VCE Applied Computing Data Analytics common mistakes
VCE Applied Computing Data Analytics · Understand the traps and how to avoid them.
When asked to distinguish primary from secondary data, the candidate labels any dataset found online as primary, or labels survey data collected by someone else as primary because it is 'about people'.
Candidates memorise 'primary = you collected it' as a slogan but never apply the test of who collected the data and for what original purpose, so provenance and purpose get conflated with format or topic.
What to do instead
Before answering, write two columns: 'who collected it' and 'for what original purpose'. Only data collected firsthand by the analyst for the current question is primary; anything reused from another purpose (including government open data, ABS statistics, company records) is secondary. Practise sorting ten real sources into the two columns and justify each in one sentence.
The candidate describes validation as 'checking the data is correct' and lists validation and verification as synonyms, so a question asking for the difference between them receives one blended definition.
Both terms sound like 'making sure data is right', and students revise from summary tables that list techniques without the underlying purpose of each.
What to do instead
Learn the one-line distinction: validation checks whether data is reasonable and within accepted rules at the point of entry (type, range, presence, length, format checks); verification checks whether data has been accurately transferred or entered by comparing it with the original source (double entry, visual proofreading). Then match each named technique to the correct category in a table and state what error each technique would catch.
When asked to clean a described dataset, the candidate removes or overwrites anomalous values without recording what was changed, and does not state how missing values were handled.
Cleaning is treated as a tidying exercise rather than a documented, reversible decision process, and candidates want a 'clean' final table to look competent.
What to do instead
Always describe cleaning as a sequence with an audit trail: identify the issue, decide a rule (delete, impute with mean/median, flag, leave), apply it, and log the count of affected records and the rationale. Mention that the original raw data is preserved and that cleaning decisions can bias results, so they must be reported alongside findings.
The candidate chooses a pie chart for data with many categories or a line graph for unordered categorical data, and does not label axes, units or a title on the described chart.
Chart types are picked by appearance or habit rather than by data type and purpose, and conventions like axis labels are treated as decoration rather than as part of the communication of findings.
What to do instead
Use a decision rule: comparison across categories = bar chart; trend over time = line graph; part-of-whole with few slices = pie chart; relationship between two numeric variables = scatter plot. For every chart described, state the title, both axis labels with units, and the intended audience takeaway in one sentence.
The candidate reports a correlation as proof of causation, writing that one variable 'causes' or 'leads to' another when only an association is shown.
The narrative of a finding feels stronger with causal language, and students do not habitually ask whether a confounding variable or the direction of the relationship could explain the pattern.
What to do instead
Phrase findings as 'is associated with' or 'moves with' unless the study design supports causation. For each conclusion, add a sentence naming at least one plausible confounding variable and one limitation of the data that restricts the claim, then state what additional evidence would be needed.
On legal and ethical questions, the candidate answers only with 'get consent' and does not identify the specific obligation, the relevant legislation or the competing ethical principle in the scenario.
Privacy is reduced to a single keyword, and students have not practised mapping a scenario to the privacy principles, the purpose of collection, and the difference between legal compliance and ethical expectation.
What to do instead
Structure every response in three parts: the legal requirement (for example, collecting only data necessary for the stated purpose, informing individuals, storing securely, allowing access and correction), the ethical consideration (fairness, transparency, avoiding harm, secondary use of data), and a concrete action the organisation should take. Name the principle, not just the slogan.
The candidate treats security and privacy as the same issue, so a question about protecting data from unauthorised access is answered with a discussion of consent forms.
Both concepts sit under 'keeping data safe' in students' notes, and the distinction between who may access data and how data is protected is rarely tested explicitly in class.
What to do instead
Separate the two: privacy concerns what data is collected, why, who is told and how it is used; security concerns how the data is protected technically and procedurally (access controls, encryption, authentication, backups, audit logs). When answering, state which concept the scenario is about and give one control from that concept's list.
When asked about project management, the candidate lists 'make a Gantt chart' and 'set a deadline' but cannot explain how milestones, dependencies, roles or a contingency plan address a described risk in the scenario.
Project management tools are memorised as artefacts rather than as responses to coordination problems, so students cannot link a tool to the specific failure it prevents.
What to do instead
For each tool, learn the problem it solves: Gantt chart shows sequence and dependencies; milestone list marks decision points for review; task allocation with named owners prevents duplication; risk register with contingency covers data access delays or missing data. In answers, name the tool, the risk it mitigates in this scenario, and what would happen without it.
The candidate describes the data life cycle as a straight line that ends at visualisation, omitting evaluation, storage, archiving and disposal, and omitting feedback loops back to earlier stages.
Textbook diagrams are often reproduced from memory as a simple sequence, and the later stages feel administrative rather than analytical.
What to do instead
Learn the cycle as collect, prepare/clean, analyse, visualise/communicate, then evaluate and manage (store, secure, archive, dispose), with the evaluation stage feeding back into improved collection. When a question asks about a stage, name the stage before and after it and state what output passes between them.
The candidate answers an extended-response question with definitions and no reference to the data, scenario or stakeholder provided in the question stem, so marks for application are lost.
Under time pressure, students default to recalled definitions and do not re-read the stem to extract the specific organisation, dataset or decision that must be addressed.
What to do instead
Underline the scenario nouns and the decision to be supported before writing. Then use a fixed paragraph pattern: name the concept, apply it to the named organisation and data, state the consequence for the decision-maker, and give one limitation. Check that every paragraph contains a word from the stem.