2.3 · Data Control — Validation, Verification & Parity

Goal: distinguish validation from verification, list common techniques for each, and explain parity-check error detection.

Why we need data control

Garbage in → garbage out. If a user types Bn instead of Bob, the system might:

• send the wrong email,
• charge the wrong account,
• print the wrong name on a graduation certificate.

Data control is the umbrella term for techniques that prevent, detect or correct errors as data enters or moves through a system.

The three families of technique

Family	Performed by	When
Validation	Software (automatic)	At the point of data entry
Verification	Human (manual)	At the point of data entry
Parity check	Hardware / software	During transmission or storage

Memory hook

• Validation = Validated by code (the Verbose computer)
• Verification = Verified by a Viewer (a human)
• Parity = the post-it on the data saying "I have an odd / even number of 1s"

Validation in detail

A validation rule is a check performed by the software against pre-defined criteria. If the input fails, the user is asked to re-enter.

Common validation types

Type	What it checks	Example
Range check	Value is within an acceptable range	Age between 0 and 150
Presence check	Field is not empty	Name field is required
Format / picture check	Matches a pattern	HK_ID = X123456(7)
Length check	Number of characters within bounds	Password length ≥ 8
Type / data type check	Numeric vs text vs date	Phone number is digits only
Check digit	Final digit calculated from the others	HKID check letter, ISBN, credit card (Luhn)
Lookup check	Value exists in a reference list	Country code is in ISO 3166
Consistency check	Two fields agree	Departure date ≤ return date

Worked example · Check digit (HKID)

The HKID check letter is calculated from the digits using a weighted sum modulo 11. The software performs the same calculation on the data you typed; if the result differs from the letter, your input is rejected — even a single typo is caught.

Validation does not guarantee accuracy

A validation rule confirms the input is plausible, not correct. "Alice's age = 17" passes a range check even if Alice is really 16.

Verification in detail

Verification is the human check that data has been transferred correctly.

Common verification techniques

Technique	How it works
Visual check	The user re-reads what they typed against the source document
Double entry	The same data is typed twice (often by two different operators); the system compares
Read-back	A call-centre agent reads the data back to the caller
Print and proofread	Common in publishing

Worked example · Password confirmation

Almost every signup form asks you to type your password twice. This is double entry — a verification step, not validation. The software cannot know whether your typed password matches what you intended; the second entry is your manual check.

Validation vs verification side-by-side

Aspect	Validation	Verification
Done by	Software	Human
Detects	Implausible values	Transcription / typing mistakes
Example	age <= 150 rule	Re-typing the password
Guarantees correctness?	No	No, but greatly reduces typing errors
Performed when	Real-time at entry	At entry or after

Parity checking

Parity is a hardware-level error-detection technique used during transmission or storage.

How even parity works

When sending an n-bit data byte, the sender appends a parity bit so that the total number of 1s (data + parity) is even.

Example (even parity, 7 data bits + 1 parity bit):

Data bits	Number of 1s	Parity bit	Sent byte
1010101	4 (even)	0	10101010
1010100	3 (odd)	1	10101001

The receiver counts the 1s in the received byte. If the count is odd (under even parity), the byte was corrupted in transit.

Odd parity

Odd parity works the same way but the total number of 1s should be odd. Either convention works as long as sender and receiver agree.

Limitations

• Detects single-bit errors only. Two flipped bits cancel out.
• Cannot correct errors, only detect them.
• More advanced schemes (Hamming code, CRC) detect and correct multi-bit errors but are beyond the syllabus.

Where parity appears

• Older modems and serial cables.
• RAM ECC (Error-Correcting Code) modules use a richer form.
• Storage devices for integrity (e.g. RAID has parity disks).

A complete pipeline

                  ┌─────────────────┐
  USER  ──INPUT──▶│ Validation rule │ Reject if rule fails
                  └─────┬───────────┘
                        │
                        ▼
                  ┌─────────────────┐
                  │ Verification    │ User re-checks / re-types
                  └─────┬───────────┘
                        │
                        ▼ stored
                  ┌─────────────────┐
                  │ Parity check    │ Detects bit flip during transmission
                  └─────────────────┘

Practice activity

Classify each as validation (V), verification (R) or parity (P):

1. The login form refuses passwords shorter than 8 characters.
2. You re-enter your new password to confirm.
3. The HKID page rejects an invalid check letter.
4. A network card detects a corrupted Ethernet frame.
5. The librarian reads the borrower's surname back to the student before scanning.
6. The membership form refuses non-numeric phone numbers.

Answers

1. V (length / range)
2. R (double entry)
3. V (check digit)
4. P (parity / similar at frame level)
5. R (visual / read-back)
6. V (data type)

Exam-style question

Q (5 marks): A clinic's online booking form collects the patient's name, HKID, phone number and preferred appointment time. Suggest one suitable validation check for each of the four fields, and one verification technique for the whole form.

Sample answer:

• Name — presence check (must not be empty); optionally length ≤ 100.
• HKID — format check (X123456(7)) and check-digit verification.
• Phone number — length check (8 digits in Hong Kong) and type check (digits only).
• Appointment time — range check (within clinic operating hours) and consistency check (not in the past).
• Verification — show a summary page and ask the patient to confirm before submission (visual check), or send a confirmation email/SMS with a link to confirm.

Common student mistakes

• Saying validation "guarantees" correct data — it doesn't.
• Listing "captcha" as validation — captchas confirm humanity, not data quality.
• Confusing parity with encryption — parity is error detection, encryption is secrecy.
• Forgetting that a check digit is a special kind of validation.

Key takeaways

• Validation = software check at entry.
• Verification = human check at entry.
• Parity = bit-level error detection during transmission/storage.
• All three increase data quality but none guarantees correctness.

➡️ Next chapter: 3 · Data Representation