Psychologists Matt N. Williams of Massey University and colleagues report in the journal Frontiers in Social Psychology the results of an unusually intensive test of the idea that conspiracy beliefs isolate people. The team followed 839 adults in the United Kingdom, Australia, and New Zealand across 12 monthly survey waves spanning 11 months, from November 2023 to October 2024. Because the study was preregistered, with hypotheses and analysis plans locked in before data collection, the findings carry more evidential weight than the typical correlational snapshot.

The stakes of the question are considerable. Loneliness is a subjectively distressing gap between the social relationships people want and those they actually have, and it is linked in prior research to impaired cognition, poorer wellbeing, and heightened vigilance toward potential threats. Social connectedness, its more positive counterpart, captures the feeling of being comfortably bonded to a wider community beyond family and close friends. If conspiracy beliefs genuinely eroded either of these, the public health and social consequences could be substantial.

The theoretical picture is genuinely two-sided. On one hand, conspiracy theories cast believers as holders of special knowledge and non-believers as gullible

The methodological design of the study deserves particular attention, because cross-lagged panel research of this kind is far less common in this literature than simple one-time surveys. Most previous claims about conspiracy beliefs and social isolation rest on cross-sectional data, in which loneliness and conspiracy belief are measured at a single moment. Such designs cannot establish which variable, if either, drives the other, since people who feel isolated may simply also be more suspicious of mainstream narratives at that same point in time. By measuring both belief and social outcomes repeatedly across nearly a year, the researchers could examine whether changes in belief within a given person predicted subsequent changes in loneliness or connectedness one month later, and whether the reverse pathway held as well.

The specific analytical tool, the random intercept cross-lagged panel model, represents a relatively recent advance in longitudinal methodology. Traditional cross-lagged panel models have been criticized for conflating stable between-person differences with genuine within-person change. If some people are simply more distrustful and more isolated than others across the entire study period, a traditional model might mistakenly interpret that stable co-occurrence as evidence that one trait causes the other. The random intercept version separates each person’s stable baseline level from fluctuations around that baseline, allowing researchers to ask a sharper question: when a particular person believes more strongly in conspiracy theories than they usually do, do they subsequently become lonelier or less connected than they usually are? This within-person framing is conceptually important, because it mirrors the question that actually matters for individuals and for interventions aimed at them.

The distinction between two kinds of conspiracy-related belief also matters for interpreting the results. Belief in specific conspiracy theories refers to endorsement of particular claims, such as beliefs about the origins of specific events or the misconduct of particular institutions. Conspiracy mentality, by contrast, is a generalized disposition to suspect that major events are secretly orchestrated by powerful groups, regardless of the specific claim involved. The two are related but not identical, and they may have different social consequences. A person might entertain one specific theory without adopting a broader conspiratorial worldview, or might hold a diffuse distrust of official accounts without endorsing any particular narrative. The study tested both constructs against both social outcomes, and found that conspiracy mentality showed no evidence of effects on loneliness or connectedness in either direction.

The one positive finding that did emerge, a negative effect of specific conspiracy belief on connectedness in the preregistered analyses, is instructive precisely because of how it behaved under scrutiny. When the researchers re-analyzed the data using alternative but equally defensible statistical specifications, the effect did not hold. This pattern, in which a result appears under the preferred model but dissolves under reasonable variations, is a familiar cautionary tale in longitudinal psychology. It suggests the initial signal may have been an artifact of particular modeling assumptions rather than a stable feature of the data. Honest reporting of such fragility, which the authors provide, is a valuable corrective to a literature in which single-model findings are often presented with more confidence than they deserve.

The null results also need to be weighed against the study’s limitations, which the authors acknowledge candidly. One constraint is statistical power. Detecting small within-person effects in cross-lagged models generally requires large samples and many measurement occasions, and even a study of more than 800 participants across 12 waves has limits on the smallest effects it can reliably detect. Effects smaller than the study’s sensitivity threshold could exist and would have gone unnoticed. A second constraint is the possibility of time-varying confounds. Something else that changes over time, such as exposure to major news events, seasonal shifts in social activity, or personal circumstances like illness or job change, could in principle drive both belief and social outcomes simultaneously, and the models may not fully remove such influences.

A third limitation is perhaps the most substantively interesting: the sample reported low average levels of belief in conspiracy theories. Most participants, most of the time, endorsed conspiracy claims only weakly or not at all. This matters because the social costs of conspiracy belief, if they exist, may be concentrated among the most committed believers rather than distributed evenly across the population. A person who occasionally voices skepticism about one official account may face little social friction, whereas someone who reorganizes their life around a conspiratorial community, or who repeatedly challenges friends and family with contrarian claims, may experience genuine estrangement. If such effects are concentrated in a small subgroup, average within-person effects across the whole sample would be diluted toward zero, and the study’s design, however rigorous, would struggle to detect them.

The possibility of subgroup effects connects to a broader theme in research on stigmatized beliefs: heterogeneity in how people hold them. Two individuals endorsing the same theory may differ enormously in centrality, in how much the belief matters to their identity, in how often they discuss it, and in whether their social ties are drawn primarily from believing or non-believing circles. Someone embedded in a community of fellow believers may find that the belief strengthens their bonds, while someone surrounded by skeptics may find the same belief corrosive. The net population-level effect could be small not because conspiracy beliefs are socially inert, but because opposing processes in different people cancel out. Disentangling these possibilities would require research designs that deliberately oversample committed believers or that track the composition of believers’ social networks over time.

The time lag built into the design also deserves consideration. The study tested effects over one-month intervals, a reasonable choice given that social relationships change gradually. But some social consequences of expressing a conspiratorial view might unfold over much shorter periods, within a single conversation or a week of strained interactions, or over much longer ones, as accumulated friction gradually reshapes a friendship network over years. Cross-lagged effects at one-month lags can miss processes operating at other timescales, and the authors note that effects unfolding at different lags remain possible and would require further research to establish.

It is equally notable that the study found no evidence of the converse pathways, meaning that fluctuations in loneliness or connectedness did not predict subsequent changes in conspiracy belief. This addresses a complementary hypothesis that has circulated in both academic and popular discussion: that loneliness drives people into conspiratorial communities seeking belonging. The idea has intuitive appeal, and online spaces organized around shared distrust do seem to offer validation and identity to the socially disconnected. Yet within this sample, becoming lonelier or less connected in one month did not foreshadow stronger conspiracy beliefs the next. As with the forward pathways, this null result applies to average within-person effects over the tested lag, and it does not rule out that particular individuals, at particular moments of crisis or transition, turn to conspiratorial communities for connection.

The broader significance of the study lies in what it suggests about the social texture of everyday conspiracy belief. The dramatic narrative of the believer losing family and friends may describe a real phenomenon, but if so, it appears to be one that unfolds in specific circumstances rather than as an automatic consequence of holding heterodox views. Many people harbor doubts about official accounts without visible social penalty, perhaps because such beliefs are common enough to be unremarkable, or because believers moderate how and with whom they express them. The stigma attached to conspiracy endorsement, documented in prior work showing that believers are judged as less competent, moral, and warm, may operate selectively rather than universally.

For researchers, the study offers a template: preregistration, intensive longitudinal measurement, multiple model specifications, and transparent reporting of robustness. For the public, the takeaway is more modest but still meaningful. The assumption that conspiracy beliefs inevitably isolate their holders, while psychologically intuitive, is not well supported by the strongest longitudinal evidence to date, at least at the level of typical believers over monthly timescales. Whether deeper costs accrue to the most devoted believers, over longer horizons, or through mechanisms the present design could not capture remains an open empirical question, and one that future research will need to pursue with similarly rigorous methods.

Subject of Research: Longitudinal effects of conspiracy beliefs on loneliness and social connectedness

Article Title: How does belief in conspiracy theories affect loneliness and connectedness? A longitudinal investigation

Article References: Williams, M. N., Kerr, J. R., Marques, M. D., Abu-Obeid, H., & Hill, S. R. (2026). How does belief in conspiracy theories affect loneliness and connectedness? A longitudinal investigation. Frontiers in Social Psychology, 4, Article 1892137. https://doi.org/10.3389/frsps.2026.1892137

Image Credits: AI Generated

DOI: 10.3389/frsps.2026.1892137

Keywords: conspiracy theories, loneliness, social connectedness, longitudinal study, RI-CLPM, social psychology, conspiracy mentality, preregistration, panel study, interpersonal relationships, isolation, wellbeing