Audit Sampling and Substantive Testing under SA 530: Documenting Sample Sizes, Tolerable Error, and Extrapolation

1. Introduction to SA 530 and the Statutory Audit Framework

In statutory and tax audits, verifying 100% of transactions is practically impossible and economically unfeasible for most corporate and commercial enterprises. Standard on Auditing (SA) 530, Audit Sampling, issued by the Institute of Chartered Accountants of India (ICAI), provides the mandatory technical architecture for applying audit procedures to less than 100% of items within a population of audit relevance. The fundamental premise of SA 530 is to ensure that all sampling units have a chance of selection, providing the auditor with a reasonable basis on which to draw conclusions about the entire population.

Audit sampling applies to both Tests of Controls (evaluating the operating effectiveness of internal controls in preventing or detecting material misstatements) and Substantive Procedures (detecting material monetary misstatements at the assertion level, including tests of details). However, sampling is not applicable to inquiry, analytical procedures, or walk-through inspections where an auditor evaluates the design of a system. With regulatory oversight tightening through the National Financial Reporting Authority (NFRA), the Quality Review Board (QRB), and ICAI Peer Review mandates, auditors can no longer rely on arbitrary rule-of-thumb sampling (such as picking the first 5 invoices of every month or auditing an ad-hoc 10% of vouchers). Workpapers must rigorously demonstrate statistical justification, risk assessment linkage, tolerable error definitions, and quantitative projection of discovered misstatements.

2. Core Concepts: Population, Sampling Unit, and Stratification

Before selecting a sample, the auditor must carefully establish the parameters of the population and the sampling units:

• Population Completeness (SA 500): The auditor must ensure that the population from which the sample is drawn is complete, valid, and appropriate for the specific audit objective. For example, when testing for unrecorded liabilities (completeness of trade payables), sampling from the recorded creditor ledger is conceptually flawed; the appropriate population consists of subsequent cash disbursements, unpaid receiving reports, and post-year-end vendor statements.
• Sampling Unit: The individual items that make up the population. A sampling unit can be a physical document (an invoice, a goods receipt note, a debit memo), a ledger balance, an individual debtor account, or even individual monetary currency units (as used in Monetary Unit Sampling).
• Stratification: The process of dividing a heterogeneous population into sub-populations (strata), each of which is a group of sampling units with similar characteristics (often monetary value). Stratification reduces the variability of items within each stratum, allowing the auditor to significantly reduce sample size while directing auditing effort toward high-value, high-risk items.

The 3-Tier Stratification Framework

Modern risk-based auditing segregates any financial ledger into three distinct evaluation tiers:

1. Tier 1 — High-Value & Key Items (100% Examination): All individual items exceeding performance materiality or specific risk criteria (e.g., related-party vouchers, round-sum year-end entries, litigation provisions). These items are tested 100% and excluded from the sampling population.
2. Tier 2 — Representative Sampling Population: The remaining homogenous population where sampling risk is formally managed. Items are sampled using statistical or structured non-statistical methods.
3. Tier 3 — Immaterial / Trivial Balances: Items falling below the clearly trivial threshold (CTT) where aggregate exposure is demonstrably negligible, requiring only high-level analytical review.

3. Statistical vs Non-Statistical Sampling: Technical Comparison

SA 530 permits both statistical and non-statistical sampling methods, provided the auditor designs and selects an audit sample, performs audit procedures on selected items, and evaluates the results to obtain sufficient appropriate audit evidence. The comparative operational characteristics are outlined below:

Audit Parameter	Statistical Sampling	Non-Statistical Sampling
Mathematical Basis	Rooted in probability theory (normal distribution, Poisson, binomial).	Rooted in auditor's professional judgment and qualitative experience.
Selection Technique	Random selection, systematic random selection, Monetary Unit Sampling (MUS).	Haphazard selection, block selection (discouraged), judgmental quotas.
Measurement of Sampling Risk	Quantified mathematically at a precise confidence level (e.g., 95% confidence).	Evaluated qualitatively without mathematical measurement of sampling error.
Defensibility before Regulators (NFRA/QRB)	Highly objective; reproducible results with clear mathematical audit trail.	Requires extensive documented justification of professional rationale and bias controls.
Extrapolation Mechanism	Rigorous statistical projection with calculated upper error limits (UEL).	Ratio or difference projection to the stratum with qualitative cushion.

4. Determining Sample Size and Managing Sampling Risk

Sampling risk is the risk that the auditor's conclusion based on a sample may be different from the conclusion if the entire population were subjected to the same audit procedure. It manifests in two critical forms:

• Risk of Incorrect Rejection (Alpha Risk / Type I Error): In substantive testing, concluding that a population is materially misstated when it is not. This affects audit efficiency, leading to unnecessary additional work.
• Risk of Incorrect Acceptance (Beta Risk / Type II Error): Concluding that a population is not materially misstated when it is in fact materially misstated. This directly impacts audit effectiveness and can lead to an inappropriate audit opinion.

The Quantitative Sample Size Formula for Tests of Details

In substantive testing of account balances (e.g., trade receivables, inventories, fixed assets), statistical sample sizes are governed by the relationship between Population Recorded Amount, Tolerable Misstatement, Expected Misstatement, and the Confidence Factor:

Sample Size (n) = (Population Recorded Amount - 100% Examined Items) × Confidence Factor / (Tolerable Misstatement - (Expected Misstatement × Expansion Factor))

Where:

• Tolerable Misstatement (TM): A monetary amount set by the auditor in respect of which the auditor seeks to obtain an appropriate level of assurance that the monetary amount set by the auditor is not exceeded by the actual misstatement in the population. TM is typically set at or below Performance Materiality (usually 50% to 75% of Overall Materiality).
• Expected Misstatement (EM): The anticipated monetary error in the population based on prior year audit findings, current year risk assessments, and internal control effectiveness.
• Confidence Factor: Derived from Poisson distribution tables corresponding to the desired level of assurance (Risk of Incorrect Acceptance):
  • High Assurance (5% Risk of Incorrect Acceptance / 95% Confidence): Factor = 3.0
  • Moderate Assurance (10% Risk of Incorrect Acceptance / 90% Confidence): Factor = 2.31
  • Low Assurance (20% Risk of Incorrect Acceptance / 80% Confidence): Factor = 1.61

5. Monetary Unit Sampling (MUS) vs Classical Variables Sampling

Monetary Unit Sampling (also known as Probability-Proportional-to-Size or PPS sampling) is the most widely adopted statistical method in corporate auditing. Under MUS:

1. Every individual rupee in the population is treated as a separate sampling unit.
2. The cumulative monetary ledger is built, and items are selected at a constant monetary sampling interval: Sampling Interval (SI) = Tolerable Misstatement / Confidence Factor
3. Any line item whose book value equals or exceeds the Sampling Interval is automatically selected for testing (Tier 1 items).
4. Accounts with larger rupee balances have a proportionally higher probability of selection, naturally biasing testing toward high-value transactions while giving every rupee an equal chance.

6. Executing Audit Procedures, Unexamined Items, and Anomalous Errors

When executing audit tests on selected samples, strict adherence to SA 530 requires specific procedural discipline:

• Handling Missing Documentation: If a supporting voucher, invoice, or third-party confirmation cannot be located, the auditor cannot simply replace it with another convenient sample item. An unexamined item must be evaluated as a misstatement unless alternative substantive procedures provide sufficient appropriate audit evidence.
• Replacement Sampling Restrictions: Replacement of a sample item is permitted only when the item is genuinely non-applicable to the audit assertion being tested (e.g., a voided cheque that was cancelled in the ordinary course of business prior to execution).
• Anomalous Errors: A misstatement or deviation that is demonstrably not representative of misstatements or deviations in a population. To establish an error as anomalous, the auditor must obtain a high degree of certainty through additional audit procedures proving that the error arose from an isolated, non-recurring event and has no systemic counterpart.

7. Mathematical Projection of Misstatements to the Population

For tests of details, the auditor is statutorily mandated to project monetary misstatements found in the sample to the population to obtain a broad view of the scale of misstatement. Two primary projection methods are applied:

Method 1: Ratio Projection

The ratio of misstatement discovered in the sample is applied across the entire sampling stratum:

Projected Misstatement = (Monetary Value of Sample Misstatements / Total Monetary Value of Sample Tested) × Total Population Monetary Value

Method 2: Difference Projection

The average misstatement per sampling unit is multiplied by the total number of units in the population:

Projected Misstatement = (Monetary Value of Sample Misstatements / Number of Sample Items Tested) × Total Number of Items in Population

Total Population Misstatement:

Total Estimated Misstatement = 100% Examined Item Misstatements + Projected Sample Misstatements + Allowance for Sampling Risk

If Total Estimated Misstatement approaches or exceeds Tolerable Misstatement, the auditor cannot conclude that the population is free from material misstatement. The auditor must request management to investigate and adjust the ledger, extend sample testing, or consider the impact on the Auditor's Report under SA 705.

8. 7 Comprehensive Worked Examples and Practical Case Studies

Case Study 1: Trade Receivables Debtor Testing using Monetary Unit Sampling (MUS)

Context: Statutory audit of a manufacturing corporate with total Trade Receivables of ₹18,00,00,000 comprising 1,200 debtor accounts. Overall Materiality is ₹40,00,000; Performance Materiality is set at 75% (₹30,00,000); Tolerable Misstatement (TM) is ₹30,00,000; Expected Misstatement (EM) is ₹3,00,000. Desired assurance is High (95% confidence; factor = 3.0).

Step 1: 100% Key Items Identification: All debtors with balances ≥ ₹30,00,000 are segregated for 100% examination. 12 debtors total ₹5,40,00,000. No misstatements were found in this tier.

Step 2: Remaining Population: ₹18,00,00,000 - ₹5,40,00,000 = ₹12,60,00,000 (1,188 debtor accounts).

Step 3: Sampling Interval (SI): TM / Confidence Factor = ₹30,00,000 / 3.0 = ₹10,00,000.

Step 4: Sample Size (n): Remaining Population / SI = ₹12,60,00,000 / ₹10,00,000 = 126 sample units.

Step 5: Testing & Tainting Calculation: During debtor confirmation and subsequent realization testing, two misstatements are discovered:

• Sample Item A: Recorded balance ₹8,00,000. Verified audit balance ₹6,00,000 (unrecorded credit note). Misstatement = ₹2,00,000. Tainting percentage = 2,00,000 / 8,00,000 = 25%. Projected error = 25% × SI (₹10,00,000) = ₹2,50,000.
• Sample Item B: Recorded balance ₹4,00,000. Verified audit balance ₹3,00,000 (billing rate error). Misstatement = ₹1,00,000. Tainting percentage = 1,00,000 / 4,00,000 = 25%. Projected error = 25% × SI (₹10,00,000) = ₹2,50,000.

Projection Summary: Total projected sample misstatement = ₹2,50,000 + ₹2,50,000 = ₹5,00,000. Adding basic precision (3.0 × SI = ₹30,00,000 at zero error) vs actual projected error. Total Projected Misstatement (₹5,00,000) is well within TM (₹30,00,000). Population is accepted as fairly stated subject to booking the ₹3,00,000 actual errors.

Case Study 2: Fixed Assets Additions Stratification and Verification

Context: Plant & Machinery additions during the year total ₹6,50,00,000 across 480 capital asset purchase vouchers. Overall Materiality is ₹25,00,000; Performance Materiality / TM is ₹15,00,000.

Stratification Scheme:

• Stratum 1 (Additions > ₹10,00,000): 18 vouchers totaling ₹4,20,00,000 (64.6% of total additions). Tested 100%. One repair expenditure of ₹12,00,000 was wrongly capitalized as plant modification. Factual misstatement = ₹12,00,000.
• Stratum 2 (Additions between ₹1,00,000 and ₹10,00,000): 112 vouchers totaling ₹1,80,00,000. Sample size selected: 30 vouchers totaling ₹52,00,000. Discovered ₹1,30,000 of ineligible freight charges capitalized. Ratio projection = (₹1,30,00,00 / ₹52,00,000) × ₹1,80,00,000 = ₹4,50,000 projected misstatement.
• Stratum 3 (Additions < ₹1,00,000): 350 vouchers totaling ₹50,00,000. Evaluated via analytical review and high-level scrutiny; zero sampling required due to negligible aggregate risk.

Audit Conclusion: Total estimated misstatement = ₹12,00,000 (Stratum 1 factual) + ₹4,50,000 (Stratum 2 projected) = ₹16,50,000. Because ₹16,50,000 exceeds TM of ₹15,00,000, management was requested to adjust the ₹12,00,000 factual error. Post-adjustment, remaining projected error is ₹4,50,000 (well below TM), allowing an unqualified audit conclusion.

Case Study 3: Inventory Valuation Pricing Test & Handling Anomalous Errors

Context: Raw material inventory of ₹9,20,00,000 comprising 2,400 distinct SKU items. Tolerable Misstatement is ₹20,00,000.

Audit Sampling Execution: A random sample of 80 SKU lines totaling ₹1,60,00,000 is selected for vendor purchase invoice price verification.

Findings: One imported chemical SKU (recorded value ₹14,00,000) showed an error of ₹7,00,000 due to a customs duty calculation software glitch that occurred strictly on August 14th during an ERP version update. Investigation revealed that only 3 import bills were cleared on that specific day under that patch, totaling ₹18,00,000, and no other days were affected.

Audit Treatment under SA 530: The auditor cannot arbitrarily label this error as anomalous without corroborating evidence. The auditor examined 100% of the transactions processed under the faulty August 14th ERP patch, identifying total errors of ₹8,20,000 across all 3 bills. Because the root cause was verified as isolated and completely resolved, the ₹8,20,000 was treated as a specific known factual misstatement rather than projected across the entire ₹9,20,00,000 inventory population. The remaining 79 sample items showed zero errors, confirming the regular population is cleanly stated.

Case Study 4: Operating Expenses Haphazard Sampling & Ratio Extrapolation

Context: Legal and Professional charges of ₹3,40,00,000 comprising 850 payment vouchers. Audit objective: verify Section 194J TDS compliance and business purpose deductibility under Section 37(1).

Method: Haphazard selection of 60 vouchers totaling ₹75,00,000. Testing revealed that 4 vouchers totaling ₹4,50,000 lacked deduction of TDS, leading to potential disallowance under Section 40(a)(ia).

Extrapolation: Misstatement percentage in sample = ₹4,50,000 / ₹75,00,000 = 6.0%. Projected misstatement across population = 6.0% × ₹3,40,00,000 = ₹20,40,000. With Performance Materiality at ₹18,00,000, the projected non-compliance exceeded tolerable limits. The auditor expanded substantive testing to 100% of all legal retainerships above ₹2,00,000, identifying an actual un-deducted TDS pool of ₹11,20,000, which management disallowed in Form 3CD, eliminating audit risk.

Case Study 5: Cut-Off Testing for Unrecorded Trade Liabilities

Context: Testing trade payables completeness at March 31. The recorded creditor listing reflects ₹12,00,00,000.

Sampling Procedure: Instead of sampling from the recorded March 31 ledger, the auditor sampled 100% of receiving reports (Goods Receipt Notes - GRNs) generated between March 24 and April 7, and all bank payments > ₹5,00,000 executed between April 1 and April 30.

Discovery: Out of 45 post-year-end payments sampled (totaling ₹4,10,00,000), 5 payments totaling ₹38,00,000 pertained to services received and goods inspected prior to March 31 for which no liability was accrued.

Extrapolation & Impact: Ratio of unaccrued liability in post-year-end disbursements = 38,00,000 / 4,10,00,000 = 9.27%. Applied across total April disbursements of ₹7,80,00,000, projected unrecorded liabilities = ₹72,30,000. Management agreed to pass a year-end audit journal entry accruing ₹38,00,000 known liabilities and an additional general accrual of ₹34,30,000, ensuring complete presentation.

Case Study 6: Internal Financial Controls (IFCoFR) Attribute Testing

Context: Testing operating effectiveness of manual supervisory credit control approval on sales orders exceeding customer credit limits. Total sales orders exceeding credit limits during the year: 1,400 transactions.

Audit Parameters: Tolerable Deviation Rate = 5%; Expected Deviation Rate = 0%; Desired Confidence Level = 95%. Based on attribute sampling tables (AICPA / ICAI), the required sample size is 60 transactions with zero allowable deviations.

Result: In the sample of 60 orders, 2 instances were found where sales orders were released without the mandatory signature or digital approval of the Chief Commercial Officer. Deviation rate in sample = 2 / 60 = 3.33%.

Evaluation: Although 3.33% is less than the tolerable rate of 5%, under attribute sampling tables, finding 2 deviations in a sample of 60 implies that the upper deviation limit is 10.3% at 95% confidence. Because 10.3% > 5%, the auditor concludes that the internal control cannot be relied upon as effective. The auditor was required to expand substantive testing of year-end trade receivables recoverability and issue a qualified operating effectiveness remark in the IFCoFR Annexure to the Independent Auditor's Report.

Case Study 7: Cash Verification and Handling Missing Vouchers at Multiple Branches

Context: Statutory audit of a retail chain with 45 branch outlets holding cumulative petty cash balances of ₹90,00,000 (average ₹2,00,000 per branch). Materiality is ₹15,00,000.

Sampling Scheme: 12 branches selected via systematic random sampling for surprise cash verification on March 31. At Branch #18, petty cash physical balance was ₹1,20,000 against recorded book balance of ₹2,00,000. Branch manager stated that ₹80,000 was disbursed for local store repair, but no vouchers or invoices were available.

Procedural Action: Under SA 530, the auditor refused to accept oral representations. The missing ₹80,000 was treated as an unexplained shortage (misstatement). Extrapolating the ₹80,000 shortage across the 45 branches gave a projected cash deficit of ₹3,00,000. Furthermore, the incident triggered fraud risk considerations under SA 240, prompting 100% physical cash verification across all remaining 33 branches, uncovering an additional ₹1,40,000 shortage, which was formally debited to employee advance and provided for.

9. Documenting Audit Sampling to Withstand NFRA and Peer Review Scrutiny

Recent audit inspection reports from NFRA emphasize that workpapers lacking documented proof of sample design, population reconciliation, and extrapolation will be treated as non-compliance with SAs. Standardized audit documentation must include:

1. Sampling Strategy Memo: Documenting the business rationale for choosing statistical vs non-statistical approaches.
2. Population Reconciliation: Tie-out of the sampled population to the audited General Ledger and Trial Balance.
3. Parameter Justification: Explicit basis for selecting Tolerable Misstatement (linked to Performance Materiality) and Expected Misstatement.
4. Sample Selection Trail: The random seed number, selection interval, or haphazard selection criteria used, allowing exact re-performance.
5. Evaluation Sheet: Complete listing of discovered misstatements, calculation of tainting percentages, extrapolation formula, and comparison against Tolerable Error.

10. Frequently Asked Questions (FAQs)

Q1: Can an auditor use rule-of-thumb sampling (e.g., selecting 10% of invoices) under SA 530?

No. Arbitrary rule-of-thumb approaches (such as picking 10% of invoices or testing the first week of every month) do not constitute valid statistical or structured non-statistical sampling under SA 530. Such methods fail to provide every sampling unit with a chance of selection and cannot quantify or manage sampling risk. Audit regulators (NFRA/QRB) routinely flag arbitrary percentage sampling as gross professional negligence.

Q2: What is the exact difference between Tolerable Misstatement and Performance Materiality?

Performance Materiality (SA 320) is set by the auditor to reduce to an appropriately low level the probability that the aggregate of uncorrected and undetected misstatements exceeds Materiality for the financial statements as a whole. Tolerable Misstatement (SA 530) is the application of performance materiality to a specific sampling procedure for an account balance or class of transactions. In practice, Tolerable Misstatement is set equal to or lower than Performance Materiality.

Q3: What should an auditor do if a selected sample voucher is missing from client records?

Under SA 530.12, if the auditor is unable to apply the designed audit procedures to a selected item, the auditor shall perform the procedure on an alternative item only if the original item was invalidly selected. If the voucher is simply missing, the auditor must treat that item as a 100% misstatement or deviation. The auditor cannot replace a missing voucher with an easily available voucher, as this creates selection bias and conceals potential fraud or control failures.

Q4: How should misstatements in Tier 1 (100% examined items) be treated during extrapolation?

Misstatements discovered in 100% examined key items must never be projected across the remaining population. Because these items were tested in their entirety, their exact monetary misstatement is known with 100% certainty. Extrapolation is strictly performed on errors discovered within the representative sampled stratum, and the factual misstatements from Tier 1 are added directly to the projected sample error.

Q5: When is it mandatory to use statistical sampling instead of non-statistical sampling?

SA 530 does not mandate statistical sampling over non-statistical sampling; the choice is left to the auditor's professional judgment. However, for large, homogeneous populations with thousands of transactions (such as retail banking loan books, e-commerce sales, or large telecom billing ledgers), statistical sampling (such as MUS) is practically essential because it mathematically optimizes sample sizes and provides objective, legally defensible results.

Q6: If total projected misstatement is less than Tolerable Misstatement, can the auditor conclude the ledger is clean?

Not automatically. The auditor must consider both the projected misstatement and the allowance for sampling risk (anomalous variance). If the projected misstatement is very close to Tolerable Misstatement, the risk that undetected misstatements could push total error above materiality remains high. The auditor must either obtain additional audit evidence or request management to correct known errors to create an adequate safety cushion.