Who Doubts Vaccines in Türkiye? Trust and Education Outweigh Health in New Survey Analysis

Vaccine hesitancy has become one of the defining public health challenges of the post-pandemic era, and a new analysis from Türkiye offers one of the clearest portraits yet of who actually doubts vaccines — and why. Using data from the 2024 Turkish General Social Survey, a nationally representative social survey of adults, researcher Ömer Faruk Sönmez of Kırıkkale University and Université Paris-Saclay examined the sociodemographic, health-related, and attitudinal correlates of general vaccine hesitancy. The study, published in BMC Public Health, moves beyond the COVID-19-focused studies and convenience samples that have dominated the Turkish literature so far, asking instead whether people agree with a blunt statement: in general, vaccines do more harm than good.

The analytical foundation of the study is a cross-sectional secondary analysis of 1,600 adults who gave valid responses to the hesitancy item in the 2024 Turkish General Social Survey. Responses were recorded on a five-point agreement scale, from strong disagreement to strong agreement. The researcher treated the item in two complementary ways: descriptively, as a binary indicator of hesitancy, and analytically, as an ordinal outcome that preserves the graded intensity of doubt. Covariates were organized into three conceptual domains — sociodemographic and socioeconomic characteristics, health-related characteristics, and information, belief, and attitudinal factors. All analyses were unweighted, and the core statistical engine was ordinal logistic regression, a technique suited to outcomes that are ordered but not continuous.

Ordinal logistic regression estimates the odds of falling into a higher rather than a lower category of an ordered outcome, and its results are expressed as odds ratios with 95 percent confidence intervals. The modeling strategy proceeded in two stages. Univariable models first screened each covariate individually for an association with hesitancy. A multivariable model then adjusted simultaneously for all covariates, allowing the researcher to identify which factors retained independent associations after accounting for the others. The researcher also reported the Akaike information criterion, a measure that penalizes model complexity and helps compare the fit of competing models, and used variance inflation factors to check for multicollinearity — the statistical problem that arises when predictor variables overlap so heavily that their individual effects become impossible to separate.

The results paint a striking picture of hesitancy as a social and attitudinal phenomenon rather than a medical one. After adjustment, higher vaccine hesitancy was independently associated with being aged 35 to 44 years, with an adjusted odds ratio of 1.67 and a confidence interval of 1.14 to 2.44. Having only a primary school education raised the odds by a factor of 1.44, with a confidence interval spanning 1.07 to 1.93, and living in the Aegean region carried an adjusted odds ratio of 1.48, ranging from 1.05 to 2.09. Higher self-rated religiosity also predicted greater hesitancy. In other words, midlife adults with limited formal education, stronger religious self-identification, and residence in a specific geographic context formed the demographic core of vaccine skepticism in the sample.

The protective side of the ledger is equally informative. University education was associated with lower hesitancy, with an adjusted odds ratio of 0.68 and a confidence interval of 0.51 to 0.90, and a master’s degree was associated with a far steeper reduction, at an adjusted odds ratio of 0.40 with a confidence interval of 0.22 to 0.73. The single most powerful attitudinal buffer was confidence in scientists: respondents with high trust in the scientific community had an adjusted odds ratio of 0.52, with a confidence interval of 0.32 to 0.85, meaning their odds of scoring higher on the hesitancy scale were roughly half those of people lacking that confidence. The gradient across educational levels suggests that each additional tier of schooling compounds the effect, a dose-response pattern that strengthens the causal plausibility of the education-hesitancy link.

Perhaps the most surprising finding concerns health itself. Most health-related characteristics — the conditions one might expect to shape personal experience with medicine — were not independently associated with hesitancy once sociodemographic and attitudinal factors were taken into account. The main exception was arthritis or rheumatism, which retained its own association with higher hesitancy. The study also found that general vaccine hesitancy was not significantly associated with COVID-19 vaccine uptake, a decoupling that suggests people may accept a specific emergency vaccine while still harboring generalized doubts about immunization, or vice versa. For public health planners, this means that the population most likely to refuse vaccines cannot be reliably identified from medical records or chronic disease registries alone.

The methodological choices deserve scrutiny because they shape what the study can claim. By analyzing hesitancy as an ordinal outcome, the study respects the psychological reality that doubt comes in degrees — someone who slightly agrees that vaccines do more harm than good is not equivalent to someone who strongly agrees. The unweighted analysis is a limitation worth noting: because the survey data were not weighted to match the national population structure, the reported associations describe the analytic sample rather than precise national estimates. The cross-sectional design also means that all findings are correlational; the study can identify which factors travel together with hesitancy, but it cannot establish whether education, for example, causes lower hesitancy or whether both are downstream of some deeper variable such as cognitive style or social trust.

The regional signal from the Aegean adds a geographic layer that is easy to overlook in attitudinal research. Türkiye’s regions differ in economic structure, health service density, and cultural orientation, and the NUTS classification used in the study allows these contextual differences to enter the statistical model directly. Why the Aegean stands out is not answered by the data, but the finding echoes a broader lesson from vaccination research: hesitancy clusters geographically, and regional context can carry information that individual-level measures miss. Clustering matters practically because localized pockets of low acceptance can sustain disease outbreaks even in populations with acceptable overall coverage, as measles resurgences in several countries have demonstrated.

The prominence of religiosity and trust in scientists places this study squarely within the international literature on the behavioral and social drivers of vaccination, the framework that the World Health Organization’s Strategic Advisory Group of Experts on Immunization has promoted through instruments such as the BeSD framework. Across many countries, generalized trust — trust in institutions, in expertise, in the people who make and regulate vaccines — has proven to be a more consistent predictor of acceptance than any particular medical fact. The Turkish findings align with that pattern: confidence in scientists cut the odds of hesitancy nearly in half, while religiosity raised them, and both effects survived adjustment for education, age, region, and health status. Attitudes, not biology, appear to carry the explanatory weight.

What emerges is a coherent set of priorities for intervention. Communication campaigns that simply restate vaccine safety may be preaching to the converted if they reach audiences whose skepticism is rooted in low institutional trust or in religious framing that the messages never address. The strong educational gradient points to the long game — investment in education as structural immunization policy — while the centrality of trust in scientists suggests that trusted local messengers, including physicians who hold community credibility, may be more effective than centralized messaging. The study’s limitations, including its cross-sectional design, unweighted analyses, and reliance on a single general hesitancy item, mean the findings should be read as a map of associations rather than a causal blueprint. But as a nationally grounded snapshot of vaccine doubt in a country of over 85 million people, the analysis delivers a clear message: to understand hesitancy, look first at who people are, where they live, and whom they trust — not merely at what illnesses they carry.

Subject of Research: Sociodemographic, health-related, and attitudinal correlates of general vaccine hesitancy among adults in Türkiye

Article Title: Vaccine hesitancy in Türkiye: sociodemographic, health-related, and attitudinal correlates in the 2024 Turkish general social survey

Article References: Vaccine hesitancy in Türkiye: sociodemographic, health-related, and attitudinal correlates in the 2024 Turkish general social survey. (n.d.). https://doi.org/10.1186/s12889-026-29795-8

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29795-8

Keywords: vaccine hesitancy, Türkiye, Turkish General Social Survey, public health, trust in scientists, religiosity, education, ordinal logistic regression, BMC Public Health, immunization, health behavior, social determinants