A total of 1497 completed responses were collected. We only included fully completed surveys. A survey was considered fully completed if a respondent reached the final question in the questionnaire and responded to all questions used as variables in the regression. This applied to 1500 respondents. Three respondents were excluded because they did not actively give consent resulting in a final sample size of 1497 respondents. 750 respondents were in the control group and 747 were assigned to the information treatment group.
The sample was targeted to be representative of the population of Pakistan in age, gender, and geographic region. The sample consists of slightly more males than females (see Table 1). Our sample is slightly older than the population, with around 35% of respondents being less than 30 years of age whereas more than 50% of Pakistan’s population is 19 years or younger42. However, this discrepancy is likely in part a result of the age restriction levied in the survey, which excluded respondents under the age of 18 from participating.
Data were collected across four main regions in Pakistan. Those regions and the number of respondents within each were: Punjab (62%), Sindh (2%), Khyber Pakhtunkhwa (27%), and Balochistan (10%). While efforts were made to ensure representativeness in the sample obtained, there were statistically different proportions of respondents in the sample residing in each region when compared to the Census targets. Nevertheless, the statistically higher portion of respondents from areas other than Punjab allow us to also gather insights from otherwise spatially under sampled communities.
67% of respondents can be classified as high frequency buyers of chicken, purchasing at least twice a week. Wet markets were the most common locations to purchase chicken (59%), although supermarkets were also commonly frequented by about 40% of respondents corresponding with the rise of supermarkets in Asia observed since the early 2000s49.
Almost 80% of our respondents previously took antibiotics and 40% indicated that at least one of their children had taken antibiotics in the past. Only 11% indicated never having taken antibiotics and 8% were uncertain if they had.
Familiarity with and perception of antibiotics and phages
85% of respondents indicated that they were familiar with antibiotics, while 69% indicated familiarity with AMR. Less than half the sample was aware of bacteriophages (48%), which was expected given the relative novelty of the technology. As we asked about respondent’s familiarity with the three concepts prior to taking the survey and thus receiving the information, no significant differences exist across the Control and Treatment groups.
Related to their familiarity, we find that 55% of respondents answered at least half of the objective knowledge questions included in the survey incorrectly, with only 1% of respondents answering all eight questions correctly (see Supplementary Table S1 in the Supplementary Material) suggesting that most participants lack significant understanding of the specificities concerning antibiotics and phages. The result held across both the Control and Treatment.
71% of respondents indicated that they would support using new technologies as alternatives to antibiotics in poultry production. Among those supporting the use of novel approaches, 76% were in support of using bacteriophages. Thus, significant support exists among respondents to not only move away from antibiotics, but potentially move towards the use of phages. Already, there are several countries that have permitted the use of bacteriophage products in food processing such as the United States, Canada, Australia, and Israel50. Au50 notes that while many countries do not regulate phage use in food production, phage products are allowed on food exports from countries where such use is authorized. Studies50 note the benefits of phage adoption in food. Like antibiotics, for example, phages can be used as preservatives to extend expiration dates of food products51. Moreover, phages are able to selectively reduce food-borne illness without harming the gut microflora52.
We found no significant difference in respondent’s support for the technologies across Control and Treatment groups suggesting that the information provided did not directly impact consumer’s acceptance. Respondent’s support for new technologies is slightly less than their self-indicated levels of concern about AMR, as around 82% of respondents indicated to be at least somewhat concerned about AMR (59% were somewhat concerned, 23% were extremely concerned).
The majority of respondents at least somewhat agree that chicken produced without the use of antibiotics is more nutritious, safer, better for kids, healthier, and environmentally friendlier (see Fig. 1). That said, the majority of respondents also consider chicken produced without antibiotics to be more expensive, which might impact their decision to seek the product out in the market. Across treatments, we only find a significant difference (determined via a Chi2-test) across groups in terms of antibiotic-free chicken being different to other chicken (45% vs. 39%) suggesting that the provided information only affected the perception of very a limited subset of attributes. This lack of differences is even more pronounced when looking at the results for respondents’ perception of how the use of bacteriophages would affect the seven quality attributes, as no significant differences are detected across the two groups. Nevertheless, in both groups we again find that respondents at least somewhat agree with most of the statements except for the bacteriophage-treated chicken being the same (Control: 41%, Treatment: 42%) or more expensive (Control: 49%, Treatment: 51%).
Respondent agreement with statements about chicken production without antibiotics or with bacteriophages. Agreement measured as a dummy where responses of “strongly agree” of “somewhat agree” on a 4-point Likert scale are captured as 1 and all others are recorded as 0.
The probit models
Table 2 presents the outcomes derived from our two Probit models, which measure the correlation between (a) the acceptance of new technologies in general and (b) bacteriophages in particular for microbial control, and the independent variables described in “Data analysis”. The first and third columns provide coefficient estimates, while the second and fourth columns display the corresponding marginal effects.
As a reminder, we modeled consumer acceptance for both new technologies and phage technology using the full sample, which includes respondents from both the Control and Treatment groups. Please also note that for the second model (Acceptance of Phage Technology), we only recorded responses from 1061 respondents, as this question was only shown to those respondents, who indicated to accept new technologies to reduce antibiotic use.
Regarding the acceptance of new technologies, we note that respondents’ familiarity with antibiotics has the most substantial marginal effect (0.147). This finding indicates that individuals who are familiar with antibiotics are more likely to embrace alternative technologies to antibiotics. Similarly, we find significant positive marginal effects for familiarity with bacteriophages (0.112) and antimicrobial resistance (AMR) (0.053). These results align with previous studies that have found a positive association between consumer’s familiarity with a technology and how suspicious they are of it53. Furthermore, our findings demonstrate that being acquainted with related concepts, such as antibiotics and AMR, significantly influences consumer perceptions.
In terms of socio-demographics, we only identify a positive marginal effect for the age variable (0.068), indicating that respondents under 30 years of age are more inclined to accept new technologies as alternatives to antibiotics compared to older respondents. These results are consistent with previous studies that have examined the relationship between technology acceptance and consumer demographics54.
Notably, we find no significant correlation between the information treatment and respondents’ acceptance. This suggests that providing general information about bacteriophages may not lead to widespread changes in perception. Our results contrast with earlier studies that explored the impact of consumer acceptance and willingness to pay for other food-related technologies, such as genetic modification55 or gene-editing45,56. However, as demonstrated by Caputo et al.45, the effectiveness of information dissemination is also dependent on the product and its processing degree, which might have influenced our results.
Shifting our focus to the acceptance of bacteriophages, we once again find significant positive marginal effects for familiarity with the technology itself (0.114) and AMR (0.077). Interestingly, familiarity with antibiotics was not a significant factor (0.006). However, it should be taken into consideration that responding to this question was conditional on responding “yes” to the acceptance of new technologies, which might have influenced our results.
Also of interest is the diametrical marginal effect for age (− 0.067), which suggests that younger consumers might be more willing to accept a new technology to reduce the use of antibiotics, but bacteriophages are not among the preferred options to be used in this case.
We further observe, that female respondents were less likely to accept bacteriophages (− 0.060), which mirrors the results of earlier studies on different new technologies57,58.
The multivariate probit models
Table 3 reports the estimates from Model 1 of the MVProbit estimation. As a reminder, this model captured respondents’ perception of various quality attributes reported above relative to the same independent variables included in the Probit model plus a dummy reporting consumer’s general acceptance of new technologies as alternatives to antibiotics.
The results show that the marginal effect of providing respondents with information about bacteriophages is insignificant across almost all quality attributes with the exception of perceiving the chicken to be the same (− 0.147), which aligns with our descriptive results. Studies have shown that consumers perceive antibiotic use in meat production to negatively impact the animal’s welfare59, which might be why informing respondents about available alternatives leads to this difference in perception.
Interestingly, the general acceptance of new technologies was not significant for any of the quality attributes. However, as for general acceptance, we find that with regards to familiarity with the three key concepts of antibiotics, bacteriophages, and AMR, the former has the most pronounced impact across quality attributes, as we found a significant positive marginal effect on the safety perception (0.314), price point perception (0.341), children suitability (0.363), health perception (0.392), and environmental impact (0.196). Familiarity with AMR had a significant but less strong impact on these attributes (aside from the price point perception), but also had a substantial positive impact on the perception of nutritiousness (0.166) and similarity understanding (0.222). This highlights how important familiarity with those two concepts is for how chicken is viewed by consumers: understanding the concept of AMR and being familiar with antibiotics is correlated with positive associations for antibiotic-free chicken.
Meanwhile, bacteriophage familiarity was only significant for two out of seven attributes (children suitability [− 0.161], and health perception [− 0.233]), but noticeably these effects were negative suggesting that being familiar with an alternative to antibiotics may also be associated with some distrust towards not using antibiotics.
Objective knowledge meanwhile had significantly positive marginal effect on the price perception (0.143) suggesting that respondents with a higher objective knowledge are aware of the increased costs associated with foregoing antibiotics. We also observe a positive marginal effect on respondents’ perceptions of the antibiotic-free chicken’s suitability for children (0.152) and safety (0.256). This suggests that respondents with greater objective knowledge may be more aware that antibiotic use in animal production can impact human health, indicating a better internalization of the concepts underlying the One Health framework. In terms of socio-demographics we find that respondents with lower incomes are more likely to perceive less of a difference between antibiotic-free and conventional chicken in terms of nutritiousness (− 0.152), safety perception (− 0.198), price point perception (− 0.325), children suitability (− 0.260), health perception (− 0.300), and environmental impact (− 0.249), which aligns with previous studies showing that a greater income is associated with a more critical view of antibiotic use in animal production60. Interestingly, while we found in our Probit model that younger respondents are more likely to accept alternatives to antibiotics, we observe significant negative marginal effects for five out of seven attributes including for similarity understanding (− 0.165). This could potentially indicate that younger respondents see a difference between antibiotic-free and conventional chicken, but not in terms of any of the attributes explored here. Their greater acceptance of alternative technologies to antibiotics could suggest that the differences they do perceive are of concern to them. In coherence with our Probit models, the significant marginal effects associated with identifying as a female are positive.
A slightly different picture emerges for our second MVProbit estimation, Model 2, which considered the same independent variables, but the dependent variables were focused on chicken production with bacteriophages (see Table 4 for the results). In contrast to model 1, familiarity with antibiotics is not significant for any of the observed attributes. Instead, familiarity with AMR has a significant and strong effect on all attributes. Moreover, we find that familiarity with bacteriophages has a positive marginal effect on perception of nutritiousness (0.160) and price point perception (0.210) which highlights how consumers’ product considerations alter when the use of a new technology is considered instead of merely the reduction of antibiotic use. Importantly, we also note that similarity understanding (0.325) has a significant positive effect, indicating that respondents who understand what bacteriophages are perceive it as a technology that does not fundamentally alter chicken products, while familiarity with antibiotics alone has the opposite effect.
Interestingly, we also note that objective knowledge has a significant and positive effect on all attributes except for similarity understanding (− 0.026), which is insignificant. suggesting that higher levels of objective knowledge are associated with more favorable opinions about bacteriophages (relative to antibiotics). The result stresses the importance of potentially establishing long-term educational interventions that aim to increase consumer knowledge about AMR, antibiotics, and bacteriophages. By enhancing objective knowledge, these interventions could help build consumer trust and acceptance of bacteriophage technology as a viable alternative to traditional antibiotic use. Again, we also find that neither the information provision nor technology acceptance were significant for any of the attributes suggesting that the factors influencing consumer perceptions and acceptance of bacteriophages may extend beyond general awareness and openness to new technologies. If so, this could indicate that more nuanced and targeted approaches are needed to effectively shape consumer opinions regarding the use of bacteriophages in poultry production.
In terms of socio-demographics, we also see a shift. While young consumers are still less likely to significantly agree with any of the attribute related statements, we find that gender has no significant effect for any of the attributes and income is only significant for health and environmental perception. Given the low familiarity with the technology, potentially these results are guided by respondent’s uncertainty regarding bacteriophages.
Results of models 3 and 4 are reported in the Supplementary Material in tables S2 and S3. As a reminder, models 3 and 4 mirror models 1 and 2, but exchange Technology Acceptance for Bacteriophage Acceptance as independent variables meaning that they only consider responses from consumers who indicated that they would accept a new technology instead of antibiotics. The results largely reflect the findings for models 1 and 2. However, for model 4, we find highly significant and positive marginal effects of bacteriophage acceptance on each of the seven attributes. Thus, respondents in this group seem to perceive chicken produced with phages as overall better for humans and the environment and not different to other chicken, but also as more expensive than chicken produced with the use of antibiotics. As such the group could represent an interesting target group for marketers and producers of bacteriophage treated chicken.