Pleading claims by sampling and extrapolation: recent guidance given by the UK Court of Appeal

The disputes concerned a construction project, between the developer and employer (Developer) and the contract administrator and leader of the design team (Contract Administrator). The Developer claimed against the Contract Administrator and two other members of the design team^[1], in respect of payments the Developer had made in respect of the contractor’s claims for additional payment and for delay-related loss and expense resulting from 3,604 variations that had been approved by the Contract Administrator.

The Developer analysed and pleaded its case by way of particulars of 167 out of the 3,604 variations. The Developer alleged that, based on its analysis, the design team as a whole was culpably responsible for 83.1% of the value of the analysed variations, of which the Contract Administrator was responsible for 81.71%. The Developer then sought to apply those percentages to the remaining, unexamined variations. The Developer’s claims based on (i) the investigated variations and (ii) the unexamined variations, were referred to by the court respectively as the Detailed Claims and the Extrapolated Claims.

The Contract Administrator sought an order to strike out the Extrapolated Claims and also sought a reverse summary judgment in respect of the Extrapolated Claims, on the bases that the Extrapolated Claims as pleaded were an abuse of process, disclosed no reasonable grounds for bringing such a claim, and/or had no realistic prospect of success. 

The first instance court dismissed the Contract Administrator’s applications and instead directed the Developer to amend its pleadings. Among other things, the Developer was ordered to analyse 160 further variations to be selected separately by the defendants. Following the further investigation, the Detailed Claims increased in value from around £12.8 million to around £32.9 million and the Extrapolated Claims dropped in value from around £20.1 million to around £1.28 million. These significant shifts in value resulted from a combination of the following:

• The further variations analysed and assessed by the Developer were added to be part of the Detailed Clams and consequently were removed from the Extrapolated Claims.
• The further analysis resulted in a much lower proportion of the remaining unexamined variations being alleged to give rise to a claim for negligence or breach of contract against the Contract Administrator. The culpability percentage was reassessed by the Developer and the lower re-assessed percentage was then applied to the remaining unexamined variations.

The Contract Administrator appealed against the dismissal of its challenges to the claims as pleaded against it.

Court of Appeal judgment

Citing the first instance decisions of Amey LG Ltd v Cumbria County Council ^[2] and Imperial Chemical Industries Ltd v Merit Merrell Technology Ltd (No.2) ^[3], the Court of Appeal confirmed that, as a matter of pleading, extrapolated claims are permissible in principle.

The Court of Appeal held that the Extrapolated Claims were a proportionate way to address the 3,604 unexamined variations. The court considered proportionality to be a real issue in this case and estimated that, had the Extrapolated Claims been investigated and pleaded in the same manner as the Detailed Claims, the pleadings and accompanying schedules would have filled over 60 lever arch files.

Despite placing great emphasis on proportionality, the Court of Appeal stressed that extrapolated claims must be pleaded in a manner that defendants will be able to understand the case they have to meet, and must have a real prospect of success.

The Court of Appeal held in the Developer’s favour regarding these points, after finding as follows:

• The Developer’s claims were standard claims for negligence and for breach of contract. No novel points of law were involved.
• The Contract Administrator fully understood the Detailed Claims and the particular defaults alleged by the Developer, the particulars of which had been set out by the Developer in “almost interminable detail.”
• Both the investigated and the unexamined variations had been issued by the same people, on the same project and in the same circumstances. Also, the underlying themes behind the Detailed Claims cut across the entire project. Therefore, the inference of the Contract Administrator’s responsibility as a whole, based on the investigated variations, was at least reasonably arguable.
• The drawing of such an inference in this case did not, as the Contract Administrator submitted, involve comparing apples with oranges. In this regard, the Contract Administrator had made no attempts to distinguish the investigated variations from the remaining variations, other than stating the obvious that the former thoroughly had been investigated and the latter had not.
• The Contract Administrator had detailed knowledge of the variations given its role in the design team and in approving the variations.

The Court of Appeal also made the following remarks:

• There is nothing special or different about professional negligence actions to suggest that sampling and extrapolation could not be used for pleading such claims.
• Allowing the Extrapolated Claims to proceed would not give rise to unfair commercial pressure on the Contract Administrator. If the Developer had investigated and pleaded all 3,604 variations in detail, the Contract Administrator would have protested against the Developer frontloading the costs to put commercial pressure on the Contract Administrator to settle the claim.
• Whether or not a particular claim has been properly pleaded on an extrapolated basis is a question of degree, which turns on the specific facts of the particular case in question. Therefore, allowing pleading by sampling and extrapolation would not open the floodgates to undetailed or global claims being advanced by plaintiffs, at the expenses of defendants not knowing the case they had to meet or being held liable on bases they never understood.
• It is not sensible, cost-effective or proportionate to require parties in detailed commercial and construction cases to plead every last detail of their claims at the outset of the proceedings. It may be necessary in some cases to investigate “the grinding details” of claims, but such investigations should only ever be commensurate with the overriding objective of enabling the court to deal with cases justly and at proportionate cost.

For these reasons, the Court of Appeal dismissed the Contract Administrator’s appeal.

The Court of Appeal also endorsed the first instance judge’s exercise of his discretion in respect of the case management decisions. If a claim can be saved in some way, the Court of Appeal explained, the courts would much rather achieve that end than to strike out the claim.

Discussion

This decision confirms that pleading by sampling and extrapolation is permissible in appropriate cases, with a view to encouraging the production of pleadings in a more proportionate and cost-effective manner. This case also demonstrates that sampling and extrapolation can be a pragmatic approach for presenting claims in complex construction disputes, especially where significant amounts of defects or variations are involved.

However, none of this amounts to a license for potential plaintiffs to take a lazy or minimalist approach to pleading their claims. Instead, whether extrapolation is permitted will fall to be assessed on a case by case basis, having regard to the circumstances of the case, the type and magnitude of claims and other practical considerations. Relevant considerations include the size and value of the samples, and whether a discernible nexus exists between the samples and the extrapolated claims.

Proving extrapolated claims at trial

Moving on from the “pleading” issues discussed above, the reality is that Extrapolated claims can be difficult to prove at trial.

This is because extrapolation in essence involves the drawing of an inference from the results of the investigation into the samples, and the burden of establishing such an inference lies with the claimant. Respondents no doubt will scrutinise, among other things, whether the samples truly are representative, the method of their selection and the quality of the analysis of the selected samples.

In fact, in the two first instance decisions cited by the Court of Appeal, the extrapolated claims had failed at trial:

• Amey (2016). The problems with the extrapolated claims in Amey included: (i) the examined claims represented very little in monetary terms of the plaintiff’s overall claims. In a category of claims identified by the court, the extrapolated claims were about 76 times the size of the examined claims. (ii) The extrapolated claims were advanced solely on the basis of extrapolation. The court in Amey observed that the plaintiff had no system or records to identify the specific works that allegedly were defective. (iii) Ultimately, the extrapolated claims were rejected after the court found on the specific facts of the case that the sampling process was not statistically genuinely random and that the samples were infected by factors introducing biases.^[4]
• Imperial Chemical Industries (2017). As for Imperial Chemical Industries, which concerned allegedly defective welding, the plaintiff’s expert considered reports for 412 welds and calculated a defective rate of 38%. Then, the plaintiff sought to apply the 38% defective rate to all of the welds in the project of the order 28,000. However, the 412 welds were picked from around 1,800 examined welds^[5], meaning that the plaintiff’s expert had failed to consider over a thousand of the welds that had been examined. The defendant further contended that the 1,800 welds were picked for examination in the first place because they were determined by the plaintiff to be defective and had been rejected. After ruling that the samples were not representative of all the welds and so the extrapolated exercise therefore was flawed from the outset, the court in Imperial Chemical Industries rejected the extrapolated claims.

These cases highlight the importance of selecting a set of samples that is sufficiently large and which properly represents the pool of allegations as a whole.

In this regard, Lord Justice Birss clarified in Building Design Partnership that there is no general requirement that extrapolated claims must be based on randomly selected samples or supported by statistical confidence. However, statistically rigorous sampling in a proper case is likely to add weight to a case based on extrapolation.

Procedural considerations

For the above reasons, potential claimants carefully should consider and assess, before commencing proceedings, (a) the appropriateness of pleading claims on an extrapolated basis (before proceedings are commenced, if possible) and (b) the set of samples to be investigated. After proceedings are commenced, parties also may wish to discuss and agree – and seek directions if agreement cannot be reached – on the appropriate procedural directions and timetable for further investigations into and development of the extrapolated claims – for example, the utility and timing for having joint expert processes.

[1] For simplicity, this article focuses on the claims by the Developer against the Contract Administrator.

[2] [2016] EWHC 2856 (TCC).

[3] [2017] EWHC 1763 (TCC); 173 Con LR 137.

[4] While there is no general requirement to use random probability sampling for establishing extrapolated claims, the court explained, random sampling was required in Amey because of the manner in which the extrapolated claims were advanced.

[5] The exact number of examined welds was disputed.