Proof by sampling in construction cases – sorting the wheat from the chaff

In dealing with a strike-out application in a professional negligence case, the Technology and Construction Court (TCC) in England has provided some useful comment on how to tackle disputes involving thousands of variations, and when and how ‘sampling’ can be used to assist the court.

29 January 2021

In dealing with a strike-out application in a professional negligence case, the Technology and Construction Court (TCC) in England has provided some useful comment on how to tackle disputes involving thousands of variations, and when and how ‘sampling’ can be used to assist the court.


Standard Life Assurance (Standard Life) entered into a building contract for a mixed retail and residential development in Newbury, Berkshire, with Costain as its main contractor. The contract price was £77.4 million, and Standard Life ultimately paid £146.4 million to settle the contractor’s final account. That amount included £28.4 million in respect of variations to the building contract (of which £25.8 million arose from contract administrator instructions (CAIs) and £2.6 million from confirmations of verbal instructions (CVIs)).

Standard Life subsequently brought a claim for professional negligence against six members of its design team, for a sum of £38.1 million. Standard Life claimed that the design team had carried out its work on the development negligently, issuing 3,600 instructions to Costain that consequently resulted in the final account increasing by over £38.1 million.  This was its primary case.


Standard Life’s particulars of claim did not set out all 3,600 instructions to Costain. It instead relied upon extrapolation from a selected sample of instructions, which it claimed it had analysed in detail. It argued that the court’s findings on these 122 samples, covering four areas of work, could be extrapolated in order to draw inferences on the remaining instructions. The selected sample related to circa £12.9 million. Standard Life argued that this was a reasonable and proportionate approach, and that requiring it to plead and prove every part of the claim would be disproportionate and absurd.

Three of the six defendants – BDP, Sutton Griffin Architects and Cundall (the Design Team) argued that the sampling was not appropriate. Firstly, they argued that, given the pleadings were limited to this specific sample (£12.9 million), Standard Life had failed to plead a case in relation to the balance of the claim (£25.2 million). They also argued the sample taken was not appropriate, as it targeted the high-value variations in order to maximise Standard Life’s claim. Finally, the Design Team took issue with the manner in which the case had been pled, claiming the pleading was ‘vague and incoherent’.

Strike-out application

Consequently, the Design Team brought applications to strike out parts of Standard Life’s claim against them, or alternatively for summary judgment dismissing those parts of its claim. The court has the power to strike out a claim if there are no reasonable grounds for bringing or defending a claim, or to give summary judgement if there is no real prospect of a party succeeding on a claim or defence.

The judge rejected the applications.  After considering the evidence and applying the principles, he found that the case to be met by the Design Team was clear enough. He noted that they had been able to work out the case against them sufficiently to be able to ‘pour elegant scorn’ upon it.

TCC comments on sampling

The judge examined the extrapolation claim and concluded that, within the ‘balance’ of Standard Life’s claim (i.e. the £25.2 million of claimed losses), it was likely that there was both ‘wheat’ as well as ‘chaff’.

Although the judge rejected the defendants’ application, he did agree that there may be an issue with the selection process for the samples. The judge therefore proposed issuing outline directions to refine the sampling process. These would help to ensure that the sample was valid, and also fairer for the defendants. He took the view that the sample of 122 variations was not enough for the required purposes and directed that there needed to be further sampling to enable the Design Team to have more detailed knowledge of Standard Life’s case on causation, to which they were entitled.

The draft directions allowed the three defendants to nominate between them 160 variations to make up the sample. In advance of this, Standard Life would have the opportunity to reduce its claim to the variations on which it genuinely thought it could rely. That would also incentivise Standard Life to dispose of its weaker claims, for fear that the defendants might otherwise nominate those as the samples to be used.

Finally, in considering Standard Life’s loss and expense claim, the judge was willing to accept that, despite it being a ‘global claim’, it was not objectionable just because the amounts claimed did not directly link to specific breaches of duty. The issue was in fact whether, and to what extent, Standard Life could discharge the burden of proof. As with the variations claim, the judge directed that further sampling should take place to provide greater clarity to the Design Team.

Points to take away

Although this case was heard in the English TCC, it should still be of interest to parties undertaking actions both north and south of the border. The judgment provides an interesting examination of the use of extrapolation from a sample of evidence as a means of establishing the full extent of a breach (or breaches) of duty. It also provides helpful commentary on sampling as a method to calculate quantum of loss; considering the appropriateness of such an approach and its potential limitations.

It should be noted that, where sampling is agreed or allowed, it will be open to the judge to prescribe or comment on the method to be used, and any sample set will need to be large enough to ensure reasonable conclusions can be drawn from the process. Therefore, although sampling is an option available to parties litigating a large dispute, considerable care should be taken at the outset to ensure the sample is: (i) of sufficient size; and (ii) is generally representative of the range of issues, covering different areas and/or types of work included in the claim.