The Statistics Were Real When You Liked Them

There is a revealing pattern in how political arguments handle evidence. A source is treated as authoritative for as long as it supports the preferred narrative. The moment it points the other way, the same source is suddenly unreliable, manipulated, or irrelevant. The British right has executed this shift with striking clarity over recent UK migration figures. The Office for National Statistics reports that long-term net migration stood at 171,000 in the year ending December 2025.[1] This was down from a revised 331,000 the previous year and a sharp fall from the peak of 944,000 in the year ending March 2023.[2] Gross immigration fell to around 813,000, a reduction of roughly 20 percent in a single year.[3] These are large movements. They occurred under Keir Starmer, building on visa restrictions initiated late in the prior Conservative government.

Many who spent years highlighting high immigration now reject the new numbers outright. The figures are dismissed as cooked, the methodology as suspect, or the drop as a political confection. The response is not measured scrutiny but outright disbelief that the data could show improvement on this metric. This reaction becomes self-undermining once the recent history of the same statistics is recalled.

The surge now called the Boriswave was documented primarily through official government data. After the post-Brexit points-based system took effect in 2021, ONS and Home Office releases showed net migration climbing rapidly, driven by non-EU inflows on work, study, and dependent visas. Record visa grants and annual net additions in the hundreds of thousands turned diffuse concerns into specific grievances. Graphs, threads, and commentary relied on these series to quantify the scale of change. The numbers supplied the concrete evidence. Without them, the narrative of unprecedented inflows would have rested on anecdote alone. Mainstream media also amplified those immigration numbers heavily. You did not know about them primarily because of social media accounts; you knew because the government released them and the BBC, along with every major newspaper, reported the releases in detail. To use the language of Walter Lippmann, your pseudo-environment on the story was framed almost entirely by these facts.[4]

The current stance therefore requires a distinction without a difference in the underlying instrument. The ONS used the same surveys, administrative records on visas and border crossings, and estimation methods to record both the rise and the subsequent fall. Nothing fundamental in the production of the data changed between the two periods. If the series was robust enough to establish a problem when numbers were rising, it remains the relevant measure when they decline. Rejecting it now while having built the earlier case upon it reveals a preference for the conclusion over consistent standards of evidence.

The more refined objection concedes a fall in the net figure but attributes it to British emigration rather than any reduction in foreign arrivals. On this account, inflows remain at or near record levels, and the headline improvement simply reflects natives leaving in large numbers. The data does not support this reading. Gross immigration declined substantially, with the largest drops among the non-EU work and study categories that had driven the earlier increase. Non-EU net migration itself fell from 511,000 to 350,000. British nationals recorded net emigration of 136,000, with 246,000 leaving and 110,000 arriving. Total emigration reached around 642,000. British leavers formed only a portion of the outflow; much of the emigration involved foreign nationals, including former students who arrived during the surge and departed after completing studies or as routes tightened. British emigration levels have remained broadly stable in recent years rather than spiking dramatically. The net reduction aligns far more closely with lower inflows than with any surge in departures. Claims that more foreigners are arriving than ever therefore conflict with the gross immigration series itself. The composition of the change points to fewer new arrivals overall and some return migration among those already here, not a British exodus masking continued high settlement.

A deeper objection still reaches beyond official aggregates to lived experience. Thomas Carlyle, in his 1839 essay Chartism, argued that statistics often obscure more than they reveal.[5] A judicious man uses them not to gain knowledge but to save himself from having ignorance foisted upon him. Where Carlyle could look with his own eyes – at the condition of the working classes, at patterns of thrift and saving – he found realities that contradicted the comforting tables. Savings-bank returns might show an increase while personal observation in specific communities revealed thrift in decline. For Carlyle, what one sees directly on the ground carries a weight that abstract numbers frequently lack. This premise has real force. Official statistics can lag, smooth over local variation, or miss qualitative shifts in social texture. Personal observation and anecdote remain essential checks on bureaucratic abstraction.

Yet the same premise is vulnerable to a powerful counter from Walter Lippmann. We do not simply see what is there; we see what our existing framework prepares us to notice and interpret. Lippmann’s Public Opinion (1922) shows how the ‘pseudo-environment’ we carry in our heads shapes which facts we register and how we weigh them. The same person who, in 2015, might have walked past a non-white couple in a provincial town and registered nothing unusual can, by 2026, interpret an identical sighting as immediate proof that immigration has not been controlled or that the government is lying about falling numbers. The observation has not changed, but the interpretive lens has. What feels like direct, unmediated evidence is often prior belief selecting and colouring the data of the senses. The same mechanism that once made rising official figures feel like confirmation now makes falling official figures feel like deception, while everyday sights are recruited to close the gap.

This selective application of evidence standards is a textbook illustration of confirmation bias. Daniel Kahneman describes in Thinking, Fast and Slow how people who accept a conclusion as true readily accept arguments that appear to support it, even when weak, while subjecting contrary information to far stricter scrutiny.[6] The intuitive, fast System 1 thinking constructs coherent stories that fit existing beliefs and resists disconfirmation. Jonathan Haidt develops a related account in The Righteous Mind.[7] Moral and political reasoning often operates as post-hoc justification rather than genuine inquiry. We tend to ask ‘Can I believe this?’ when evidence suits our priors and ‘Must I believe this?’ when it does not. In both frameworks, the credibility assigned to a source (or to one’s own eyes) tracks the congeniality of its output more than the consistency of its method.

The episode does not require anyone to credit the current government with a transformative success or to accept 171,000 as an ideal level. Policy on legal routes, enforcement, and the longer-term consequences of earlier high inflows remain legitimate subjects of debate. One can credit visa changes begun under the previous administration or argue that further reductions are necessary. Those positions stand without needing to discard the statistical trend when it moves in an inconvenient direction. Nor is this pattern confined to one side. Elements on the left have at times downplayed or reframed inconvenient data on integration outcomes, certain categories of crime, or the fiscal effects of high migration when those figures complicated preferred narratives. The temptation to sort evidence after the fact is widespread.

What the pattern ultimately reveals is that much political conviction sorts evidence according to whether it flatters settled beliefs. The same data series that established the scale of recent inflows now records a material decline. Treating one reading as decisive and the other as fraudulent, without a change in the underlying production of the numbers, substitutes preference for consistent reasoning. Personal observation matters; it can reveal what aggregates miss. But observation itself is not immune to the pseudo-reality Lippmann described. The statistics were treated as reliable when they aligned with the narrative of unchecked increase. They remain the same statistics when they show movement in the opposite direction. The task is to apply the same standards of scrutiny in both directions, or else the evidence we claim to follow simply follows us.

---

[1] ONS, ‘Long-term international migration, provisional: year ending December 2025’ (21 May 2026): https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/bulletins/longterminternationalmigrationprovisional/yearendingdecember2025.

[2] Electronic Immigration Network, ‘UK net migration nearly halves to 171,000 in year to December 2025, ONS says’: https://www.ein.org.uk/news/uk-net-migration-nearly-halves-171000-year-december-2025-ons-says.

[3] Statistica, ‘Long-term migration figures in the United Kingdom from 1964 to 2025’: https://www.statista.com/statistics/283287/net-migration-figures-of-the-united-kingdom-y-on-y/?srsltid=AfmBOorFhhF7ALmKlCOrmlvPK3vEolf70Vbx3ADS8w69VAqT3IaqwqLc.

[4] Walter Lippmann, Public Opinion (New York: Macmillan, 1922), p. 15.

[5] Thomas Carlyle, ‘Chartism’ (1839) in Essays on Politics and Society, ed. Chris R. Vanden Boscsche (Oakland, CA: University of California Press, 2022), pp. 61-130.

[6] Daniel Kahneman, Thinking Fast and Slow (New York and London: Penguin, 2011).

[7] Jonathan Haidt, The Righteous Mind (New York and London: 2012).