Designing Accurate Data Entry Procedures Notes

Code as a Tool for High-Quality Data Entry

Data quality axiom: “ $\text{High-quality input} \; \Rightarrow \; \text{High-quality output}$ .”
Four analyst objectives for better input:
- Create meaningful codes.
- Design efficient data-capture approaches.
- Assure complete & effective capture.
- Assure data quality via validation.
Quality of data = degree to which data remain consistently correct within preset limits.

Purposes Humans Have for Coding

Keeping track (identification only).
Classifying information.
Concealing information.
Revealing information.
Requesting (triggering) appropriate action.

Major Code Families & Illustrative Examples

Simple sequence code
- Arbitrary or sequential numbers w/ no intrinsic meaning.
- Ex: Order # $5676$ vs. full verbal description of furniture order.
- Advantages: uniqueness, implicit timing.
Alphabetic derivation code
- Digest meaningful substrings (ZIP, consonants, street digits) into account #.
- Ex: $68506KND7533TVG$ for magazine subscriber.
- Pitfalls: short names (ROE→RXX), data change revokes key.
Classification code
- Single symbol representing a mutually-exclusive class.
- Ex: Tax categories ${I,M,T,C,D,S}$ .
- Danger: duplicate letters → forced, unnatural symbols (computer=P, insurance=N, etc.).
Block sequence code
- Sequential numbers reserved in “blocks.”
- Ex: Software types — browser $100\text{–}199$ , DB $200\text{–}299$ .
- Hybrid benefits: quick next-number, intrinsic grouping.
Cipher code
- Direct substitution (letters↔digits) to hide info.
- Store markdown price of 25.00\(retail) → cipher “BIMC” → 18.75\.
- Security through obscurity; easy for insiders.
Significant-digit subset code
- Concatenate mini-codes with specificity.
- Example garment ticket $202\,395\,40\,10$ → dept 202 (maternity) | style 395 | color 40 (red) | size 10.
Mnemonic code
- Human memory aid; mix letters/numbers meaningfully.
- Blood-center city list: BUF, ROS, KEN … minimize shipping errors.
Unicode & glyph codes
- ISO character set $0\text{–}65{,}535$ accommodates all writing systems.
- Web notation: “こ” → Japanese “こ”.
- Empowers multilingual input & display.
Function code
- Triggers system action.
- Inventory example: $3$ ⇒ “Spoiled”, $8$ ⇒ “Journal Add”.

Eight General Guidelines When Designing Codes

Concise.
Stable (rarely change).
Unique.
Sortable.
Non-confusing (avoid O/0, I/1, Z/2).
Uniform format.
Modifiable/expandable.
Meaningful (unless secrecy needed).

Effective & Efficient Data Capture

Decide What to Capture

Two data types:
- Variable/transaction-specific (e.g., quantity).
- Differentiating/static (e.g., SSN + 3-letter last-name key).
Capture only once; let system store & reuse.

Let the Computer Do Repetitive Work

Auto-insert date/time from system clock.
Retrieve stored item descriptions; operator enters only key.
Example OCLC: one catalog record per title shared by thousands of libraries.

Avoid Bottlenecks & Extra Steps

Provide adequate capacity at “neck.”
Eliminate forms if online capture feasible; otherwise streamline form.

Start with a Good Form / GUI

Source documents logically arranged, machine-ready.
Online forms: radio buttons w/ default, drop-down with “--Select--” sentinel.

Contemporary Data Entry Technologies

Keyboards

QWERTY plus function keys, macro keys, ergonomic & wireless variants.

Bar Codes (1-D)

UPC encodes manufacturer + product + check-digit.
Mobile cams + apps (e.g., Delicious Library) now read them.

2-D / Matrix Codes (QR etc.)

Denser, orientation-independent.
QR recognisable via three nested corner squares.
Free to create; used for URLs, coupons, ticketing, study-room scheduling, etc.
Security reminder: malicious stickers possible → use secure scanners (e.g., Norton Snap).

RFID (Radio Frequency Identification)

Tag = antenna + chip; passive (< $0.05$ ) or active (> $1$ ).
Uses: Walmart inventory, toll passes, cattle tracing, supply-chain blockchain logging.
Privacy & ethics: potential person-tracking.

NFC (Near Field Communication)

Two-way RFID within ~10 cm.
Mobile wallet (Apple Pay, Google Pay), transit cards, coupon delivery.
Security via short range + cryptographic layers.

OCR (Optical Character Recognition)

Scans print/handwriting → text; boosts speed $60!\%–90\%$ over keystrokes.

MICR (Magnetic Ink Character Recognition)

Bank checks; magnetic ink line.
Resistant to stray marks; doubles as security feature.

Mark-Sense / OMR Forms

Bubble sheets (surveys, exams).
Pros: low training, high volume.
Cons: stray marks; limited alpha capture; alignment errors.

Ensuring Data Quality through Validation

Validate Input Transactions (macro-level)

Wrong data for system.
Unauthorized submitter.
Request to perform unacceptable function.

Validate Input Data (field-level)

Missing data test.
Field length test.
Class / composition (numeric vs. alpha).
Range / reasonableness (day $1!\leq!d\leq!31$ ; age < $120$ ).
Invalid value list (active/inactive/closed).
Cross-reference (price ≥ cost; area-code ↔ state).
Compare w/ stored data (part # exists).
Check-digit / self-validating codes.

Check-Digit Example (Luhn Algorithm)

Original numeric code $d<em>1d</em>2\dots d_n$ .
Double every second digit from right; if >9, add digits (e.g., $14→1+4=5$ ).
Sum all digits $S$ .
If $S \bmod 10 = 0$ ⇒ valid.
- Visa $16$ -digit, AmEx $15$ -digit numbers embed Luhn check digit.

Validation Order & Techniques

Syntax first (missing, length, composition).
Semantics next (range, value, cross-ref, check digit).
Regular expressions for pattern tests.
- JS snippet: /^[A-Za-z0-9]\w{2,}@[A-Za-z0-9]{3,}\.[A-Za-z]{3}$/ for email.
XML validation via DTD or powerful schema constraints.

Data Accuracy Advantages in Ecommerce

Self-service entry: customer knows own data best.
Stored info (cookies, autocomplete) ⇒ faster, fewer typos.
Reuse through entire fulfillment chain (invoice → pick list → shipping → re-order).
Immediate electronic feedback (confirmations, status) lets customers fix errors quickly.

Ethical, Practical & Forward-Looking Implications

Privacy vs. utility: RFID & cookies collect rich data; analysts must balance benefit w/ consent.
Coding stability vital for longitudinal databases; redesigns disrupt analytics (consulting story “Summer Code”).
Transparent mnemonic or revealing codes enhance job satisfaction, but may expose sensitive info.
UX design in ecommerce must weigh frictionless input against security (password, card verification value).
Blockchain integration with RFID offers tamper-evident supply chain but raises scalability & data-governance issues.

Quick Reference: Common Validation Problem Sources

Transposition (digits swapped).
Data truncation (field too short).
Mis-classification (code ambiguous).
Overlapping class ranges.
Look-alike symbols (O/0, S/5, Z/2).

Formulas & Notation Cheat-Sheet

Check-digit remainder: $\text{Check} = (10 - (S \bmod 10)) \bmod 10$ .
Unicode entity: \text{{&#xNNNN;} where $\text{NNNN}$ is hex.
Range test example: $0 \le \text{quantity} \le 9999$ .

Study Connections

Links to Chapter 8: data dictionary entries store code definitions, aid uniqueness & validation tables.
Links to Chapter 12: form design principles overlap with data-capture preliminaries.
Quality assurance theme (Part V) culminates here; accurate entry is first gate of QA lifecycle.