In order to bring a claim for negligence or breach of contract, you must gather and present all the evidence of the wrongdoing. However, in situations involving a very large number of instances of wrongdoing, it might not always be practicable, economical or proportionate to present every single one.

These situations crop up relatively often in construction disputes, so we should look at the way a court or arbitration tribunal might consider a case based on sampling. Here, the claiming party presents only a sample of those alleged breaches and extrapolates the sample to create its whole claim.

There are obvious objections: how is the sample chosen? Is it big enough to support extrapolation? How do we know that it is representative of the whole, in terms of the breaches complained of? How do we know that the other instances (outside of the sample) even exist?      

Also as a matter of litigation due process, a defending party will argue that sampling is an abuse of process because the defending party cannot know the case against it; or that the burden of proof (relating to the non-sampled instances) has been reversed, so that the defending party has somehow to prove that those other instances do not exist or are not as serious as the sample extrapolation would indicate.   

These are the same type of objections we have looked at before in connection with “global claims” (see Gulf Construction, July 2016). This is a shortcut presentation of a claim based on a large number of events causing delay and disruption. In a global claim, the claiming party presents the events and the impacts of them on an aggregate basis, that is, without making individual cause-and-effect connections between each event and its cost. There is some common law judicial support for global claims, but subject to strict conditions.

Support for sampling has been boosted by the focus in courts and in arbitration on cost-efficient disposal of disputes. As with global claims, the task is to find the right line between economy and doing justice between the parties.

This task was illustrated recently in a dispute about a mixed-use development (Standard Life Assurance Ltd vs Gleeds and Others). The final account for the works was settled at £146 million ($203 million); almost double the original contract price of £77 million. The developer then brought an action against its design team to recover £38 million, which the developer said was excess cost arising from 3,600 written and verbal instructions given negligently by the design team.

The developer’s claim was based on giving full details of just 122 instructions, presenting these as a representative sample of the 3,600 instructions. They contended that this is a reasonable and proportionate approach. The design team objected to this, both as a matter of litigation due process (using defences similar to those mentioned above) and alleging that the sample was skewed to the high-value instructions, and was not representative. They made an application to strike out the action. 

The judge did not strike out the action but gave directions and made some interesting comments on proof by sampling. He agreed that sampling was appropriate but that there were valid concerns about the size and selection of the sample presented by the developer. He ordered additional sampling to be done, allowing the defendants to choose 160 instructions for examination. However, before that selection was to be made, he allowed the developer to withdraw any instructions it so wished. The logic was that the developer would withdraw its weaker items, to remove the danger of the design team choosing them for the sample.     

That looks like a neat solution to the issue of sample selection. It might be said it put too much control into the hands of the defending party but equally, that loss of control might be a fair exchange for being allowed to use sampling.

It still leaves the question of whether a sample can be representative of the whole claim. This is maybe easier to demonstrate in defects claims, where systemic poor practice can create a pattern of the same type of defects; and in my view, such a claim is not necessarily a global claim. But in a negligence action such as this one, it looks more difficult to show the common thread needed to justify the extrapolation. The judge described this as a global claim because it did not link individual alleged breaches to a specific loss. However, he ordered additional sampling, to allow the defendants to better understand that link.

Sampling is a viable option if the circumstances justify it, but we have seen here that the claiming party might not keep control of the process. 

* Stuart Jordan is a partner in the Global Projects group of Baker Botts, a leading international law firm. Jordan’s practice focuses on the oil, gas, power, transport, petrochemical, nuclear and construction industries. He has extensive experience in the Middle East, Russia and the UK.