Random drug and alcohol testing is one of the most widely used and legally scrutinised elements of any workplace testing program. The premise is straightforward: by selecting employees at random, organisations create a credible deterrent without targeting individuals. But here is the uncomfortable truth — many organisations that believe they are conducting random testing are not. Their selection processes contain flaws that undermine randomness, introduce bias, and create legal vulnerability.

This article examines the most common flaws in random selection processes, the legal consequences of getting it wrong, and what genuinely defensible random selection looks like in practice.

Why Randomness Matters

The entire legal and ethical justification for random testing rests on one principle: every eligible employee has an equal chance of being selected on any given occasion. If this principle is not upheld — whether through bias, procedural error, or inadequate methodology — the testing program’s credibility collapses.

In Fair Work Commission proceedings, employees who have been dismissed following a positive drug test frequently challenge the circumstances of their selection. If the employer cannot demonstrate that the selection was genuinely random, the positive test result — and any disciplinary action based on it — may be set aside, regardless of whether the employee was actually impaired.

Common Flaws in Manual Selection

Many organisations still select employees for random testing using manual methods. These methods are not merely outdated — they are legally dangerous.

Drawing Names from a Hat

While this may seem intuitively fair, it fails on multiple counts. The physical mechanics of drawing — how names are folded, placed, and extracted — introduce variables that can skew results. More critically, there is no verifiable audit trail. If challenged, the employer cannot prove who was in the pool, how the draw was conducted, or that it was not manipulated.

Manager Discretion

Some organisations allow site managers or supervisors to “select randomly” at their discretion. This is not random selection; it is subjective selection with a random label. Conscious or unconscious bias — selecting employees who are perceived as likely users, or avoiding employees who are in favour — is virtually unavoidable. This approach is indefensible in any legal forum.

Rotating Through a List

Sequential selection from an alphabetical or numerical list is predictable, not random. Employees can anticipate when their turn will come, which entirely defeats the deterrent effect of the program.

The Excel RAND() Problem

Many organisations have moved beyond purely manual methods to using spreadsheet-based selection, typically using Excel’s RAND() or RANDBETWEEN() functions. This is an improvement over drawing names from a hat, but it introduces its own set of issues.

Recalculation Risk

Excel’s RAND() function recalculates every time the spreadsheet is modified. Unless the user immediately copies and pastes the values before any other action, the original selection is lost — and with it, the audit trail. In practice, many users do not understand this behaviour, and the selection they record may not be the selection that was originally generated.

Pool Integrity

A spreadsheet-based selection is only as good as the data in the spreadsheet. If the employee list is not current — if it includes terminated employees, excludes new starters, or contains duplicates — the selection is drawn from an inaccurate pool. This is a common problem in organisations where the testing spreadsheet is maintained separately from the HR system.

Lack of Audit Trail

A spreadsheet does not inherently record when it was used, by whom, or what the precise state of the data was at the time of selection. Reconstructing the circumstances of a selection months or years after the fact — as may be required in legal proceedings — is extremely difficult with spreadsheet-based systems.

Bias Risks: Conscious and Unconscious

Even where an organisation uses a technically sound random method, bias can be introduced at other points in the process:

• Substitution — when a selected employee is unavailable (on leave, travelling, or at a remote location), who decides the replacement? If the replacement is chosen by a manager rather than by a random mechanism, bias has been introduced.
• Timing — if selections are consistently run at times when certain groups of employees are known to be absent (for example, during school holidays when parents are more likely to be on leave), the effective pool is skewed.
• Exclusion — are senior managers, executives, or specific teams excluded from the selection pool? Unless there is a documented, policy-based reason for exclusion, this undermines the program’s credibility and may constitute discrimination.

Legal Cases Where Selection Was Challenged

The Fair Work Commission and Australian courts have considered the integrity of random selection processes in numerous cases. While each case turns on its own facts, several consistent themes emerge:

• Employers bear the burden — when selection methodology is challenged, the employer must demonstrate that the process was genuinely random. The absence of an audit trail is treated as a significant deficiency.
• Pattern evidence matters — if an employee can show that they were selected disproportionately often relative to other employees, the employer must explain why. “Random chance” is a difficult argument when the data suggests otherwise.
• Procedural fairness is paramount — even a genuinely random selection can be undermined by procedural failures in how it was conducted, communicated, or documented.

In several reported decisions, dismissals following positive test results have been found to be unfair where the employer could not adequately demonstrate the integrity of the selection process — even where the employee’s impairment was not in dispute.

What Genuinely Random Selection Looks Like

A legally defensible random selection process has several characteristics:

• Algorithmic generation — the selection is produced by a software algorithm that generates genuinely random outcomes, not pseudo-random approximations.
• Current pool data — the selection is drawn from an accurate, up-to-date employee list that includes all eligible employees and excludes only those with a documented, policy-based exemption.
• Immutable audit trail — the system records the date and time of the selection, the complete pool at the time of selection, the algorithm used, and the employees selected. This record cannot be altered after the fact.
• No human intervention — the selection process does not involve any discretionary step where a person can influence the outcome.
• Documented substitution rules — where a selected employee is unavailable, the substitution process is itself random and documented.

Audit Trail Requirements

An audit trail is not a nice-to-have; it is a legal necessity. For each selection event, your records should include:

• The date and time the selection was generated
• The identity of the person or system that initiated the selection
• The complete list of employees in the pool at the time
• The selection criteria (percentage, number, site, department)
• The employees selected
• Any substitutions made and the reason for each
• The date and time each selected employee was notified and tested

This documentation should be produced automatically by your selection system, not assembled manually after the fact.

Is your random selection process truly defensible? Start a free trial of FairTest and experience genuinely random, fully auditable selection — with an immutable audit trail that stands up to any challenge.