NSSCO Computer Studies (6134) common mistakes
NSSCO Computer Studies (6134) · Understand the traps and how to avoid them.
When converting between binary, denary and hexadecimal, the candidate reverses the place values (treating the leftmost bit as 2^0) or reads a hexadecimal digit as its denary face value instead of its value (e.g. writing B as 11 but then computing with 1 and 1 separately).
Candidates memorise 'multiply by 2' and 'divide by 2' as isolated procedures without a secure grasp of positional notation, so under exam pressure they apply the operations in the wrong direction or lose track of which end is the least significant digit.
What to do instead
Before every conversion, write the place-value headings (128 64 32 16 8 4 2 1 for one byte; 16^1 16^0 for a two-digit hex number) and place the digits under them. For hex, always write the denary value of each letter above it (A=10 ... F=15) before calculating. Check by converting back the other way as a self-verification step, and practise converting a single byte in all three directions until the headings are automatic.
The candidate calculates a two's complement negative number by simply adding a minus sign to the magnitude, or negates a binary value by flipping the bits but forgets to add 1, producing an answer that is off by one.
The sign-and-magnitude method learned earlier feels intuitive, and the 'invert and add one' rule is memorised as a chant without understanding that the leftmost bit carries a negative place value.
What to do instead
Learn the negative place-value reading: in an 8-bit two's complement number the leftmost bit is worth -128. Read 1111 0110 as -128 + 64 + 32 + 16 + 4 + 2 = -10. Do five conversions each way using this method, then confirm with the invert-and-add-one procedure, so both routes agree. Always state the bit width, because the same pattern means different values at 4, 8 and 16 bits.
The candidate describes the fetch-decode-execute cycle vaguely ('the processor does the instruction') and omits the roles of the program counter, memory address register, memory data register, current instruction register and control unit, or places them in the wrong order.
Register transfer is abstract and invisible, so candidates learn a summary sentence instead of the data flow, then cannot answer a question that asks what happens to the program counter during a jump or where an operand is held.
What to do instead
Draw the cycle as a numbered diagram with the buses as arrows: PC to MAR, memory read, data to MDR, MDR to CIR, decode, execute, PC incremented. Annotate what each register holds at each step and where the increment happens relative to a branch. Rehearse the diagram from memory and then answer 'what if' questions (interrupt, branch, immediate addressing) against it.
The candidate writes 'RAM stores data permanently' or 'ROM is volatile', or confuses cache with main memory and virtual memory with secondary storage when asked to compare them.
Everyday use of 'memory' and 'storage' blurs the technical distinction, and the four-way comparison (speed, volatility, capacity, cost per byte) is learned as a table to be recited rather than an ordering to be reasoned from.
What to do instead
Build a single ladder from registers, cache, RAM, SSD/HDD, off-line backup, and write for each rung: volatile or not, typical size, relative speed, cost per gigabyte, and what it holds. Answer comparison questions by moving along that ladder. For virtual memory, state explicitly that it is a portion of secondary storage used as an extension of RAM, which is why it is slower.
The candidate answers a network question by naming a device or protocol without matching it to the layer or purpose: for example saying a router 'connects computers in a LAN', a switch 'assigns IP addresses', or treating TCP and IP as interchangeable.
Device and protocol names are learned as a vocabulary list, so the candidate retrieves a familiar word rather than reasoning about addressing, forwarding and reliability.
What to do instead
Make a two-column table: device or protocol, and the single job it does plus the address it uses (MAC for switch, IP for router, port for a service). For TCP/IP, write one sentence each: IP moves packets between hosts, TCP provides ordered, reliable delivery with acknowledgements and retransmission. When a question names a scenario, first decide the layer, then choose the device or protocol.
The candidate describes a database as 'a collection of files' and cannot distinguish a primary key from a foreign key, or draws an entity-relationship diagram in which a foreign key is placed in the wrong table or duplicated into both tables.
The concept of referential integrity is invisible in a flat table, and candidates copy example diagrams rather than deriving the key placement from the one-to-many relationship.
What to do instead
For every relationship, state the cardinality in words first ('one customer places many orders'), then apply the rule: the primary key of the 'one' side becomes a foreign key on the 'many' side. Check referential integrity by asking what happens when a parent row is deleted. Practise with three entities and verify each design by writing the actual table columns with keys marked PK and FK.
When writing an algorithm, the candidate produces pseudocode that mixes languages (Python print with C-style braces), uses = for comparison, or omits initialisation of a total or counter before a loop, so the logic fails on a dry run.
Candidates practise by typing code into an interpreter that forgives or flags errors instantly, so they never build the habit of desk-checking a written algorithm against a trace table.
What to do instead
Adopt one pseudocode convention and write it at the top of your answer. Before moving on, hand-trace with a trace table containing every variable plus output, using two or three test values including a boundary case (zero items, one item, maximum). Explicitly write initialisation lines (total = 0, count = 0) and use == or the words 'is equal to' for comparison.
The candidate writes a loop whose condition is wrong at the boundary, producing an off-by-one error: a FOR loop that processes n+1 items, a WHILE loop that never terminates because the counter is not incremented, or an array traversal that reads index n of an n-element array.
Loop bounds are reasoned about loosely as 'repeat until done', and the candidate does not track the value of the control variable at the moment the condition is tested.
What to do instead
For every loop, write the values of the control variable at the first and last iteration, and state whether the loop is count-controlled or condition-controlled. For arrays, state the valid index range explicitly (0 to n-1) before writing the loop. Trace the terminating case by hand. Check the loop body updates whatever the condition depends on.
On a security question, the candidate lists 'a firewall and antivirus' as a complete answer, or defines phishing, pharming, spyware and a Trojan as synonyms, without linking each threat to a specific mechanism and countermeasure.
Security is learned as a list of software products rather than as threats matched to vulnerabilities, so answers stay generic and miss marks for the mechanism (e.g. deceptive email link versus DNS redirection versus keystroke logging).
What to do instead
Build a threat table with columns: how it reaches the victim, what it does, warning signs, and the countermeasure that actually addresses it (user education and sender verification for phishing, DNS security and certificates for pharming, anti-spyware and permission review for spyware, avoiding untrusted downloads for Trojans). Add the human and procedural controls (strong unique passwords, two-factor authentication, backups, patching) alongside technical ones.
The candidate answers an ethics or 'impact of ICT' question with a one-sided list of benefits, or gives an opinion with no justification, when the question asks for both advantages and disadvantages or for a reasoned judgement.
Such questions look like invitations to write everything known about the topic, so candidates dump facts and ignore the command word (discuss, evaluate, justify) and the marks available for balance and conclusion.
What to do instead
Underline the command word and the number of marks, then plan one line per mark. For discuss or evaluate questions, write at least one advantage and one disadvantage, each with a consequence ('automation reduces labour cost, but removes low-skill jobs, so retraining is needed'), and finish with a short judgement naming the condition under which one side outweighs the other. Keep each point tied to the stated context rather than generic statements.