become a global issue, as psychological tests and even college entrance and graduate school admission tests have found their way online.
Notes
1To avoid confusion, the term technology-based is used throughout this article to refer to these various types of tests as a group. When referring to them individually, the specific term is used.
2In this article a “high-stakes” test refers to any exam that produces a score that has important consequences for the examinee and/or other stakeholders. In education, these include course midterms and final exams, U.S. state assessments, admissions tests to college or graduate programs, and others. Outside of education, they include certification and licensure exams, pre-employment screening assessments and employment tests, and clinical psychological exams administered for a variety of purposes including employment, educational placement, and legal. In contrast, low-stakes tests have absolutely no important consequence.
3Although it has a more specific definition, for the purposes of this article, “unproctored” testing refers to Internet (or online) testing that has very little or no security, including no authentication of the test taker, and no monitoring of test taker behavior.
4Online Proctoring refers to several technologies meant to discourage cheating and to immediately detect and deal with it when it occurs. Online Proctoring can be divided into three components: (a) strong authentication through Keystroke Analytics, (b) machine lockdown and automated monitoring of attempts to breach the system and cheat, and (c) human monitoring of test taker behavior through individual Webcam viewing.
5Of course, this hesitancy may also be closely related to laxity in putting stringent security measures in place: If you catch cheaters effectively you then have the unpleasant and risky task of punishment.
6In a personal communication, Dennis Maynes of Caveon Test Security described to the author a new forensics analysis method for a certification program that had exams composed of both new items and older, exposed items. He calculated two scores for each test taker and a “difference” statistic that identified those candidates who had taken the test with pre-knowledge of the exam questions. Those candidates would have significantly higher scores for the older items. With such information the program could reliably identify the truly competent candidates.