ENADE Ciência da Computação common mistakes
ENADE Ciência da Computação · Understand the traps and how to avoid them.
On objective questions that present a computing scenario (a code snippet, an algorithm description, or a system requirement), the candidate answers from a memorized definition of the named concept instead of tracing what the described artifact actually does.
ENADE items often wrap a familiar term (e.g., 'recursão', 'escalonamento', 'normalização') inside a situation whose correct answer depends on the specific data or constraints given; candidates pattern-match the buzzword and stop reading.
What to do instead
Before choosing, restate the scenario in one sentence in your own words, then verify the option against that restatement. For code or algorithms, simulate one small input by hand until the output is fixed.
The candidate treats the discursive (dissertative) questions as open essays and writes everything known about the topic, without addressing the command verb (compare, justify, propose, analyze) or producing the requested artifact.
Discursive items are scored against explicit criteria tied to the command verb; broad exposition earns partial credit at best, and candidates under time pressure default to dumping content.
What to do instead
Underline the command verb and the number of required elements in the prompt, then structure the answer as numbered points matching each required element. Reserve 30 seconds to check that every requested element appears.
When asked to analyze an algorithm's complexity, the candidate reports the complexity of a known algorithm by name rather than deriving it from the loop structure or recurrence actually shown.
Candidates memorize a table of classic complexities and apply it even when the presented variant (extra nested loop, different recurrence, different base case) changes the result.
What to do instead
Count operations from the innermost statement outward, or expand the recurrence two or three levels until the pattern is visible. Only then compare with the memorized table.
In database questions, the candidate normalizes or writes SQL against the tables as imagined rather than the schema and functional dependencies explicitly given, producing a decomposition that loses a dependency or a join that drops rows.
Textbook examples are memorized with canonical schemas; real items supply unusual attribute sets and dependencies that break the memorized pattern.
What to do instead
List the given functional dependencies and candidate keys explicitly on scratch paper first. After proposing a decomposition, test each original dependency against the new relations and test the join for losslessness.
In operating-systems and concurrency questions, the candidate reasons about the intended behavior of the mechanism instead of executing the given interleaving or scheduling trace step by step.
Semaphore, mutex, and scheduler questions look like conceptual questions but are decided by the exact order of operations in the trace; candidates answer the general principle and miss the specific outcome.
What to do instead
Draw a timeline with one column per process/thread and advance it instruction by instruction, tracking each variable and lock state. Answer only from the final state of the timeline.
The candidate answers software-engineering and project-management items with personal workplace habits or tool preferences instead of the process model, metric, or technique the question names.
These topics feel like common sense, so candidates substitute anecdotal experience for the formal definition (e.g., what a sprint review produces, what a use-case actor is).
What to do instead
Identify which formal artifact or phase the question is about, recall its defined inputs and outputs, and map the options onto that definition. Treat personal experience as irrelevant unless the prompt explicitly asks for it.
In computer-architecture and digital-systems items, the candidate computes in decimal or ignores representation details (two's complement range, overflow, endianness, cache line alignment) and selects a plausible-looking number.
Mental arithmetic defaults to signed decimal, and representation constraints are easy to overlook when the question is phrased around a value rather than a format.
What to do instead
Convert all operands to binary or hexadecimal before computing, state the word width, and check the result against the representable range. Verify the sign and overflow explicitly as a separate step.
On questions about ethics, legislation, and professional responsibility (including the code of ethics and the role of the professional council), the candidate answers from general moral intuition rather than the specific normative rule.
These items test knowledge of formal norms and their scope of application; intuition often conflicts with the letter of the rule, and candidates do not study this content because it feels obvious.
What to do instead
Study the applicable normative texts as a checklist of duties, prohibitions, and competencies. For each scenario, identify which specific rule applies and quote its condition before judging.
The candidate spends disproportionate time on a single hard objective question, leaving the discursive questions and the answer sheet incomplete.
Objective items give immediate feedback of progress, so candidates over-invest in them; discursive items are deferred and then truncated, and the answer sheet is filled hastily or partially.
What to do instead
Set a fixed per-item time budget, skip and mark anything exceeding it, and reserve a protected block for the discursive questions. Transfer answers to the sheet in batches and verify numbering at the end.
On interdisciplinary items that combine computing with another field (health, environment, society, engineering), the candidate ignores the contextual data and answers only the technical fragment of the question.
ENADE deliberately frames items in applied contexts; candidates trained on isolated technical drills filter out the context as noise, missing the constraint that determines the right option.
What to do instead
Read the context first and extract the explicit constraints and the stakeholder affected. Then check each candidate answer against those constraints, not only against technical correctness.