A random selection – Choosing samples for inspection

by Michael Ogus

Experts are frequently asked to provide opinions on projects where there are defects in multiple locations. In these circumstances, it is often not feasible to inspect every location and sampling is proposed as a reasonable basis for assessment.

The advantages of sampling are clear. The parties can quickly clarify the issues and quantum of the defects without having to inspect in every location. Appropriate sampling helps to limit the costs of expert evidence to a reasonable level, as required by the court in pre-action proceedings.

The process by which sample locations are selected, however, is not without its pitfalls.

A selection process based only on expediency or practical criteria, such as avoiding disruption to occupants; choosing only locations where access is straightforward on the grounds of safety or cost; or only opening-up parts of the works that can be reinstated with minimum time, cost and damage, may result in a range of samples that will not allow the court fully to understand the issues.

A selection process based purely on these criteria is unlikely to persuade the court that the range of selected samples is adequate. For example, Experts often encounter situations where details that should be identical have been constructed by different teams or different sub-contractors, leading to variations in workmanship. Such variations may be missed if too few samples are taken.

Whilst practical aspects must be considered, it is important to remember that the overriding purpose of expert evidence is to assist the court.

Guidance was provided by the court in the case of Amey v Cumbria County Council.[1] Here the contractor had repaired and maintained the county’s road network over a seven-year period. The area of roads repaired during that time ran into several million square metres with instructions running into the tens of thousands. The contract was terminated and Amey started proceedings to recover sums deducted from a payment. Cumbria counterclaimed alleging that some of Amey’s work was defective.

The Council relied on two types of statistical sampling; probability sampling and non-probability sampling. The Court considered whether this was sufficiently random and free from bias to be relied upon.

HHJ Davis accepted that it was impractical to take core samples from every location where defects were alleged because of the sheer number and geographical spread of the locations. Also, some of the earliest repairs could not be located because of subsequent re-patching or lost records. The judge agreed that the council could in principle rely on evidence gathered from sampling.

“I accept Cumbria’s primary submission that it would have been quite impracticable for it to have visually inspected and/or undertaken core testing to all of the patches laid by Amey over the duration of the contract …”

He continued –

“… there is no principle of law nor a statistical theory that a claim or proposition can only be established by statistically random sampling.I accept that it is perfectly open to a claimant to seek to establish a claim by reference to representative sampling…

However, he went on to say that as some of the samples were not genuinely random it was not possible to extrapolate the percentage of defective work claimed.

The judge was also critical of some of the aspects of sampling that Cumbria and its experts relied on. He pointed out that the sampling had been conducted over a long period and that variations in the age of the patching had not been adequately considered. Sample locations were selected using GPS. This produced a bias towards newer samples because works carried out before the introduction of GPS could not be located. Inadequate consideration was given to the size and composition of the sample that was tested or to variables, such as weather conditions at the time of the works, drainage and foundation issues and geographical area. Log sheets showed that insufficient time had been allocated to inspect, record and photograph the defects. The judge commented that this time pressure may have led to errors in the process. There was a concern about the quality of the inspections which were described in some instances as “perfunctory” and contained a number of assumptions. Because the inspection sheets were poorly designed and compiled under considerable time pressure it was doubtful whether the experts or court could rely on them. HHJ Davis concluded –

“… in my view Cumbria has failed to demonstrate that the sampling exercise undertaken on its behalf in this case is a sufficiently reliable exercise to justify the court in making the finding as against Amey …”

So, the lessons are clear. Whilst the courts are prepared to accept that sampling can be used as the basis to advance a claim, care must be taken at the outset to ensure the selection process and methodology is given proper consideration and is sufficiently robust, random, widespread and free of perceived bias to withstand judicial scrutiny. Wherever possible, the sampling exercise should be agreed between the parties to assist in overcoming later objections.

[1] Amey LG Ltd v Cumbria CC [2016] EWHC 2856 (TCC) (11 November 2016)