Since the shift to online teaching, assessment has become one of the most challenging areas for EFL teachers. In my context of teaching adults online in Naples, I realized that traditional tests, when moved online, often become neither fair nor accurate. An assessment is fair when all learners have an equal opportunity to show their ability, and it is accurate, or valid and reliable, when it actually measures what it claims to measure. This reflection explores how we can ensure both in online EFL settings.1. Ensuring Accuracy: Validity and ReliabilityAn accurate assessment must have validity, meaning it tests what we taught and what learners need to do with the language. A common mistake I made was using online grammar quizzes that tested only recognition, e.g., multiple-choice: "I ___ to Rome yesterday: a) go b) went". While easy to mark online, this has low validity if my learning objective was for learners to be able to tell a story about their past.To ensure validity online, I now align my assessment with my learning outcomes and use a wider range of task types. For example:To assess writing, I don't just use gap-fills. I use Google Docs where learners write a short email. I assess it with a clear rubric that looks at task achievement, grammar range, and vocabulary, not just accuracy.To assess speaking, I don't use written tests. I use Flip or voice notes on WhatsApp where learners record a 1-minute response to a question. This is a more valid test of speaking than a multiple-choice test.To assess listening, I use Edpuzzle with authentic videos and ask comprehension questions, not just discrete grammar questions.Reliability means the test gives consistent results. Online, reliability is threatened by technical issues and unclear instructions. To improve reliability, I now provide clear models, use rubrics for all productive skills, and standardize conditions. For example, before a speaking assessment, I send a video showing exactly what a good answer looks like at B1 level, so students know the expectation.2. Ensuring Fairness: Accessibility and Academic IntegrityFairness online has two main dimensions.a) Accessibility and Inclusivity: Not all learners have the same internet connection, device, or quiet space. An assessment is unfair if it advantages learners with better technology. In my classes, some students join from their phone while commuting. To make it fair, I offer flexibility. For a listening test, I allow two attempts and allow them to download the audio in advance. For timed quizzes, I add 25% extra time for all students, not just those who request it, following Universal Design principles. I also provide instructions in three channels: oral explanation, written instructions in the chat, and a short Loom video. This ensures that a learner who missed an oral instruction due to a connection issue can still access it.b) Academic Integrity and Cheating: This is the biggest fear for teachers online. If students use Google Translate or AI, the assessment is no longer measuring their ability. Punitive approaches like lockdown browsers create anxiety and are not suitable for adult learners.Instead, I use assessment design to make cheating difficult and irrelevant:Design tasks that are personalized: Instead of "Write about your last holiday," I ask "Write an email to your partner describing what you did last weekend. Use two of the past time expressions we learned." It is hard to translate something so personal and specific.Use low-stakes, formative assessment frequently instead of one high-stakes test: I use weekly 5-question formative quizzes on Quizizz for vocabulary retrieval. Because they are low-stakes, there is less pressure to cheat, and I get more accurate data about learning over time.Use synchronous, live assessment for productive skills: For final speaking assessments, I use 5-minute one-to-one live conversations in breakout rooms, where I can ask follow-up questions. You cannot fake a live conversation with Translate.3. The Role of Feedback and Learner InvolvementFinally, a fair and accurate assessment must include transparent feedback and involve learners. An assessment that only gives a score of 7/10 without explaining why is not helpful. Online, I use screencast feedback: I record my screen for 2 minutes while I correct their writing and explain their errors orally. Learners find this more personal and clear than written comments.I also involve learners through self and peer assessment. Before submitting a writing task, students use a checklist: "Did I use past simple and past continuous? Did I check my spelling?" This develops autonomy and makes assessment a learning tool, not just a judgment.ConclusionEnsuring fair and accurate online assessments is not about finding the perfect anti-cheating software. It is about good assessment design. Accuracy comes from ensuring our tasks are valid and reliable, meaning they truly test communicative ability with clear rubrics and models. Fairness comes from designing for accessibility, offering flexibility, personalizing tasks to make cheating difficult, and using assessment for learning through formative, low-stakes checks and transparent feedback. When we design assessments with these principles, online assessment can be not only fairer and more accurate than traditional paper tests, but also a much more motivating and authentic experience for our learners.
