Cameroon GCE O-Level Computer Science common mistakes
Cameroon GCE O-Level Computer Science · Understand the traps and how to avoid them.
When converting a binary number to hexadecimal, the candidate regroups the bits starting from the leftmost bit instead of from the right, so 1011011 is split as 101 1011 and answered as 5B rather than 5B being checked against the correct 0101 1011 grouping from the right.
Textbooks show the grouping table left-to-right and candidates copy the visual pattern without internalising that each hex digit maps to a fixed 4-bit block anchored at the least significant bit; padding with leading zeros on the left feels like 'adding something that wasn't there'.
What to do instead
Practise a fixed ritual: write the binary string, mark groups of four from the rightmost bit with a pencil stroke, pad the leftmost group with zeros to make four, then substitute each group using a written 0-F table. Do ten conversions and check each by converting the hex back to binary.
The candidate writes that RAM 'stores data permanently' or that ROM is 'where the current program and its variables live', mixing the roles of main memory and secondary storage in definitions and in questions about what happens when power is switched off.
Everyday language calls a phone's 128 GB 'memory', so the everyday meaning (storage) overrides the technical meaning (volatile working memory). Candidates also memorise 'RAM = main memory' without attaching volatility and read/write behaviour to it.
What to do instead
Build a three-column table (RAM, ROM, secondary storage) with rows for volatility, read/write, typical contents, and speed, and fill it from memory before checking notes. Then answer 'what happens on power loss' for each. Use the phone example explicitly: 8 GB RAM is working memory, 128 GB is secondary storage.
When tracing a loop, the candidate executes the body one time too many or too few, for example counting a FOR loop from 1 TO 5 as six iterations, or treating a WHILE loop's condition as tested after the body.
Candidates read the loop as English ('repeat until you reach five') rather than as a precise machine rule, and they do not distinguish the pre-test WHILE from the post-test REPEAT...UNTIL, which differ by exactly one execution when the condition is initially false.
What to do instead
Trace on paper with a three-column table (iteration number, variable values, condition result) and write the condition result before each pass. For every loop, state aloud whether the test happens before or after the body. Verify by writing the expected output first, then tracing to confirm it.
The candidate describes a database table as a 'list of records' but cannot say what a primary key is for, allows duplicate values in the primary key column, or stores several values (such as three phone numbers) in one field.
Candidates learn the vocabulary (field, record, file) as a definition list and never connect keys to the integrity problems they prevent, so the concept has no function attached to it and is forgotten or applied wrongly.
What to do instead
Work a small example table of five students, deliberately insert a duplicate ID and a multi-valued phone field, then state what goes wrong when you try to update or delete a record. Rewrite the table into two tables linked by the key and note which column is unique and non-null.
When asked to write SQL, the candidate writes SELECT * FROM Students WHERE Class = 'Form 4' AND Age > 15 ORDER Class, omitting the BY keyword, or puts the WHERE clause after ORDER BY, producing a statement that is close to correct but will not run.
Candidates write SQL as loose English word order and have often only read queries rather than typed them, so the fixed clause order (SELECT, FROM, WHERE, GROUP BY, ORDER BY) has never been enforced by an error message.
What to do instead
Memorise the clause order as a chant and write it as a skeleton before filling in columns: SELECT ... FROM ... WHERE ... ORDER BY .... Type each query into any free SQL tool or a paper trace and check keyword by keyword, especially BY, quotation marks around text values, and the position of ORDER BY.
The candidate draws a flowchart or writes pseudocode for a problem and never initialises the accumulator or counter before the loop, or uses an undefined variable name, so the algorithm would fail or give a wrong total even though the logic 'looks right'.
Attention is on the visible control structure (the loop, the decision) and the setup steps feel like boilerplate; candidates also rename variables casually between steps and do not re-read for consistency.
What to do instead
Adopt a checklist applied to every algorithm: all variables declared and initialised, names spelled identically everywhere, loop has a terminating condition, both branches of every decision produce an output or a next step. Apply it to your own answer before moving on, and dry-run with the smallest possible input (one item) and with zero items.
The candidate defines a local area network by size alone ('a network in one room') and cannot distinguish a switch from a router, or a hub from a switch, when asked to explain how devices are connected in a star topology.
Topology diagrams in notes show shapes without labelling the central device's behaviour, so candidates memorise the picture and the word 'star' but not what the central device does with a frame, and 'small area' is the only LAN criterion they retain.
What to do instead
Learn each device by one action: hub copies a frame to every port, switch forwards to the port holding the destination MAC address, router forwards between different networks using IP addresses. Then re-describe a star LAN using those verbs, and state LAN criteria as ownership plus geographic span, not room size.
The candidate answers a health-and-safety question with 'computers are dangerous' or lists only eye strain, missing repetition strain injury, back and neck problems, and the electrical and fire risks of cables and overloaded sockets, or confuses health with security threats such as viruses.
Health and safety is taught late and briefly, and candidates lump all 'bad things about computers' together, so data loss and malware get written where posture and cabling are being examined.
What to do instead
Sort hazards into three labelled lists: physical environment (posture, seating, lighting, cables, sockets, ventilation), health effects (eye strain, RSI, back pain, stress), and security (malware, phishing, data loss). For each, write one preventive measure. Answer only from the list the question names.
When asked to test a program, the candidate writes 'I would run it and see if it works', offering no test data, no expected result, and no distinction between normal, boundary, and erroneous data.
Candidates have rarely seen a test plan, so 'testing' means 'running once with whatever I type', and boundary values are not obvious until someone points out that a loop from 1 to 10 should be tested at 0, 1, 10 and 11.
What to do instead
For any input with a stated range, produce a table with columns: test data, type (normal/boundary/erroneous), expected result, actual result. Always include the lowest and highest valid values, one below and one above, and one clearly invalid entry such as a letter where a number is required.
The candidate answers 'what is an operating system' with a single function ('it runs programs') and cannot name memory management, file management, process scheduling, or the user interface, and similarly gives one-word answers for utility programs and application software categories.
The topic is treated as a definition to be memorised in one line, and the functions are invisible in daily use, so candidates have no concrete example to attach each function to and default to the vaguest true statement.
What to do instead
For each function, attach a visible example: memory management is why a large program fails to open when many tabs are open; file management is the folder tree; scheduling is why music keeps playing while a document saves; the interface is the desktop. Write the function plus its example as a pair, then answer in full sentences naming at least three functions.