The vast majority of IT certification exams require that candidates be tested in a controlled environment that is monitored by a testing proctor. This is meant in part to ensure that every candidate taking the exam will do so in an environment as similar to all other exam candidates as possible. Obviously, a significant portion of the rationale is to eliminate (or at least reduce) the number of candidates who cheat on the exam.
That said, there are a handful of certification exams that can be taken in an unproctored environment. Usually it is possible to take these via any computer with an internet connection. A significant number of professionals feel that any certification program that includes one or more of these exams is considerably less valuable. Their logic is that anything that makes it easier for candidates to cheat on an exam also makes it more difficult to trust the competence of people who have passed a certification track that includes one.
It is hard to argue with that thinking. If I were going to write about whether or not it is easier to cheat when taking an unproctored exam versus a proctored one, this would be a very short article. The question I am exploring, however, is what value we should attach to such tests and, by extension, whether they have a valid place in a certification program.
At one time, I most likely would have dismissed the possibility that these exams held any value. However, a number of years ago, I was looking to change jobs. A recruiter had a position as a PL/SQL developer that seemed interesting. The company asked her to have me take an exam to gauge my knowledge of PL/SQL before even considering my resume. The recruiter sent me a URL for an online exam.
After reading the instructions, I was floored. The instructions specifically noted that it was legal to use Oracle documentation, personal notes and web searches while taking the exam. I wondered how an open book exam like that could possibly determine anything about my skills as a PL/SQL developer. As I took the exam, it became clear that the ‘open book’ aspect was not quite the free pass it seemed. Every question had an individual time limit. This was years ago and I do not recall the exact amount, but it was under two minutes. There is only so much research that can be performed with that little time.
I received a high percentile score on the exam and the company was interested in me. Ultimately, I was not interested in the company. The test, however, I found intriguing. In fact I told my wife afterward that it was possibly one of the most realistic exams for testing development capabilities that I had ever taken. The reason I say that is because what the exam tested was a microcosm of how I do my job as a developer.
I write enough code that most of the common syntax and functions that I need are in rote memory. However, if I require a function that I seldom use and do not recall the syntax, I look it up in the manual… or more commonly these days, I use Google to pull up an example. If I have a question about coding and I can find an answer in less than two minutes, no one is going to care that the information came from a manual or a website rather than from my memory.
Part of being an experienced developer is knowing how to find the answers to questions. While the exam I took was not part of a certification program, it is an example of an exam that did an excellent job of minimizing the opportunities for cheating.
The Oracle certification program has a handful of exams that can be taken unproctored. The most common of these are the SQL Fundamentals exams: 1Z0-051 and 1Z0-061. These exams are also the most controversial as an SQL exam requirement is a cornerstone of both the Oracle DBA and developer certification tracks. While Oracle exams do not have a timer for each question, the overall time allotted for these two exams gives candidates an average of a bit over a minute-and-a-half per question.
Since some questions on any multiple-choice exam are obvious if a candidate has prepared for it, candidates will be able to spend more than the average time allotted on harder questions. However, as with the PL/SQL exam I took, there is not a huge amount of time to look through various resources for answers.
On Oracle’s certification forum, a candidate recently shared his experiences with taking (and failing) the 1Z0-051 exam. He had assumed that at the end of the exam there would be plenty of time remaining to go back through answers he was unsure of. Instead he barely finished the exam in time and was not able to go back. Even if he had more time, he noted that most of the questions did not involve facts about SQL but rather contained SQL statements as part of either the question or the answer. This is not something that you can readily answer with either the manual or with Google. As with the PL/SQL exam I took, the opportunity for using manuals and other resources is minimized.
The biggest problem with unproctored exams that I see would be substitution. Did Fred really take the exam or was it his brother Ed, or his friend Ted? I do not know how Oracle verifies the identity of the exam taker on unproctored exams. What I do know is that the SQL Fundamentals exam is often the first exam taken by Oracle certification candidates (and the easiest). It does not confer a certification by itself, so the candidate must take at least one more exam that will be proctored.
Even if Fred gets someone else to take his SQL Fundamentals exam, the next Oracle exam he takes is liable to come as a rude awakening. I cannot say if this is a widespread abuse, but I do think that any candidates that have a ringer take the exam for them are doing themselves more harm than good.
To answer my own question, I believe that so long as a professional certification path is not composed solely of unproctored exams, they can be used without significantly reducing its value in the marketplace