The role of information presentation for wildlife knowledge, attitude, and risk perception

Public opinion can have a decisive influence on conservation actions leading to a need to understand how public opinion is formed. In a survey with a representative sample of the German population, participants answered questions about foxes in two consecutive years. Different versions of a leaflet about foxes were distributed to 2448 participants before the second interview. We compared a narrative text presentation to a non‐narrative list of facts and examined the use of photographs and schematic graphs. We assessed how the presentation format and socio‐demographic factors affected the probability that participants read the leaflet. Using a before‐after/control‐impact design, we examined whether the leaflet affected people's fox‐related knowledge, attitude, and risk perception. The results show that participants were more likely to read the leaflet with increasing age and a higher educational level. Reading probability also increased with attitude toward foxes. Participants who read the leaflet completely gained more knowledge about foxes than those who read it only partly. Photographs also contributed to a higher knowledge gain, but schematic graphs did not. Moreover, participants who read a fact list gained more knowledge compared to the control condition. Furthermore, the combination of visual and textual features had an effect on attitude toward foxes. However, we found no evidence that any treatment affected risk perception. We discuss implications and derivations for science communication to improve conservation actions.

examined whether the leaflet affected people's fox-related knowledge, attitude, and risk perception.The results show that participants were more likely to read the leaflet with increasing age and a higher educational level.Reading probability also increased with attitude toward foxes.Participants who read the leaflet completely gained more knowledge about foxes than those who read it only partly.Photographs also contributed to a higher knowledge gain, but schematic graphs did not.Moreover, participants who read a fact list gained more knowledge compared to the control condition.Furthermore, the combination of visual and textual features had an effect on attitude toward foxes.However, we found no evidence that any treatment affected risk perception.
We discuss implications and derivations for science communication to improve conservation actions.and wildlife species (Soulsbury & White, 2016;Suraci et al., 2019;Venter et al., 2016).To achieve a good coexistence of humans and wildlife, different parties must be involved in conservation and management actions (Decker et al., 2016;Enck et al., 2006;Salafsky, 2011).Thus, information flow between these groups is crucial, including from research to the general public (Decker et al., 2016;Saunders et al., 2006;Toomey, 2016;Toomey et al., 2017).The challenges are to reach target groups with different social backgrounds and to achieve changes in attitude or behavior.Accordingly, the question arises as to how science, NGOs, or conservationists can reach non-scientists to increase their knowledge about certain wildlife species, their understanding of risks associated with these species, and their attitude toward the species and management methods.
Knowledge, attitude, and risk perception can be influenced by information on wildlife species (Flemming et al., 2018;Robinson & Kiewra, 1995).Communication can be purely fact-based but may also include elements of emotionalization, such as narrative presentations and pictures of young and cute animals (Dahlstrom, 2014;Greving & Kimmerle, 2020, 2021;Nummenmaa et al., 2006).Emotionalization is an important factor for knowledge transfer and changes in attitude and risk perception (Keller et al., 2006;Kim & Morris, 2007;Ryffel et al., 2014).It is well known and strongly supported from previous research that emotions may influence attitudes by changing them in a way that is in accordance with the emotion's valence (Walther et al., 2011).Moreover, emotions may be used as reasons to follow a particular line of argument (Petty & Briñol, 2015).But the specific elements of textual and visual presentation formats that impact people's knowledge, attitude, and risk perception are not well understood.In the study presented here, we aimed to examine how these target variables can be influenced by transferring information in different presentation formats.
For our study we chose red foxes (Vulpes vulpes, hereafter referred to as foxes) as a relevant example of human-wildlife coexistence (Basak et al., 2022;Soulsbury & Statham, 2023) and sent leaflets about foxes to a representative sample of the German population.We conducted a longitudinal intervention study in which the participants answered an identical questionnaire about red foxes in two subsequent years.In the second year, some participants received leaflets of different formats but identical content, about 6 weeks before they answered the questionnaire again.The treatments differed regarding text type with a narrative article or a fact list and regarding visualization with graphs, photographs, graphs and photographs, or no visualization, resulting in eight treatment groups and one control group.In the analysis, we investigated how textual and visual features of information influenced (a) the probability that recipients read the leaflet, and (b) how the different formats influenced their knowledge about foxes, attitude toward foxes, and risk perception of fox-related diseases.
For knowledge transfer to be successful, target groups first need to take in the communicated information.The willingness to read the provided information could be influenced by a multitude of factors.We expected a. that a more positive attitude toward foxes would lead to a higher probability that the leaflet was read (Jones & Sugden, 2001).b. that visual cues would catch the attention of readers (cf.Myers & Sar, 2013;Zillmann et al., 2001), leading to a higher reading probability.c. that a narrative text would catch the attention of readers (cf.Dahlstrom, 2014), leading to a higher reading probability.
In addition, we expected socio-demographic factors such as age, gender and education level to influence reading probability, particularly because a previous study found that these factors correlated with attitudes toward foxes: With a higher education level and a younger age, the attitude toward foxes was more positive (Kimmig et al., 2020), and gender-related differences in attitude were also observed with men holding more positive attitudes toward foxes than women (Kimmig et al., 2020).
Once recipients read the material, the presentation format may also influence the absorption of information and its effects in terms of factual knowledge, attitude and risk perception.Flemming et al. (2018) showed an interaction effect of text type and visual features on knowledge gain in a laboratory study.Narrative information has been shown to be better understood, which results in a deeper processing of information and more knowledge (Dunlosky et al., 2013;Graesser et al., 1980;Kintsch, 1994).When adding visual features, information can be better memorized (Myers & Sar, 2013).With increased knowledge, attitude has been shown to improve as well (Kimmig et al., 2020), even though there is also frequently a gap between environmental knowledge and pro-environmental behavior (e.g., Kollmuss & Agyeman, 2002).Moreover, emotionalization has an impact on attitude (Ryffel et al., 2014).Finally, risk perception with respect to red foxes has been shown to decrease with more knowledge or more positive attitudes (Kimmig et al., 2020).Flemming et al. (2018) found that risk perception was lower after reading information presented as a fact list and when photographs were presented.For the impact of presentation format on the target variables knowledge, attitude, and risk perception, we developed the following hypotheses, referring to participants who read the leaflet: H1a. Reading the leaflet will lead to more knowledge.
H1b. Material with visualizations will lead to more knowledge.
H1c. Information presented in a narrative format will lead to more knowledge.

H2 (attitude)
H2a.A narrative text will lead to a more positive attitude.
H2b. Material with photos will lead to a more positive attitude.

H3 (risk perception)
H3a.A fact list will lead to lower risk perception.
H3b. Material with visualizations will lead to lower risk perception.

| METHODS
Participants were recruited via the Socio-Economic Panel (SOEP) that holds a socio-demographic long-term dataset of the human population in Germany.It is a longitudinal study that annually (re-)interviews up to 30,000 adults (Goebel et al., 2018).In two consecutive years (2016, 2017) a sub-sample of the same participants answered an identical questionnaire about foxes.To analyze the effect of different presentation formats on knowledge, attitude, and risk perception, randomized participants received leaflets with different presentation formats 6 weeks before the second interview in 2017 (see Figures A1 and  A2 in the appendix for examples of the original leaflets).Thus, in 2017 participants either received different versions of the leaflet (eight different treatment groups, Table 1) or did not receive any further information (control group).To investigate the impact of the treatments on the change in knowledge, attitude, and risk perception, we analyzed our data according to a Before-After / Control-Impact (BACI) design (Green, 1979).
Therefore, we split up the data of 2016 into the groups of 2017 (eight treatment groups and the control group).
We used text type (narrative article vs. fact list) and visualization (no visualization vs. photographs vs. graphics vs. photographs and graphics) as between-group factors.Together with a control group that did not receive any material [N = 313], this procedure resulted in nine conditions.2448 persons participated [age range: 17-96 years; mean age = 54 years; SD = 18.61; gender: 1320 (54%) female; 1128 (46%) male].2135 persons received the leaflet; out of those participants 393 participants stated that they read the leaflet completely (18%), 358 participants read it partially (17%), 1367 did not read it at all (64%), and 18 participants did not answer the questionnaire.Distribution of participants per treatment group is shown in Table 1.For the impact analysis according to a BACI design, we calculated for 929 participants out of 1032 participants: These variables not only accounted for the change between 2016 and 2017 (before-after) but also corrected for temporal changes (control-impact).The participants attended voluntarily and anonymously.Ethical approval was obtained from the Local Ethics Committee of the Leibniz-Institut für Wissensmedien (approval number: LEK 2017/001).
The knowledge test consisted of eleven statements about foxes, with six statements that were correct and five statements that were wrong.For each statement, the participants had to indicate whether it was true or false.For each item that participants correctly identified as true or false, they received one point.All points were added and divided by the number of items, resulting in a scale from 0 (no knowledge) to 1 (perfect knowledge).The variable was normally distributed.
The attitude questionnaire consisted of four items that participants rated on seven-point Likert scales (ranging from 1 = do not agree at all to 7 = totally agree; Joshi et al., 2015).All ratings were summed up and the score was reduced by 4 (in order to have the lower end at the value of 0), and then divided by the number of rating points (7) and by the number of items (4), resulting in an attitude score that could range from 0 to 1, with higher values indicating a more positive attitude.Internal consistency (Cronbach's alpha) of this scale was α = 0.73.The variable was normally distributed.
For the risk perception questionnaire, participants were instructed to rate the danger of four diseases (echinococcosis, distemper, foot-and-mouth disease, and mange).Answer format was a 6-point Likert scale with the option to answer "not able to estimate."All ratings were summed up and reduced by the number of items that were evaluated by the participant (0-4) and divided by the number of items (4) and the number of rating points (5) resulting in a risk perception scale that ranged from 0 (low risk perception) to 1 (high risk perception).To reach a normal distribution, the variable was square root transformed.
To analyze the probability that recipients read the material and to evaluate our expectations (a-c), we conducted generalized linear models (GLM).As predictor variables, we included demographic factors (education [Table A1], gender, year of birth), the presentation formats (text type and visualization with interactions between the terms), as well as attitude and knowledge scores from the first measurement.First, we examined the participants who read the material completely or not at all (binomial GLM, link function: logit).Thus, the categories "read completely" and "read not at all" were set to 1 and 0, respectively.Second, we additionally included the participants who stated that they only partially read the leaflet.To account for the uncertainty of the variable "partially read" we set up three different linear models coding 'read partially' either as 0.25, 0.5, or 0.75, indicating that 25%, 50%, or 75% of the leaflet was read.We set up a GLM using "ArcSin(Square root (x))" as the link function.
To analyze the impact of the presentation formats on our target variables BACI (Equation ( 1)), we ran a multivariate analysis of variance (MANOVA) for the experimental factors text type and visualization with an interaction term between these variables as predictors on changes in knowledge, changes in attitude, and changes in risk perception as dependent variables with Tukeycorrection.Further, we included the extent to which the leaflet was read (categorical variable coded as: "completely," "partially," control group) as additional interaction term to account for possible differences among these groups.As a test statistic, we used Wilk's Λ. Afterwards, we conducted univariate testing (ANOVA) for each target variable and the same set of independent variables.By performing the Tukey honest significant difference test on the ANOVAs, we looked more closely at significant differences among the treatments to test our hypotheses.All statistical analyses were conducted in R version 4.1.3(R Core Team, 2022).The significance level was set to α = .05.

| Reading probability
We analyzed how different presentation formats and socio-demographic factors influenced the probability that the material was read (see Table A2 for all statistics).The reading probability was not affected by text type, visualization, or the interaction between those factors.There was also no significant difference regarding gender.However, the probability to read the leaflet increased with age and with education level (Figure 1).
The attitude toward foxes given in the pre-survey in 2016 had a significant positive influence on the probability of reading the material in 2017, thus reading probability increased with attitude (Figure 1).There was no difference with respect to participants' prior knowledge.Including the participants in the analysis who only partially read the leaflet did not make a difference compared to the binomial model using information of participants that either read the leaflet completely or not at all (Table A2, Figure A3).The results were also not sensitive to the assumption of which percentage of the leaflet constituted "partially read" (i.e.read 25%, 50%, or 75% of the text).

| Knowledge
On average, participants' knowledge was M = 0.42 (SD = 0.14) in 2016.Univariate testing indicated that the reading extent had an effect on change in knowledge (F = 9.033, p = .003;partial η 2 = 0.010).The post-hoc Tukey test revealed that participants who read the leaflet only  A2. partially had no difference in knowledge change compared to the control group, whereas participants who read the leaflet completely, showed an increase in knowledge compared to the control group (Figures 2 and 3), which supported H1a.
Visualization and text type also influenced knowledge (visualization, F = 4.1862, p = .006;partial η 2 = 0.013; text type, F = 6.416, p = .002;partial η 2 = 0.014).Findings of the post-hoc Tukey test supported H1b partially: Adding photos lead to a stronger knowledge increase compared to participants who received leaflets without visual features (Figure 3).However, adding graphs had no effect on knowledge change.Moreover, the post-hoc test rejected H1c: Compared to the control group, only participants who read a fact list showed an increase in knowledge (Figure 2).

| Risk perception
The participants' average risk perception was M = 0.29 (SD = 0.22) in 2016.Univariate testing revealed that the reading extent (F = 1.874, p = .171)and the presentation formats with visualization and text type had no influence on risk perception (visualization, F = 0.974, p = .404;text type, F = .622,p = .537).Thus, neither of the two hypotheses (H3a, H3b) was supported.In the present study, we analyzed how leaflets about foxes can be perceived by recipients and how textual and visual presentation formats impact people's knowledge, attitude, and risk perception with respect to foxes.For the latter the participants answered an identical questionnaire about red foxes in 2016 and 2017.In 2017, some participants received leaflets about 6 weeks before they answered the questionnaire for the second time.
Out of all recipients, 393 participants stated that they read the leaflet completely, 358 participants read the leaflet partially, while the majority of recipients stated that they had not read the leaflet at all (1367).However, there was no difference regarding the reading probabilities irrespective of whether the second group was in-or excluded in our analyses.Since the replies were self-assessments, we have to consider that participants may have falsely stated that they read the leaflets, at least partially, because they felt this was the socially desired answer.Participants could also have stated that they did not read the leaflet despite having read it, to avoid having to answer additional questions in the survey.
The results show that the material was read more frequently by participants with increasing age, with a higher educational level, and with a more positive attitude toward foxes.This is in line with previous research indicating that with higher educational level, people were more motivated to read new information (Price & Zaller, 1993).Additionally, when information supports the existing attitude, the reading probability is higher, as indicated in studies by Spink et al. (1998) and Jones and Sugden (2001), for example.Since the material handed out in the present study emphasized a more positive attitude toward foxes, our results support these previous findings.
Although attitude and knowledge are linked as well (Kimmig et al., 2020), there was no significant influence of prior knowledge on reading probability.For emotionalization reasons, it could be argued that the narrative article with pictures should be most attractive to read as stated by Dahlstrom (2014) who suggested that researchers should rely on narrative formats to reach a non-expert audience.It has further been demonstrated in previous studies that pictorial stimuli catch readers' attention (Garcia & Stark, 1991;Zillmann et al., 2001), which could motivate reading.However, in the study presented here, narrative text formats alone did not result in a higher reading probability and adding photos or graphs also did not increase reading probability.Adding a combination of photos and graphs did not increase reading probability -this may be due to an overdose of visual cues, where too many visual effects outperform one visual effect (Davy, 2006;O'Reilly, 1980).
There may also be other factors that influence the reading probability of material, which we did not include in our analysis.For example, people's disposition to read a leaflet could be affected by the quality and acceptance of the source of information (Borgatti & Cross, 2003).It has been shown that motivation can vary due to individual circumstances (Davy, 2006;Szulanski, 2000).The effects of our treatments may have been masked by pronounced individual variation.We conclude that with the leaflet used in this study, we reached mainly older people, recipients with a higher educational background and people who already had a positive attitude toward foxes.
Looking more closely at the impact of the provided information on knowledge, attitude, and risk perception, our results show that the provided leaflet only had effects on participants who read the entire leaflet.Participants had more knowledge when they read the distributed material completely, which supported our hypothesis concerning the impact of information on knowledge.With regard to textual features, the positive impact on knowledge was only detectable for the fact list when distributed without visual effects.This contrasts with Negrete and Lartigue ( 2010) who showed that scientific information transferred via narrative articles was memorized for longer time spans than when presented as fact lists.Similarly, Webb (2008) found that information transfer resulted in an increase in knowledge when information was embedded in a contextual background, which occurs when information is presented as a narrative article.
There was a significant effect on knowledge gain when photos were added to the leaflet.However, adding graphs or the combination of graphs and photos did not result in an increase in knowledge.As for the reading probability, this could be caused by an "overdose" of visual cues, which limited the uptake of information (Davy, 2006;O'Reilly, 1980).This finding is also in line with research on the seductive details effect (e.g., Sundararajan & Adesope, 2020), which indicates that learners who were exposed to attractive but irrelevant information perform worse in a knowledge test than learners who were not exposed to such seductive details.
In the present study, neither a narrative article nor a fact list led to a more positive attitude toward foxes, which rejected our hypothesis regarding the impact of information on attitude.The differences between fact list and the narrative format were not significant.When visualizations were presented, attitude did not change.Thus, textual features and the additive effect of visual features increased people's knowledge, but there was no effect on attitude.Childers and Houston (1984) and Robinson and Kiewra (1995) found a positive impact of visual cues on knowledge, which is in line with the results of this study.However, the absent impact of visual features on attitude is in contrast to Mitchell (1986) who found that visual effects had a predominant effect on attitude formation.Since we focused in the present study on the additional effects of visual features to the textual features, a higher sample size may have resulted in a stronger additive effect.
In contrast to our hypothesis and previous results by Flemming et al. (2018), we did not find any influence of the material on risk perception.Considering that knowledge, attitude, and risk perception have been shown to influence each other (Kimmig et al., 2020;Miao et al., 2020), we could surmise that with increased knowledge about a specific topic, attitude toward foxes should be adjusted and thus people should be able to realistically assess the risks associated with the topic.However, the expected effect may not have occurred since risk perception was already low in the study presented here.
We showed how information can be perceived by participants with different and which effect textual and visual features can have on knowledge about and attitude toward foxes.When transferring information to the public for conservation purposes, it might be good to test the impact of the material on a representative sample size of the target group in advance of conducting an information campaign, since different formats have different impacts.This was also suggested by Kidd et al. (2019), where the authors emphasize the importance of segmenting the audience in advance to reach a target group with a specific presentation format.Communication of scientific information with combinations of textual and visual features can impact knowledge and attitude about specific wildlife and conservation measures, which are important to enable human-wildlife coexistence in the future.Taking the right framing into account (Kusmanoff et al., 2020), attitude and behavior are influenced to improve conservation actions.As the attitude of people affects the chance that future information material is attended to, an information campaign might initiate a positive feedback loop.However, one may argue that conservationists may face the dilemma of only reaching out to a minority that already supports their views and actions (see also Veríssimo, 2013).Thus 2 Middle Degree from a secondary modern school from class level 5-10 ("Realschule") 3 High University-entrance diploma from a secondary modern school from class level 5-12 or 13 ("Gymnasium" with degree "Abitur" or "Fachhochschulreife") T A B L E A 2 Results of binomial (information material was read completely [1]; information material was not read at all [0]) and general linear models, depending on the extent to which the leaflet was read.The general linear model uses ArcSin(Square root (x)) as the link function.As the true extent to which participants read the leaflet is unknown for the partial readers, we used three different values, that is 0.25, 0.5, and 0.75 respectively.In this context 0.5 means that 50% of the leaflet was read.

K
E Y W O R D S attitude, knowledge, presentation format, risk perception, science communication, survey, wildlife management 1 | INTRODUCTION Human expansion comes with intensive land use, resulting in an increased level of interactions between humans Danny Flemming and Miriam Brandt contributed equally to the manuscript.

F
I G U R E 1 Effect plots of the binomial GLM showing the reading probability depending on year of birth (left), education level [1 = low, 2 = middle, 3 = high, see Table A1] (middle), and attitude (right).The shaded area and the bars denote the 95% confidence intervals.Results of the binomial model are presented in Table

F
I G U R E 2 Knowledge change and attitude change in the two text type conditions compared to the control group.The bars denote the 95% confidence intervals of the Tukey honest significant difference test on the ANOVA.Groups where the bars do not cross the zero line are significantly different from the control group.

F
I G U R E 3 Knowledge change and attitude change in the four visualization conditions compared to the group which received leaflets without visual features.The bars denote the 95% confidence interval of the Tukey honest significant difference test on the ANOVA.
, the question of how to reach other parts of society remains open and might decide about the success and failure of conservation measures.AUTHOR CONTRIBUTIONS S.E.K., M.B., F.O., R.H., J.K., and U.C. designed the research.R.H., F.O., and D.F. conducted data analysis.F.O. wrote the manuscript.S.E.K., M.B. and J.K., developed the questionnaire.All authors contributed substantially to the drafts and gave final approval for publication.APPENDIX A T A B L E A 1 Educational level classification according to school type attended.certificate or a degree from a secondary modern school from class level 5-9 in Germany ("Hauptschule")

F
I G U R E A 1 Example of the leaflet (condition: fact list including graphs).F I G U R E A 2 Example of the leaflet (condition: narrative article including photos).OEHLER ET AL.F I G U R E A 3 Effect plots of the GLM including participants who read the leaflet partially, showing the reading probability depending on year of birth (left), education level (middle) and attitude (right).The shaded area and the bars denote the 95% confidence intervals.
Number of participants per treatment group who read the leaflet completely, partly, or not at all.
T A B L E 1 The value Pr(>jzj) represents the probability that the value of the Gaussian distribution is greater than z or smaller than z.Thus it equals the two-sided p-value.Values of Pr(>jzj) smaller than the significance level (α = .05)are highlighted in bold. Note: