Validation of the behavioural and social drivers of influenza vaccine survey in Australia: Cognitive interviews and psychometric evaluation

ABSTRACT

Seasonal influenza vaccination reduces the risk of contracting influenza. In Australia, vaccination coverage is consistently low among the most at-risk groups eligible for the free influenza vaccine. To establish an annual national monitoring of the behavioral and social drivers of influenza vaccination, we validated the World Health Organization’s Behavioral and Social Drivers of Influenza Vaccination survey. Ours is the first study reporting the validation results of the newly developed survey tool. Two validation stages occurred: first, cognitive interviews and survey item revisions based on feedback; second, collecting quantitative data, performing psychometric analysis, including factor analysis, and assessing the association between survey items and influenza vaccination intention using polychoric correlation and a random forest model to determine each item’s importance. We modified nine of the 31 survey items based on cognitive interviews with 16 adults. Based on the psychometric evaluation of 2055 respondents, we identified and validated five factors that underlie the survey structure and aligned well with the four domains of the BeSD framework: Thinking and Feeling, Social Processes, Motivation, and Practical Issues. Each item correlated with the intention to receive influenza vaccine, particularly those related to perceived vaccine importance and social responsibility. We recommend further validation in various contexts, especially to assess the survey’s predictive validity regarding actual vaccination uptake.

Background

Influenza presents a significant public health challenge, leading to considerable morbidity and mortality annually.¹ It is estimated that unvaccinated individuals have a higher risk of seasonal influenza infection, with nearly 1 in 5 unvaccinated children and 1 in 10 unvaccinated adults estimated to contract the infection each year.² Based on modeling, it has been estimated that a vaccine with 40% effectiveness could avert a mean (SD) of 32.9% (0.9%) of seasonal influenza burden under high transmission and 41.5% (3.4%) under low transmission.³ In Australia, influenza is characterized by seasonal patterns and the potential for severe complications, especially among vulnerable populations such as adults aged 65 and over, children aged 5 and under, pregnant people, individuals with chronic health conditions, as well as Aboriginal populations and Torres Strait Islanders.⁴ Although the influenza vaccine remains the most effective intervention to reduce the morbidity and mortality of the disease¹ and is provided free of charge to these groups in Australia, vaccination coverage is consistently low among these target populations.⁵

Understanding the underlying behavioral and social drivers of influenza vaccine uptake in individuals is essential for developing effective interventions to increase vaccination coverage.^6,7 The World Health Organization (WHO) recommends that all countries regularly collect quantitative and qualitative data using Behavioral and Social Drivers of Vaccination (BeSD) surveys and priority indicators, focusing on areas and sub-populations that have gaps in vaccination coverage and inequities.⁸ The BeSD survey is a globally standardized and validated set of questions designed to measure modifiable beliefs and experiences related to vaccination measured within four domains: thinking and feeling about vaccination, social processes, motivation, and practical issues. Some studies have used individuals’ intentions, willingness, and hesitancy to receive vaccinations as outcome indicators,⁹ particularly when it is not feasible to collect data on current vaccination status. These indicators are part of the motivation domain in the BeSD.⁸

In this context, WHO has also developed Behavioral and Social Drivers of Influenza Vaccination survey tools.⁹ These tools were originally based on the BeSD tools for childhood and COVID-19 vaccination.⁸ The survey was developed by an expert group, informed by a literature review and prior studies of BeSD tools in different contexts,¹⁰ with the questions subsequently adapted to assess influenza vaccination.

The current study is part of a larger research project using mixed methods aimed at providing a foundation for annual national monitoring of the behavioral and social drivers of influenza vaccination. We followed the World Health Organization’s recommendation to assess the new and adapted BeSD survey tool for influenza vaccination.¹¹ To establish content validity and determine item clarity, we conducted cognitive interviews to refine the survey questions for the Australian context. A modified survey was delivered in the field. From the responses, we performed a psychometric analysis to evaluate the survey items in Australia. We did not assess confirmed influenza vaccination uptake via the Australian Immunization Register (AIR) linkage or follow-up because restricted access to individual-level AIR records made linkage infeasible in this online cross-sectional survey; therefore, criterion validity was assessed using intention to receive influenza vaccination.

To our knowledge, this is the first study reporting the validation results on the newly developed survey.

Methods

This paper describes two stages of validation: cognitive interviews and psychometric evaluation of the survey items. First, we conducted cognitive interviews and revised the survey items based on the feedback. Subsequently, we collected quantitative data using the revised items and performed psychometric analysis.

Cognitive interviews

We conducted cognitive interviews to pre-test survey questions and ensure that items generated the intended information and that response options were appropriate. This allowed researchers to adjust and refine the draft questions ahead of psychometric evaluation. Cognitive interviews sought to address issues with:

• Language – how people interpret and understand words or phrases.

• Inclusion – whether certain concepts are considered within the intended scope of the question.

• Temporal patterns – the time period that applies to a question.

• Computation – mental arithmetic required to answer a question.

Sample and recruitment

Adults from across Australia were recruited through an external agency. An online advertisement was distributed to their panel, and those who agreed to participate provided written consent and arranged a time for an online interview with a researcher. The 16 participants were adults aged 18 y and over, recruited from across Australia through an external panel provider, with stratification in demographic background, including age, gender, geographic location (including multiple states and both metropolitan and regional areas). The sample was purposively diverse to capture a range of perspectives relevant to the target population of the survey (Table 1).

Table 1.

Summary of participant characteristics (n = 16).

Characteristic	Participants in cognitive interviews (n)	Participants in the survey (n (%))
Age group
18–24	2	51 (2.48)
25–34	2	324 (15.77)
35–44	3	417 (20.29)
45–54	3	411 (20.00)
55–64	2	368 (17.91)
65+	4	484 (23.55)
Location by State
New South Wales	2	656 (31.92)
Victoria	3	588 (28.61)
Queensland	4	390 (18.98)
Western Australia	2	159 (7.74)
South Australia	1	178 (8.66)
Tasmania	2	50 (2.43)
Australian Capital Territory	2	26 (1.27)
Northern Territory	0	8 (0.39)
Type of area
Metropolitan	12	1623 (78.98)
Regional	4	318 (15.47)
Rural	0	104 (5.06)
Remote	0	10 (0.49)
Gender
Man	8	980 (47.69)
Woman	8	1070 (52.07)
Non-binary	0	4 (0.19)
Prefer not to say	0	1 (0.05)

Data collection and measures

During the cognitive interview, a single trained researcher in qualitative methods and cognitive interviewing (MCE) read aloud each survey question, which was also presented visually to the participant via a shared screen. The researcher recorded the interviewees’ responses to the survey question before asking them between one and three questions designed to ascertain how they understood the meaning of the question and how they interpreted and answered it. Some examples of common cognitive interview questions included: “How did you decide on your answer to this question?” and “Did the response options fit with the answers you wanted to give?” For some questions, a visual analogue scale was employed to help quantify the participant’s responses, as shown in Figure 1.

Figure 1.

Visual analogue scale presented to participants.⁸.

Data analysis

The researcher took extensive notes during the cognitive interviews. Following the completion of each interview, the interviewer recorded field notes regarding how the interview progressed and recorded any issues or observations related to the performance of the survey items. Responses and notes relating to each question were combined into a spreadsheet organized by item. Members of the research team engaged in regular meetings to discuss findings as they emerged. Careful inspection of each item and its associated notes revealed issues that were either (1) repeatedly observed across participants or (2) affected few participants but still threatened validity. Once cognitive interviews were completed, researchers collectively decided how to improve the validity of survey items for use in the Australian context.

Psychometric evaluation of the survey items

Sample and recruitment

Participants were recruited from an external panel provider to create a sample of 2,055 Australian adults over 18 y of age. The sample aimed to reflect proportions in the national population based on age, gender, and location. Detailed sampling and recruitment procedures for this national survey have been reported elsewhere¹² and are summarized briefly here for completeness.

Data collection and measures

Data were collected in March 2024 using an online survey. The WHO’s Behavioral and Social Drivers of Vaccination survey tool, adapted for influenza vaccination, was administered. Annex 1 lists the full set of items in the survey.

Data analysis

We followed the psychometric approach used by WHO in the validation of the BeSD tools for childhood and COVID‑19 vaccination, as described in Background paper on behavioral and social drivers of vaccine uptake.¹³ SPSS version 29 was used to generate frequencies of the survey items and perform exploratory factor analysis (EFA) to explore the structure of the newly developed survey. Confirmatory factor analysis (CFA) was also conducted with an oblique rotation (Promax) on the set of survey items using SPSS Amos v29. Promax rotation was selected based on the a priori assumption that the underlying factors are correlated rather than orthogonal. This decision was informed by the psychometric validation of the BeSD tools for childhood and COVID-19 vaccination, which demonstrated correlations among the underlying constructs. The names of the underlying factors were refined and agreed upon through a structured discussion and consensus between the co-authors, ensuring their consistency with the theoretical construct of BeSD. To assess criterion validity, the bivariate correlation was conducted using the polychoric correlation command on Stata SE18. This assessed the association between the survey items and the intention to receive influenza vaccine.

A random forest model was computed using the ‘forest’ package and command in Stata SE18 to estimate the importance of each item in predicting the intention to receive the influenza vaccine within the full set of items. Random forest is a machine-learning algorithm for identifying which predictors are highly predictive of a dependent variable. The syntax for the ‘rforest’ command was obtained from Schonlau et al.¹⁴

The psychometric analysis included 21 of the 31 survey items (Table 2). The item for vaccination uptake/status (vs.1), items with multiple response options (SP7, P02, P04, P08, P10) and those not applicable to all participants (vs.2, vs.3, vs.3b, and P09) were excluded. SP8 (decision autonomy) was excluded from polychoric correlation and random forest model because it is categorical variable. In addition, for items that included a “not sure” response option, responses were dichotomized, with “Yes” coded as one category and “No/Not sure” combined as the comparison category.

Table 2.

Bivariate polychoric correlations with an intention to receive influenza vaccine.

Results

Cognitive interviews

A total of 16 adults from across Australia were recruited to participate in interviews. Table 1 summarizes the characteristics of participants in the cognitive interviews.

Changes made to the survey items in Australia

We made changes to nine survey items based on cognitive interviews [see Additional file 1]. One survey item (SP5-B: what types of communication would you trust to give you accurate information about the influenza vaccine?) was removed because it did not adequately capture trust in communication as conceptualized in the survey. Five items were rephrased for various reasons, including being not applicable to the Australian context (D7: education background), allowing answers from both those who received and those who had never received an influenza vaccine (vs.2: place of receiving the vaccine), ambiguity in interpretations (vs.3: payment made), inconsistencies with previous questions (SP6: recommendation from health worker), and unclear temporal references (M1: intention to get the vaccine).

We also added response options to three questions. We added two response options (Not sure and Not applicable) to SP3B (religious leaders’ influence) to better accommodate the Australian context where having no religion is common. We also added one response option (Not sure) to SP3 (community leaders’ influence) and one response option (The vaccine is not available) to P6 (Factors that make it hard for you to get an influenza vaccine) to reflect recurring participant suggestions.

Psychometric evaluation

Participant characteristics summary

Following the cognitive interviews, the modified survey questions were answered by 2,055 Australian adults. Detailed sampling procedures and a comprehensive description of the study population have been reported previously.¹²

The largest group of participants was from metropolitan areas (79%), followed by those in regional areas (15.5%). Most of the participants were from New South Wales (31.9%), followed by Victoria (28.6%) and Queensland (19%). About 52.1% of the participants were female, 47.7% were male, and 0.2% identified as non-binary. Participants’ age ranged from 18 to 92 y (mean = 51.4, SD = 16). About one-third (33.7%) held a bachelor’s degree, followed by those with a trade certificate/apprenticeship and high school or equivalent (23.7% and 22.8%, respectively).

Demographic information also allowed identification of several higher-risk groups, including participants aged 65 y and over (23.6%), Aboriginal and Torres Strait Islander participants (1.3%), pregnant participants (0.9%), and participants with chronic health conditions (36.5%).¹² Information on whether participants were parents or carers of children aged 5 y and under was not collected. Findings were not stratified by these higher-risk groups in the present analysis, as subgroup comparisons were outside the scope of this study and are reported elsewhere.¹²

The frequency distribution of the survey items is summarized [See Additional file 2 for a summary of the distribution of survey items].

Criterion validity

Bivariate associations

Because all items are either ordinal or binary, polychoric correlations were used to assess the correlations of the survey items with the intention to receive influenza vaccine. All items are either ordinal or binary. As shown in Table 2, the correlation between the items with the intention to receive the influenza vaccine ranged from low (r = 0.33) for TF2 (Perceived influenza severity) to high (r = 0.86) for SP1 (social responsibility).

Multivariate associations and variable importance

A random forest model was performed to identify the importance of each item in predicting the intention to receive influenza vaccine within the full set of items. The random forest classification model used 1,000 decision trees and considered five randomly selected predictors at each split. By combining the predictions from these trees, the model helped identify which items were most informative for distinguishing vaccination intention. Variable-importance estimates from the random forest model indicated that TF1 (perceived influenza risk) contributed most strongly to the classification of respondents’ influenza vaccination intention. TF1 was therefore the most informative indicator for differentiating respondents who intended to be vaccinated from those who did not intend to vaccinate or were unsure. Predictors ranked next in descending importance were community leader norms (SP4), confidence in vaccine safety (TF7), and confidence in health workers (TF8) (Figure 2).

Figure 2.

Importance of items in predicting intention to receive the influenza vaccine.

Structure and model fit of the survey items

Exploratory factor analysis

Exploratory Factor Analysis (EFA) was used to identify underlying constructs and to examine how survey items grouped together and loaded on these constructs. Items were considered to load on a factor when they showed a strong association with it, indicating that they contributed meaningfully to measuring the same underlying concept. Five interpretable factors (i.e. constructs) with Eigenvalues greater than one were identified with oblique (Promax) rotation. Results from the 5-factor solution are shown in Table 3. Factor loadings reflect how strongly an item is correlated with a particular construct.

Table 3.

Exploratory factor analysis results for item loadings in the five-factor structure.

The first factor (Beliefs) was associated with nine items, including five (TF4-TF8), that were also categorized as Thoughts and Feelings; the wanting a flu vaccine (M1) item was considered to be a Motivation item, and three (SP1-SP3) were also categorized as Social Processes items. The second factor, feasibility and accessibility of getting influenza vaccine (Access), was associated with three items (P05-P07), which were considered Practical Issues. The third factor (the connection to community and resources) was correlated with practical knowledge of the influenza vaccine (TF3) and locations for getting the vaccine (P03), in addition to the influence of community and religious leaders (SP4 and SP5). Two items (TF1 and TF2) were loaded on the fourth factor (Perceived threat) of the disease. The fifth factor (Prompts) loaded three items, including two items related to receiving recommendations and reminders from the healthcare system (SP6 and P0, respectively), in addition to one item (SP8) related to the respondent’s autonomy to make a decision about vaccination.

Confirmatory factor analysis (CFA)

CFA was conducted by imposing restrictions on factor loadings to ensure model fit, indicating that most items were associated with only one underlying factor. CFA for a 4-factor model with the hypothesized structure (i.e., the BeSD framework with thinking and feeling, social processes, motivation, and practical issues) was not performed because this model did not converge in the validation of the original BeSD.¹² We performed CFA to test the 5-factor model suggested by the EFA.

Table 4 summarizes the inter‑correlations among the latent constructs identified in the exploratory factor analysis. Overall, the correlations are moderate and of similar magnitude, indicating that the constructs are related yet distinct. Consistent with our a priori choice of Promax rotation, the observed inter-factor correlations suggest that the constructs are not orthogonal. These correlations are presented to characterize the overall pattern of associations among constructs, rather than to imply meaningful differences in strength between specific pairs.

Table 4.

Inter-correlations among the five constructs identified in the exploratory factor analysis.

	Beliefs	Access	Connection to community and resources	Perceived threat	Prompts
Beliefs	1
Access	0.514	1
Connection to community and resources	0.498	0.466	1
Perceived threat	0.501	0.102	0.217	1
Prompts	0.518	0.251	0.185	0.425	1

CFA showed the best goodness of fit for the survey items with five factors (Chi-square (χ2) = 2939.3, degrees of freedom (df) = 179, p-value < .001, Root Mean Square Error of Approximation (RMSEA) = 0.087, Comparative Fit Index (CFI) = 0.855, Tucker-Lewis Index (TLI) = 0.830). Table 5 shows that all factor loadings were significant at p < .001 except SPels8 (the main decision maker for the influenza vaccine) which did not load significantly on Factor 5 (prompts) (r = 0.032).

Table 5.

Standardized confirmatory factor analysis loadings for the five-factor model.

Discussion

The cognitive interviews and the psychometric evaluation provide support for the content validity and overall construct validity of the Behavioral and Social Drivers of the Influenza Vaccination survey in Australia.

Findings from cognitive interviews helped improve the validity of the survey items within the Australian context by enhancing their clarity and relevance. This highlights the need to adapt survey questions for specific cultural, social, and health system contexts before conducting the survey on a larger sample in order to enhance the generalizability of the survey’s findings. For instance, we added “Not sure” or “Not applicable” to the question about religious leaders. In responding to this question, participants often found a forced “yes” or “no” choice challenging since about 39% of Australians have no religion.¹⁵ In general, the BeSD development group avoided having a “no opinion” response option because it often does not improve response quality, may introduce response bias in some population groups, and complicates analysis.¹⁶ However, it may be necessary in contexts where forcing a substantive response in such cases may increase measurement error because respondents lack direct exposure, relevance, or a well‑formed attitude. For example, Elkjær and Wlezien¹⁷ in their survey experiment found that providing a “Don’t Know’’ option enhanced respondents’ confidence in their answers for issues of low salience. When such options are omitted, researchers risk collecting forced responses that carry greater uncertainty, with implications for the accuracy of opinion estimates among low-information subgroups. We opted to add no‑opinion options to these questions for the Australian population.

From the psychometric evaluation, we identified and validated five factors that underlie the survey structure. These include individual and social beliefs about the vaccine, access to the vaccine, connection to community and resources about the vaccine, perceived threat from the disease, and prompts to get the vaccine. They align well with the domains of the BeSD framework: Thinking and Feeling, Social Processes, Motivation, and Practical Issues.

The findings also demonstrated that each item correlated with the intention to receive influenza vaccine, particularly those related to perceived vaccine importance and social responsibility. Furthermore, the findings revealed the importance of each item in predicting the intention to receive influenza vaccine within the full set of items.

The social responsibility item was newly developed for this survey. WHO added it to measure the sense of responsibility to vaccinate as a pro-social motive.¹¹ A previous study explained the role of “collective responsibility,” which refers to an individual’s willingness to protect others, in vaccination behavior.¹⁸ Interestingly, this item performed well during the cognitive interviews and in the psychometric analysis, supporting its value as an addition to the survey tool.

Only one item in the survey, the SP8 (autonomy of decision-making about influenza vaccine) was not significantly correlated with its underlying factor (prompts) in the CFA. This finding was similar in the BeSD for childhood vaccination testing.¹³ It therefore does not fit with the scale, but we recommend retaining this question to provide descriptive information about the extent to which an adult is autonomous in decision-making, particularly in contexts where gender issues intersect with vaccination decision-making. For example, a study in India found that most pregnant women were not the primary decision-makers regarding pregnancy-related healthcare, including tetanus vaccination.¹⁹ Similarly, a study in Kenya found that pregnant and lactating women expressed a desire to make their own maternal vaccination decisions, although partners, family members, peers, and healthcare providers remained important influences.²⁰ Furthermore, the item may support the decisions about which groups should be targeted with recommendations in information campaigns. We also recommend validating the other versions of this survey in different countries.

Strengths and limitations

This study has several strengths, including its rigorous mixed-method design that combines cognitive interviews with large-scale psychometric analysis. Our results provide strong and informative evidence supporting the validity of the tool, and they are highly consistent with the structure and domains of previously validated BeSD instruments used in childhood and COVID-19 vaccination. This is particularly important as the influenza BeSD tool builds directly on these earlier validated tools, adapting them to a new context while maintaining conceptual alignment. To our knowledge, this study represents the first comprehensive psychometric validation of this newly developed influenza-specific tool, offering a significant contribution to the literature.

However, caution is needed with respect to the representativeness of the sample, which may limit generalizability beyond the study context. We were not able to canvas all population groups in Australia. We surveyed very few First Nations people. Moreover, people from culturally and linguistically diverse backgrounds would not have been able to take part in the cognitive interviews or survey. Also, remote communities and people experiencing other forms of social exclusion with limited access to the Internet were more likely to be under-represented. A further limitation is that a conventional response rate could not be calculated. The panel provider used passive invitation methods, meaning that individuals who were invited may not have viewed the invitation before recruitment closed. As a result, it was not possible to determine how many invitees were aware of the study or to assess differences between respondents and non-respondents. Despite this, we did not observe substantial differences between the characteristics of the survey sample and available population-level data.¹² It was not possible to assess predictive validity because we did not measure subsequent 2024 influenza season vaccine uptake in this study.

Conclusion

The cognitive interviews and psychometric evaluation of the Behavioral and Social Drivers of the Influenza Vaccine survey demonstrated validity in an Australian population. Our process of refining survey items for clarity and adding nuanced response options was essential in improving the validation. The survey’s underlying factors aligned well with the BeSD framework, particularly in measuring individual beliefs, social influences, and practical issues related to vaccination. Most items, including a newly developed social responsibility item, are significantly associated with the intention to receive the influenza vaccine, affirming their relevance in understanding vaccination intent. However, the item related to decision-making autonomy did not correlate significantly with its underlying factor, suggesting it may be more useful for descriptive purposes rather than as a predictive variable.

A key implication of this study is the development and validation of a standardized survey instrument that is both locally relevant and internationally aligned. In the Australian context, the tool generates robust data to identify behavioral and social drivers of influenza vaccination across population subgroups. At the same time, alignment with the WHO Behavioral and Social Drivers of Vaccination (BeSD) framework and survey question structure enhances comparability with WHO-recommended indicators, supporting the use of this instrument as a complementary resource to existing BeSD survey tools and its potential applicability in other settings.

An overall limitation of the validation was the inability to assess predictive validity due to the lack of prospective vaccination uptake data. We recommend further validation of this survey in various cultural and health system contexts, with a particular focus on assessing its predictive validity by linking the survey items to actual vaccination uptake following survey administration.

Biography

Majdi M. Sabahelzain has a diverse background across academic and programmatic settings and currently serves as a Research Fellow in the Social and Behavioural Insight in Immunisation research group at the Sydney School of Public Health in Australia. Since 2023, he has been involved in operationalising the Behavioural and Social Drivers of Vaccination (BeSD) survey tools in both high‑income and low‑ and middle‑income countries. He has conducted and reviewed psychometric analyses and validation of multiple vaccine‑demand survey tools, including the Vaccine Confidence Index, the Vaccine Hesitancy Scale, and the Parent Attitudes about Childhood Vaccines (PACV) survey. Recently, he conducted the first psychometric analysis to validate the English version of the WHO’s newly developed BeSD influenza vaccination tool in Australia.

Previously, he served for more than a year as a national Social and Behavioural Change (SBC) consultant for UNICEF in Sudan, where he co‑developed the National Strategy for Risk Communication and Community Engagement (RCCE) for the Federal Ministry of Health in 2022–2023.

 .

Additional file 1: Changes made to survey items based on cognitive interviews

Item number	Original item	Issues identified from interviews	Revised item
D7	What is the highest level of education you have completed? □ Did not enter school/no education □ Primary school (attended or completed) □ Preparatory/middle school (attended or completed) □ Secondary (high) school (attended or completed) □ College degree/university, including associate degree, or trade/technical school □ Post graduate degree	Respondents remarked that “preparatory/ middle school” was not applicable in Australia. One participant questioned why the 5th and 6th options were not “attended or completed” as they were in the process of completing a degree and not sure which item to select. Three participants who attended TAFE (Australian technical school) remarked that the options were not suitable for them.	What is the highest level of education you completed? □ Less than high school □ High school or equivalent □ Trade certificate/apprenticeship □ Bachelor’s degree □ Graduate degree (undertaken after a bachelor’s degree, e.g. Master’s degree, doctorate)
VS2	Where did you last receive an influenza vaccine (select one)? □ Your workplace □ Pharmacy □ Doctor’s office □ Public health clinic □ Hospital □ Other: Specify___________	Participants who had never received an influenza vaccine were unable to answer this question.	Where did you last receive an influenza vaccine? (select one) □ You have not had an influenza vaccine (skip to M1) □ Workplace □ Pharmacy □ Doctor’s office □ Health center □ Hospital □ Other: Specify____________
VS3	Was there any payment made for getting the influenza vaccine? □ No □ Yes	Participants who had never received an influenza vaccine were unable to answer this question. Some participants expressed that this item was ambiguous regarding what was meant by “payment made.” This could mean payment to the clinic or vaccine. Additionally, some participants assumed the question was asking if they paid for the vaccination themselves, and thus answered ‘no’ even if someone else paid. Others cited ambiguity in time point. For example, they may have paid for their last influenza vaccine but none of their previous vaccinations.	When you last received an influenza vaccine, did you, or anyone else, pay for it? □ No (skip to M1) □ Yes
M1	Do you want to get an influenza vaccine during the next flu season? □ No □ Yes □ You are not sure □ You already got an influenza vaccine	Some participants did not know when the next influenza season was due to begin. Influenza vaccination was not available at the time of data collection and some participants remarked that it would be impossible to choose the last response.	Do you want to get an influenza vaccine in the upcoming flu season (May-September 2024)? □ No □ Yes □ Not sure
SP4	Do you think the community leaders want you to get an influenza vaccine? □ No □ Yes	Three respondents would have preferred an “I don’t know” option.	Do you think your community leaders want you to get the influenza vaccine? □ No □ Yes □ Not sure
SP5	Do you think the religious leaders want you to get an influenza vaccine? □ No □ Yes	Most respondents would have preferred an “I don’t know” option. Additionally, those who did not belong to a religious group expressed that they would have preferred the option “not applicable.” Adding “Not sure” is to be avoided in BeSD surveys unless true uncertainty is likely to be prevalent. For this question in Australia, it was necessary.	Do you think your religious leaders want you to get the influenza vaccine? □ No □ Yes □ Not sure □ Not applicable
SP6	Has a health worker recommended that you get an influenza vaccine? □ No □ Yes	Some respondents noted that they were not sure if this referred to the last two years per question VS1, or if a health worker had ever recommended they get an influenza vaccine.	In the last two years, has a health worker recommended that you get the influenza vaccine? □ No □ Yes
SP7-B	What types of communication would you trust to give you accurate information about the influenza vaccine? (check all that most closely match answers) □ Video □ Internet/webpage □ Social media □ Television or radio □ Printed materials (newspapers, magazines, leaflets, poster) □ Other: Specify__________	Many participants mentioned that the type (mode) of communication is irrelevant. The source providing the communication was most important to them. For example, some respondents who would trust government websites or videos were reluctant to choose the available options as this might suggest that they trust all internet content or videos. Similar concerns were raised about the printed media option. Participants remarked that the examples in brackets were starkly different to each other and that while one may trust a leaflet in their doctor’s office for example, they would not trust magazines.	This item was removed from the survey as it did not measure trust in communication source as intended.
P6	What makes it hard for you to get an influenza vaccine? [respondents may select multiple response options.] □ Nothing, it’s not hard □ Making an appointment is hard □ The vaccination place is hard to get to □ The opening time is inconvenient □ The waiting time takes too long □ I cannot take time away from work □ It is expensive □ Something else: Specify _____________________	Three participants cited “vaccine availability” as something that makes it hard. An item was added accordingly.	What makes it hard for you to get an influenza vaccine? [respondents may select multiple response options.] □ Nothing, it’s not hard □ Making an appointment is hard □ The vaccination place is hard to get to □ The opening time is inconvenient □ The waiting time takes too long □ Cannot take time away from work □ It is expensive □ The vaccine is not available □ Something else: Please specify _____________

Additional file 2: Behavioural and social drivers of influenza vaccine item frequencies (N = 2055)

Code	Item and Response Option		Frequency (%)
Thoughts and Feelings
TF1	How concerned are you about getting seasonal influenza, also known as “the flu?”	Not at all concerned	534 (26.0)
		A little concerned	869 (42.3)
		Moderately concerned	424 (20.6)
		Very concerned	228 (11.1)
TF2	How sick would you become if you had influenza?	Not at all sick	128 (6.2)
		A little sick	976 (47.5)
		Moderately sick	851 (41.4)
		Very sick	100 (4.9)
TF3	Before today, had you heard of the influenza vaccine?	No	143 (7.0)
		Yes	1882 (91.6)
		Not sure	30 (1.5)
TF4	How much protection do you think the influenza vaccine will offer you?	No protection at all	185 (9.0)
		A little protection	540 (26.3)
		Moderate protection	877 (42.7)
		A lot of protection	453 (22.0)
TF5	When a person gets the influenza vaccine, how much do you think it protects people in the community?	No protection at all	194 (9.4)
		A little protection	589 (28.7)
		Moderate protection	843 (41.0)
		A lot of protection	429 (20.9)
TF6	How important do you think the influenza vaccine is for your health?	Not at all important	301 (14.6)
		A little important	436 (21.2)
		Moderately important	580 (28.2)
		Very important	738 (35.9)
TF7	How safe do you think the influenza vaccine is for you?	Not at all safe	166 (8.1)
		A little safe	349 (17.0)
		Moderately safe	602 (29.3)
		Very safe	938 (45.6)
TF8	How much do you trust the health workers who would give you the influenza vaccine?	Not at all	129 (6.3)
		A little	279 (13.6)
		Moderately	539 (26.2)
		Very much	1108 (53.9)
Vaccine uptake status
VS1	How many times did you get an influenza vaccine for yourself in the past 2 y?	None	549 (26.7)
		One time	401 (19.5)
		Two times	1089 (53.0)
		Prefer not to say	16 (0.8)
VS2	Where did you last receive an influenza vaccine?	You have not had an influenza vaccine	366 (17.8)
		Workplace	298 (14.5)
		Pharmacy	515 (25.1)
		Doctor’s office	753 (36.6)
		Health center	82 (4.0)
		Hospital	24 (1.2)
		Other	17 (0.8)
VS3	When you last received an influenza vaccine, did you, or anyone else, pay for it?	No	1065 (51.8)
		Yes	624 (30.4)
VS3b	Who paid for the influenza vaccine that you received?	You	337 (16.4)
		A family member	45 (2.2)
		Your employer	202 (9.8)
		Health insurance	23 (1.1)
		Other	8 (0.4)
		Not sure	9 (0.4)
Intention
M1	Do you want to get an influenza vaccine in the upcoming flu season (May-September 2024)?	No	522 (25.4)
		Yes	1290 (62.8)
		Not sure	243 (11.8)
Social Processes
SP1	Do you think you have a responsibility to get the influenza vaccine in order to protect others?	No	719 (35.0)
		Yes	1336 (65.0)
SP2	Do you think most adults you know will get the influenza vaccine if it is recommended for them?	No	746 (36.3)
		Yes	1309 (63.7)
SP3	Do you think most of your close family and friends want you to get the influenza vaccine?	No	699 (34.0)
		Yes	1356 (66.0)
SP4	Do you think your community leaders want you to get the influenza vaccine?	No	1356 (11.3)
		Yes	1149 (55.9)
		Not sure	673 (32.7)
SP5	Do you think your religious leaders want you to get the influenza vaccine?	No	234 (11.4)
		Yes	348 (16.9)
		Not sure	605 (29.4)
		Not applicable	868 (42.2)
SP6	In the last two years, has a health worker recommended that you get the influenza vaccine?	No	1087 (52.9)
		Yes	968 (47.1)
SP7	Who would you trust the most to give you accurate information about the influenza vaccine?	Ministry of health/health authority	495 (24.1)
		Doctor or other health worker	1652 (80.4)
		Spouse or partner	225 (10.9)
		Family member	192 (9.3)
		Friend/neighbour	74 (3.6)
		Other	83 (4.0)
SP8	When it is time for you to get the influenza vaccine, who is the main decision maker?	You	1862 (90.6)
		Your spouse or partner	111 (5.4)
		Your mother or mother-in-law	36 (1.8)
		Your father or father-in-law	14 (0.7)
		Another family member	9 (0.4)
		Someone else	23 (1.1)
Practical Issues
P01	In the last 12 months, have you ever been contacted about being due for the influenza vaccine?	No	1665 (81.0)
		Yes	390 (19.0)
P02	How did you get the reminder?	Mail	25 (1.2)
		Email	156 (7.6)
		Text message	124 (6.0)
		Phone	34 (1.7)
		In person	97 (4.7)
		Other	(0.5)
P03	Do you know where to go to get an influenza vaccine for yourself?	No	10 (12.3)
		Yes	1802 (87.7)
P04	Where would you go to get an influenza vaccine for yourself?	Your workplace	298 (14.5)
		Pharmacy	855 (41.6)
		Private health clinic or doctor’s office	1133 (55.1)
		Public health center	177 (8.6)
		Hospital	74 (3.6)
		Other	15 (0.7)
P05	How easy is it to get the influenza vaccine for yourself?	Not at all easy	77 (3.7)
		A little easy	311 (15.1)
		Moderately easy	506 (24.6)
		Very easy	1161 (56.5)
P06	How easy is it for you to pay for the influenza vaccination?	Not at all easy	247 (12.0)
		A little easy	429 (20.9)
		Moderately easy	467 (22.7)
		Very easy	912 (44.4)
P07	How easy is it to afford other costs associated with influenza vaccination?	Not at all easy	253 (12.3)
		A little easy	425 (20.7)
		Moderately easy	509 (24.8)
		Very easy	868 (42.2)
P08	Reasons for low ease of access.	Nothing, it’s not hard	1318 (64.1)
		Making an appointment is hard	171 (8.3)
		The vaccination place is hard to get to	93 (4.5)
		The opening time is inconvenient	160 (7.8)
		The waiting time takes too long	155 (7.5)
		Cannot take time away from work	124 (6.0)
		It is expensive	229 (11.1)
		The vaccine is not available	74 (3.6)
		Something else	98 (4.8)
P09	How satisfied are you with influenza vaccination services?	Not at all satisfied	12 (0.8)
		A little satisfied	173 (11.6)
		Moderately satisfied	463 (31.1)
		Very satisfied	842 (56.5)
P10	What is not satisfactory about the influenza vaccine services? (If received one or two vaccines in the last two years)	Nothing, you are satisfied	1060 (51.6)
		The vaccine is not always available	175 (8.5)
		The health facility does not open on time	68 (3.3)
		Waiting times are long	161 (7.8)
		The health facility is not clean	35 (1.7)
		Staff are poorly trained	41 (2.0)
		Staff are not respectful	20 (1.0)
		Staff do not spend enough time with people	40 (1.9)
		Something else	56 (2.7)

Funding Statement

This work was supported by the Australian National Health and Medical Research Council (NHMRC) [grant number: 2010212].

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2026.2739738.

Disclosure statement

Julie Leask has received travel support from Sanofi in 2024 and Pfizer in 2025. All other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability statement

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Once the data are on the repository, we can refer any queries to the University of Sydney library dataset link.

Ethical statement

The ethical approval was granted by The University of Sydney Human Research Ethics Committee [2023/HE000668]. All research participants provided digital informed consent.