When Does National Identity Matter?
Economic Context and the Conditional Activation of Immigration Attitudes
University of Glasgow
The Puzzle
National identity is one of the strongest predictors of restrictive immigration attitudes
A decade of BES panel data shows:
- However, the strength of this relationship is not constant
- Furthermore, the within-individual association fluctuates wave to wave
- It sometimes reverse sign, turning sharply negative only after 2022
- Yet national identity itself remains relatively stable
If the same level of national identification predicts markedly different immigration attitudes depending on when it is observed, the question is not whether identity matters, but when, why, and through what mechanisms.
The answer proposed here: economic conditions determine when identity becomes politically consequential
The Framework: Conditional Activation
National identity = stable predisposition
Economic conditions = situational trigger (Sniderman et al. 2004)
Q1: Which economic channel?
- Egotropic: personal insecurity activates nationalist exclusion (Scheve & Slaughter 2001)
- Sociotropic: collectively experienced conditions activate it (Dancygier & Donnelly 2013; Hopkins 2010)
Q2: Which identifiers respond?
- Threat amplification account: strong identifiers get reinforced (Shayo 2009; Sniderman et al. 2004)
- Attitude constraint account: weak/moderate identifiers have room to move (Converse 1964; Feldman 1988)
Key contribution
Existing work tests egotropic and sociotropic channels in isolation, and typically cross-sectionally.
We test both simultaneously, within the same persons, over 11 years, and separately for Britishness and Englishness.
Hypotheses
H1: Egotropic Activation
The identity–ATI relationship is stronger among individuals experiencing greater personal economic insecurity: lower income and higher perceived unemployment risk
H2: Sociotropic Activation
The identity–ATI relationship is stronger under adverse collectively experienced conditions: local area economic deterioration and national economic downturns
Two rival predictions for which identifiers are activated:
H3: Weak identifier mobilisation: activation concentrates among weaker/intermediate identifiers (Converse 1964; Feldman 1988)
H4: Strong identifier amplification: activation concentrates among stronger identifiers (Shayo 2009; Sniderman et al. 2004)
Warning
All hypotheses estimated separately for Britishness and Englishness following (Gidron & Shayo 2026) and on micro (egotropic) and meso and macro levels (sociotropic).
::::
Data and Design
British Election Study Internet Panel (2014–2025)
- 30 waves, ~86,000 unique respondents, ~453,000 person-wave observations
- 18 waves with immigration attitudes; both Britishness and Englishness across most waves
Three DVs: Self-placement on immigration levels (10-pt, main), economic ATI (7-pt), cultural ATI (7-pt)
Two key IVs: Britishness and Englishness, 1–7 continuous scale
Estimation strategy: Two-way fixed effects
\[y_{it} = \alpha_i + \lambda_t + \beta \cdot \text{Identity}_{it} + \gamma \cdot (\text{Identity}_{it} \times \text{Econ}_{it/t}) + \varepsilon_{it}\]
- \(\alpha_i\): absorbs all time-invariant individual differences
- \(\lambda_t\): absorbs all common temporal shocks
- Identification = within-person change only
Robustness: categorical re-specification, random effects, Mundlak correlated random effects (within/between decomposition)
Economic Variables
| Level | Variable | Source |
|---|---|---|
| Micro | Equivalised household income | BES survey |
| Micro | Perceived unemployment risk | BES survey |
| Meso | Local authority unemployment rate | ONS/APS |
| Meso | Claimant count rate | ONS benefits records |
| Meso | Median weekly earnings | ASHE |
| Macro | National unemployment rate | ONS |
| Macro | CPIH inflation | ONS |
| Macro | GDP per capita | ONS |
| Macro | Claimant count | ONS |
Meso variables matched by local authority and fieldwork year. Macro indicators merged at wave level; main effects absorbed by \(\lambda_t\), identification rests on interactions.
The Puzzle in the Data: Cross-Sectional Pattern
Both identities predict restriction throughout, Englishness consistently larger
A stable predisposition becoming more politically consequential in specific economic periods
The Puzzle in the Data: Within-Individual Pattern
The same person — more identified → more permissive for most of the panel, then sharply more restrictive after 2022.
This is what conditional activation tries to explain: economic timing is not a coincidence.
Result 1: Baseline Within-Individual Effects
Within individuals, increases in national identification predict more restrictive immigration preferences:
| Britishness | Englishness | |
|---|---|---|
| Main effect (β) | −0.042*** | −0.068*** |
| Consistent across all DVs? | ✓ | ✓ |
| Consistent across all specs? | ✓ | ✓ |
| Consistent across all robustness checks? | ✓ | ✓ |
Clustered SE at individual level. p < 0.001 throughout.
The Englishness–Britishness gap holds across all specifications, all DVs, all robustness checks, including categorical re-specification which shows a monotonically ordered gradient for both identities, with Englishness uniformly larger at every position
Result 2: The Egotropic Channel Fails
H1 not supported. Neither household income nor perceived unemployment risk conditions the identity–ATI relationship in any specification. Interaction terms ≈ 10⁻⁸, SE ≈ 10× larger.
But the null is theoretically informative and provides a further discriminating test:
Key auxiliary finding: Perceived unemployment risk has a small negative effect on national identification itself (β ≈ −0.006 to −0.008, p < 0.05)
When individuals feel personally economically insecure, they become marginally less nationally identified — the opposite of what status-compensation predicts (Shayo 2009)
Result 3: Testing H2 — The Sociotropic Channel
Claimant count panel: Clear Britishness–Englishness asymmetry
- Englishness slope: increasingly negative as claimant counts rise
- Categorical specifications confirm: effect concentrated among the weakest English identifiers
Local authority earnings panel: Both slopes become more negative as local earnings rise
Result 4: Testing H2 — The Sociotropic Channel (National Level)
The pattern confirms meso results at national scale: unemployment + inflation strengthen gradient; GDP prosperity compression parallels local earnings pattern. Englishness more broadly sensitive than Britishness.
Interpretive caveat — read before the figure
Macroeconomic conditions also move identification itself within persons (GDP raises both; unemployment reduces Englishness).
This means macro interactions may partly reflect compositional change in the identity distribution, not pure activation.
The meso results provide the more robust identification. Treat macro as confirmatory.
H3 vs H4: Which Identifiers Drive Activation?
| Level | H4 (strong identifiers) | H3 (weak/moderate identifiers) |
|---|---|---|
| Micro | — (no conditional effects) | — (no conditional effects) |
| Meso | ✗ | ✓ Claimant and income effects concentrated among weakest identifiers |
| Macro | Partial: effects not limited to strong identifiers | ✓ Effects strongest among weak/moderate, but extend more broadly than at meso level |
H3 receives stronger and more consistent support, particularly at the meso level. The attitude constraint mechanism operates most cleanly in local contexts, where cues are more diffuse. At the macro level, the potency of national economic shocks is sufficient to shift even strongly identified individuals — but H4 is not supported as a general account.
Three Broader Implications
1. Egotropic vs sociotropic mechanisms in identity politics
Personal insecurity does not activate exclusionary nationalism, shared economic conditions do. This speaks directly to the literature on left-behind places, regional economic grievances and economic voting (Lewis-Beck and Stegmaier, 2007; Jennings & Stoker 2017; Eatwell & Goodwin 2018). The political salience of national boundaries is sharpened by the collective economic fate of communities, not by individual economic biography.
2. Who gets activated matters
Economic deterioration expands the base of restriction by mobilising the weakly attached, not by intensifying the already-committed. The political consequences of economic crises depend not just on how bad conditions get, but on the distribution of national identification in the affected population.
3. Predispositions can be stable while their expression is not
Accounts of opinion stability (Kustov, Laaker & Reller 2021) and instability are both partially right, operating at different levels of the attitude system. Economic context is the switch that connects a stable predisposition to variable political consequences
Limitations and Next Steps
Current limitations:
- Single-country panel: generalisability to other contexts remains to be established
- ~11 macro observations constrain precision and preclude clean separation of activation from compositional change in the identity distribution
- Wave fixed effects absorb political framing — cannot distinguish structural economic activation from discursively mediated activation
Future directions:
- Multi-country panel replication (ESS; comparative European panels)
- Quasi-experimental designs exploiting localised economic shocks (plant closures, regional deindustrialisation)
- Supply-side variation: does political salience of immigration rhetoric condition the economic activation mechanism?
Conclusion
The question is no longer whether national identity predicts immigration preferences.
The question is under which economic conditions its attitudinal consequences become most pronounced, through which mechanisms, and for which identifiers.
Personal economic insecurity does not switch the identity–ATI link on. Collectively experienced economic conditions do and they do so mainly among those with the most attitudinal latitude, not the already-committed.
Thank you !
Contact & draft
Thales Lima
University of Glasgow
t.souza-lima.1@research.gla.ac.uk
Draft available on request
Northern EPOP Workshop · York · June 2026