Gill, R (On the Application Of) v Cabinet Office [No.3]

A public body must apply its published policy unless there are good reasons to depart from it. The policy must disclose the matters an affected person needs to make informed and meaningful representations. Minor changes to weighting or scoring do not necessarily require separate publication where the decision-maker has announced that the methodology will be reviewed and updated.

Where different criteria are designed for different decisions, applying one set to the wrong decision is not established merely by occasional use of similar terminology. A court reviewing expert or technical research must identify a serious logical or methodological error that goes to the heart of the decision and could make a substantive difference. The application for judicial review was dismissed.

The claimant challenged the Census Order 2020 because it did not provide a Sikh tick-box under the ethnic group question. He alleged that the ONS had failed to apply its published public-acceptability criterion, had applied an unpublished prioritisation policy, had acted inconsistently, and had relied unlawfully on the Kantar Report.

The challenge concerned the legality of the process leading to the Order, rather than the merits of including a Sikh tick-box. The court considered the distinction between criteria for census topics and question design, and criteria for prioritising response options.

1. Outcome. The claim for judicial review was dismissed. None of the four grounds was established.
2. Published policy. The ONS had distinguished between criteria for selecting topics and designing questions, and the prioritisation tool used to select tick-box response options. Read as a whole, the published material indicated that an updated version of the 2011 prioritisation tool would be used. Occasional references to “public acceptability” did not amount to a promise to apply the topic criterion to response options. The ONS had therefore applied its published policy and had not used a hidden or secret policy.
3. Changes to the prioritisation tool. The ONS had stated that the methodology would be reviewed and updated. Changes to organisation, weighting, wording and the use of a Red-Amber-Green assessment for acceptability, clarity and quality were minor and did not materially affect the information needed to make representations. There was no Lumba-style failure to publish or apply policy.
4. Consistency. The claimant’s comparisons with sexual-orientation questions did not involve like-for-like exercises. Topic and question-design testing addressed all respondents and the effect of a question, whereas response-option testing addressed the suitability of a particular tick-box for the relevant group. Qualitative focus-group evidence could not be directly compared with quantitative survey evidence.
5. Kantar Report. The report’s Amber rating for clarity was rationally based on limited identification with the proposed tick-box. Its assessment of confusion was not unlawfully narrow. The decision in Mandla v Dowell Lee did not require census respondents to identify themselves according to a legal definition of ethnic group; census responses were based on self-identification.
6. Relief. Had relief fallen to be considered, a quashing order would very likely have been refused because postponing the census would cause exceptionally serious detriment to good administration and substantial additional cost.

First-instance judicial review. Earlier related claims were dismissed or withdrawn, but those decisions were not the subject of determination in this judgment.