1

1 The Effects of Factual Knowledge and the Illusion of Knowing on Knowledge and Pro-

2 environmental Behavioral Changes: A Longitudinal Study

3 Abstract

4 As the world’s largest greenhouse gas emitter, the Chinese public’s understanding of

5 climate change is crucial for mitigation. Drawing on the Positive Illusion Model and the

6 Knowledge Deficit Model, this study examines the effects of the illusion of knowing and

7 factual knowledge on changes in climate change knowledge and pro-environmental behaviors

8 over time. A two-wave longitudinal survey with a final sample of 427 participants was

9 conducted from November 2021 to January 2022. The illusion of knowing positively

10 predicted knowledge change, while the pathways from the illusion of knowing and factual

11 knowledge to changes in pro-environmental intentions differed significantly among different

12 groups, with a positive relationship in the high pro-environmental self-identity group and a

13 negative in the low. The findings of this study offer crucial theoretical contributions by

14 advancing the Knowledge Deficit Model and revealing the moderating role of pro-

15 environmental self-identity, which may guide future effective communication campaigns and

16 policy initiatives.

17
 2

18 1. Introduction

19 Climate change, driven by rising greenhouse gas emissions, is a pressing global issue

20 with severe environmental impacts (Arbuckle et al., 2013; Bostrom et al., 2019; Jørgensen &

21 Termansen, 2016; Metag & Schäfer, 2018). China, the world’s largest greenhouse gas emitter,

22 plays a crucial role in addressing this challenge (Copsey et al., 2013; Gong et al., 2020).

23 Despite the Chinese government’s commitment to achieving carbon neutrality by 2060 and

24 peak emissions by 2030, which has sparked increased media coverage and public discussion

25 (Pan et al., 2021), some segments of the Chinese population remain skeptical or hold negative

26 attitudes toward climate change due to factors such as economic concerns, lack of

27 knowledge, and conspiracy beliefs (Hoekstra et al, 2024; Otto & Pensini, 2017; Pan et al.,

28 2022, 2023; Wang et al., 2019). Therefore, exploring the Chinese public’s knowledge and

29 understanding of climate change is essential for ensuring effective climate actions.

30 Recent studies have highlighted a concerning trend where many individuals possess

31 limited factual knowledge about climate change, leading to an illusion of knowing (i.e.,

32 overconfidence about their knowledge) (Glenberg et al., 1982; Trémolière & Djeriouat, 2021;

33 Yang et al., 2020a, 2020b). Factors such as attention to media messages, message elaboration,

34 climate change skepticism, and socio-demographics might drive this phenomenon

35 (Trémolière & Djeriouat, 2021; Yang et al., 2020a, 2020b). However, the consequences of the

36 illusion of knowing about climate change remain unexamined. Investigating these

37 consequences could provide insights into the cognitive processes underlying climate change

38 knowledge, perceptions, and actions, offering conceptual evidence for future theoretical

39 development and informing the design of communication campaigns and policy initiatives

40 that foster accurate understanding and encourage pro-environmental behaviors.

41 Another notable limitation in the existing literature is the static nature of previous

42 research on climate change, which predominantly aimed at predicting behavior and

 3

43 knowledge at a single point in time (e.g., Lu, 2022; Ma et al., 2023b; Steg et al., 2023; Xu et

44 al., 2022). Our focus shifts towards understanding the changes in behavior and knowledge

45 over time. The distinction is crucial as determinants of behavior at a single point may differ

46 from those driving behavior change over time (David & Rundle-Thiele, 2019; Kriznik et al.,

47 2018; Michie et al., 2008). Existing communication and behavioral theories centered on

48 explaining behaviors could lead to a misallocation of resources by practitioners, especially

49 when the ultimate goal is to promote behavioral changes.

50 Therefore, this study has two purposes. First, we examine the effects of the illusion of

51 knowing about climate change on changes in knowledge and behavior over time. Second, we

52 explore how factual knowledge leads to behavior changes over time based on the Knowledge

53 Deficit Model. Answering these two questions makes three interconnected contributions.

54 First, innovatively employing changes in knowledge and behaviors as the dependent variable

55 advances the Knowledge Deficit Model by validating it from the perspective of behavioral

56 changes. Second, we examine this illusion’s influences on individuals’ changes in knowledge

57 and behaviors longitudinally, offering a crucial theoretical contribution. Lastly, we reveal the

58 moderating role of pro-environmental self-identity in these effects, enriching our

59 comprehension of the complex interplays among these factors.

60 2. Literature Review

61 2.1 Illusion of Knowing in the Context of Climate Change

62 The concept of the illusion of knowing, originating from the Cognitive Miser Model

63 in cognitive psychology, posits that individuals, as “cognitive misers,” often rely on cognitive

64 heuristics or mental shortcuts, such as self-perception of knowledge, to form opinions or

65 make decisions in order to conserve cognitive resources (Corcoran & Mussweiler, 2010;

66 Glenberg et al., 1982; Orbell & Dawes, 1991; Yang et al., 2020a, 2020b). However, this self-

67 perception is susceptible to psychological influences, such as self-serving bias, which can

 4

68 lead to an overestimation of one’s perceived knowledge, known as the illusion of knowing,

69 where individuals misjudge themselves as well-informed despite having limited actual

70 knowledge (Brosowsky& Egner, 2021; Orbell & Dawes, 1991; Trémolière & Djeriouat,

71 2021; Yang et al., 2020a, 2020b).

72 The illusion of knowing has gained attention in recent communication research

73 (Avhustiuk et al., 2018; Trémolière & Djeriouat, 2021; Yang et al., 2020a, 2020b). Scholars

74 argue that this illusion may impact how individuals interpret and respond to information,

75 which is particularly relevant in climate change communication, where accurate

76 understanding and engagement are crucial for fostering meaningful behavioral changes

77 (Moser, 2010, 2016; Steg et al., 2023; Yang et al., 2020a). The multifaceted nature of climate

78 change, encompassing scientific, political, and socio-economic dimensions, presents a

79 cognitive challenge for individuals to fully comprehend (Metag & Schäfer, 2018; Moser,

80 2016; Steg et al., 2023). Given this complexity, individuals may rely on heuristics or mental

81 shortcuts and overestimate their comprehension, cultivating an illusion of being well-

82 informed.

83 2.1.1 Why It Matters in Climate Change Communication

84 The illusion of knowing matters in climate change communication as it can shape

85 attitudes and behaviors towards mitigation efforts, potentially impacting the effectiveness of

86 strategies promoting informed public engagement and actions (Trémolière & Djeriouat, 2021;

87 Yang et al., 2020a, 2020b). The Positive Illusion Model suggests that despite the inaccuracies,

88 this illusion in self-evaluation can lead to positive outcomes by bolstering motivations,

89 promoting certain behaviors, and encouraging heightened engagement in discussions and

90 debates, potentially fostering a conducive environment for collective climate action (Betts et

91 al., 2009; Taylor & Armor, 1996; Yang et al., 2020a, 2020b).

92 However, the illusion of knowing may also present challenges in climate change
 5

93 communication and public policy. Individuals with an illusion of knowledge might be less

94 open to new information, leading to the propagation of misinformation or resistance to

95 adopting climate-friendly behaviors. This could hinder efforts to enhance public

96 understanding and promote meaningful engagement on climate change issues. Moreover,

97 individuals who overestimate their understanding of climate change may lack a

98 comprehensive grasp of the necessity or intricacies of certain policies, potentially affecting

99 their level of support or adherence, thus hindering the effective implementation of climate

100 policies.

101 In sum, while the Positive Illusion Model sheds light on the potential advantages of

102 illusory knowledge, the prevalence and potential impacts of the illusion of knowing warrant

103 more scholarly attention in climate change communication research (Betts et al., 2009; Taylor

104 & Armor, 1996; Trémolière & Djeriouat, 2021; Yang et al., 2020a, 2020b). With a better

105 understanding of this illusion and its impacts, policymakers and communicators can devise

106 more effective strategies to promote informed public engagement and support for climate

107 change mitigation efforts. The next section reviews previous research and gaps in

108 understanding the illusion of knowing in the context of climate change.

109 2.1.2 Previous Research on the Illusion of Knowing about Climate Change

110 Yang et al. (2020a, 2020b) have delineated the cognitive, media-related, and

111 demographic factors that contribute to the illusion of knowing in the context of climate

112 change. They found a positive relationship between individuals’ attention to climate change

113 messages in media, elaboration of these messages, and the illusion of knowing, with

114 elaboration moderating the relationship between media attention and the illusion of knowing

115 (Yang et al., 2020a). Additionally, Yang et al. 2020b identified demographic disparities in

116 climate change knowledge and illusion distributions, with males exhibiting more actual

117 knowledge, while less educated individuals, those from low-income households, and middle-
 6

118 aged and elderly adults had more illusion of knowing but less factual knowledge about

119 climate change compared to their counterparts.

120 Despite the potential dual effects of the illusion of knowing in the context of climate

121 change, there is a scarcity of research examining its consequences, which stifles further

122 theoretical development and the establishment of evidence-based policies and practices.

123 Trémolière and Djeriouat (2021) addressed this gap by investigating the role of analytic

124 cognitive style, knowledge, overconfidence, and political partisanship, revealing a significant

125 relationship between overconfidence in climate change knowledge and climate change

126 skepticism. However, the paucity of research and their cross-sectional design highlight the

127 importance and relevance of the current study.

128 By exploring the outcomes of the illusion of knowing in the climate change scenario

129 using a longitudinal design, this study aims to enrich the understanding of this phenomenon

130 and its consequences, potentially guiding future research, theory-building, and policy

131 development. As the illusion of knowing stems from individuals’ inaccurate estimation of

132 factual knowledge about climate change, the following section reviews the Knowledge

133 Deficit Model and the role of factual knowledge in climate change communication.

134 2.2 Knowledge Deficit Model

135 The Knowledge Deficit Model, also known as the Information Deficit Model, is a

136 one-way communication model and has been an integral part of science communication

137 research and practice (Davies, 2008; Hetland, 2014; Suldovsky, 2017; Tøsse, 2013). This

138 model posits that information flows from experts to the public, and gaining knowledge could

139 change individuals’ attitudes, beliefs, or behaviors regarding a particular topic, such as

140 climate change. At the core of this model lies the assumption that a deficit in understanding

141 or knowledge is the primary reason for insufficient concern or engagement in addressing the

142 issue. Also, this model assumes that the public is inherently either ignorant of or hostile
 7

143 towards science and technology (Hansen et al., 2003; Suldovsky, 2017)

144 2.2.1 Previous Mixed Findings Regarding the Model

145 The historical basis of science communication on the Knowledge Deficit Model has

146 been critically re-evaluated on empirical evidence (Hetland, 2014; Suldovsky, 2017; Yang et

147 al., 2020b). Three major counterarguments have been directed toward this model. First, an

148 oversimplification of the issue is challenged, with numerous studies highlighting that other

149 factors, including social, political, affective, cognitive, and cultural aspects, significantly

150 impact individuals’ attitudes and actions (Hansen et al., 2003; Poortinga et al. 2011;

151 Suldovsky, 2017; Tøsse, 2013). Second, many studies have found a weak or non-significant

152 relationship between increased knowledge and subsequent behavioral intentions (Fabrigar et

153 al., 2006; Holland et al., 2007; Suldovsky, 2017). Lastly, the model has been criticized for

154 unfairly portraying those opposed to scientific activities as lacking knowledge or ill-informed

155 (Priest, 2001; Suldovsky, 2017).

156 Specific to the climate change context, various studies have noted the limitations of

157 applying the model (e.g., Hart & Nisbet, 2012; Moser, 2016; Poortinga et al., 2011;

158 Suldovsky, 2017). For instance, research conducted by Poortinga et al. (2011) in Britain

159 revealed that climate skepticism is significantly tethered to individuals’ core values and

160 worldviews rather than just a deficiency in information concerning climate change. Similarly,

161 Hart and Nisbet (2012) identified that using the deficit model in climate change

162 communication can induce a boomerang effect among those ideologically opposed to climate

163 change. This scenario presents a particular challenge, given that communication efforts are

164 often directed toward these skeptical segments of the public.

165 Despite the consensus on the limitations of the Knowledge Deficit Model, two

166 significant gaps remain: the reliance on cross-sectional evidence, which only allows for

167 inferring correlations and limits understanding of the directional relationships between
 8

168 knowledge, attitudes, and behaviors over time, and the focus on predicting behavioral

169 intentions or behaviors at a single point rather than actual changes over time. The following

170 section draws insights from contemporary social marketing research (David & Rundle-

171 Thiele, 2019; Kriznik et al., 2018) to argue why behavior and change of behavior are distinct

172 constructs, emphasizing the need to shift focus toward understanding behavioral changes. It

173 also explores how knowledge and illusion of knowledge could potentially drive these

174 changes, laying the groundwork for revising the Knowledge Deficit Model with fresh

175 perspectives.

176 2.3 Changes in Pro-Environmental Behaviors

177 In communication research, even social sciences research in general, the prevailing

178 conceptual and methodological emphasis is placed on behavior, such as smoking, health

179 information seeking, or engaging in pro-environmental behaviros at a particular time rather

180 than behavioral change over time (e.g., Chen & Yang, 2019; Ma et al., 2023b; Shi et al.,

181 2023). However, many communication studies and practices aim to instigate behavioral

182 change, in essence, such as encouraging individuals to quit smoking, seek more health

183 information, or engage more in pro-environmental behaviros than before. Given these goals,

184 it is surprising that a scant amount of research attention is directed toward understanding

185 behavioral change (Rundle-Thiele et al., 2019).

186 2.3.1 The Distinction between Behavior and Change of Behavior

187 To understand the difference between behavior and behavioral change and to

188 recognize the value of this distinction, it is essential first conceptually to delineate the two.

189 Behavior can be viewed as a specific action executed at a given time. An example is the

190 Theory of Planned Behaviour (TPB, Ajzen, 1991), a springboard from which numerous

191 communication theories have developed. The TPB suggests that attitudes, subjective norms,

192 and perceived behavioral control collectively contribute to intentions that drive behavior
 9

193 (Ajzen, 1991). Nonetheless, Ajzen (2015) acknowledges that while TPB is a helpful

194 framework for interventions, it does not qualify as a theory of behavioral change since the

195 TPB was crafted to explain and predict intentions and behavior at a single point in time.

196 Behavioral change examines behaviors across multiple instances in time. It is defined

197 as altering the targeted behavior over time (Sarafino, 1996; Sundel & Sundel, 2017). The core

198 difference between static and dynamic behavior is temporal (David & Rundle-Thiele, 2019;

199 Rundle-Thiele et al., 2019; Ployhart & Vandenberg, 2010). Behavioral change is often

200 operationalized by calculating a change (or difference) score. This score serves as an

201 estimation of actual change, determined by the difference (T2-T1) between time point 1 (T1)

202 and time point 2 (T2). For instance, researchers may be interested in discerning whether the

203 public engages in more pro-environmental behaviors following exposure to specific climate

204 change information compared to their actions before (i.e., desired behavior change).

205 David and Rundle-Thiele (2019) explored data from 161 caregivers reporting changes

206 in their child’s walking behavior after a month-long social marketing program. They found

207 that while intentions and barriers explained behavior at a single time point (i.e., T2),

208 injunctive norms explained the change in walking behavior over time (i.e., T2-T1). Similarly,

209 David et al. (2019) investigated a food waste program aiming to reduce fruit and vegetable

210 waste through a two-week intervention. They identified attitudes and self-efficacy as possible

211 driving factors behind the transition of 43.8% of participants from fruit and vegetable wasters

212 to non-wasters post-intervention.

213 Past research has primarily focused on predicting behaviors using cross-sectional

214 designs (Gustafson & Rice, 2020; Ma et al., 2023a; van Giesen et al., 2018), which falls short

215 of offering insights into the antcedents of behavior change over time. These time-invariant

216 methodologies overlook changes in behaviors and cannot establish the time order among

217 variables of interest. Due to the conceptual and operational differences between behavior and
 10

218 behavior change, factors influencing behavior at a singular time may differ from those

219 driving behavior change (Rundle-Thiele et al., 2019; Ployhart & Vandenberg, 2010). The

220 following review delves into the concept of factual knowledge change stemming from these

221 behavioral changes.

222 2.4 Change of Factual Knowledge as A New Concept

223 Introducing the concept of change in factual knowledge as a new variable in our study

224 is a pivotal step to deepen the understanding of knowledge acquisition and its subsequent

225 influence on pro-environmental behavioral changes. This initiative stems from a similar logic

226 applied to the concept of behavioral change in the preceding section (David & Rundle-Thiele,

227 2019; Rundle-Thiele et al., 2019). The essence of this new variable lies in capturing the

228 evolution of knowledge over time, which potentially plays a critical role in shaping

229 individuals’ attitudes and behaviors toward environmental protection. Unlike a static

230 measurement of knowledge at a single time, the notion of change in factual knowledge aims

231 to capture the progression or regression of individuals’ understanding of climate change.

232 Examining factual knowledge changes over time, instead of merely focusing on

233 knowledge levels at a single time, opens up multiple benefits (Diamond & Urbanski, 2022;

234 Gustafson & Rice, 2020; Milfont, 2012). It accurately depicts how individuals’ understanding

235 matures and adapts. Also, this longitudinal lens allows us to move beyond a snapshot view to

236 a more flowing narrative of knowledge acquisition and adjustment. Moreover, it aligns with

237 the real-world scenario where individuals continually interact with new information, reassess

238 their understandings, and modify their actions accordingly.

239 2.4.1 Illusion of Knowing and Change of Factual Knowledge

240 The illusion of knowing could positively drive factual knowledge changes over time

241 by various mechanisms (Milfont, 2012). The first one is competence. The illusion of knowing

242 may initially motivate one to engage in climate change. This motivation could lead to
 11

243 pursuing more factual knowledge over time as individuals seek to align their perceived and

244 actual knowledge levels (Gifford & Comeau, 2011). Also, the illusion of knowing might

245 propel individuals into discussions, debates, or further reading to defend their perceived

246 knowledge, inadvertently leading to more factual knowledge over time (Moser, 2014; van der

247 Linden et al., 2015).

248 In addition, according to the positive illusion model, the illusion of knowing could

249 foster a sense of self-efficacy and control, encouraging individuals to engage more with

250 environmental topics (Betts et al., 2009; Taylor & Armor, 1996; Yang et al., 2020b). This

251 engagement may, in turn, lead to a gradual accumulation of factual knowledge over time.

252 Therefore, we present the first hypothesis below:

253 H1: The illusion of knowing in time 1 is positively related to changes in factual

254 knowledge.

255 2.5 Factual Knowledge and Change of Pro-Environmental Actions

256 Factual knowledge may significantly contribute to the changes in pro-environmental

257 actions over time rather than merely influencing static behaviors for two reasons. First,

258 acquiring factual knowledge may catalyze change, prompting them to re-evaluate their

259 actions and possibly adopt more environmentally friendly practices. However, the impact of

260 factual knowledge may not be typically immediate or direct on a specific behavior. Instead, it

261 may foster a gradual change, nurturing a more informed and conscious approach to pro-

262 environmental actions as individuals continue to learn and adapt over time.

263 On the other hand, simply possessing factual knowledge at a single point in time may

264 not directly translate to pro-environmental behaviors due to various intervening factors such

265 as personal beliefs and social norms (Hansen et al., 2003; Poortinga et al., 2011; Suldovsky,

266 2017; Tøsse, 2013). The knowledge might not be enough to override these factors and lead to

267 a change in behavior at that particular moment. However, as individuals assimilate this
 12

268 knowledge, reflect on it, and perhaps encounter supportive social contexts or additional

269 information, they may alter their actions towards more pro-environmental practices over

270 time.

271 Exploring the relationship between factual knowledge at an initial time and the

272 changes in pro-environmental actions helps advance the Knowledge Deficit Model with a

273 new perspective, which shifts the examinations from merely understanding the immediate

274 effects of knowledge on behavior to a more profound comprehension of how knowledge can

275 instigate a continuous change in pro-environmental actions over time. Therefore, we present

276 the hypothesis 2:

277 H2: Factual knowledge in time 1 positively relates to the changes in pro-

278 environmental intentions.

279 2.6 Illusion of Knowing and Change of Pro-Environmental Actions

280 The concept of the illusion of knowing may uniquely impact the change of pro-

281 environmental actions over time. Based on the Positive Illusion Model, the illusion of

282 knowing could foster a heightened sense of control and self-efficacy regarding one’s ability

283 to contribute to environmental conservation (Betts et al., 2009; Taylor & Armor, 1996; Yang

284 et al., 2020b). This increased self-efficacy may fuel the motivation to initiate and sustain

285 changes in pro-environmental actions over time (Taylor & Brown, 1994).

286 Also, individuals with the illusion of knowing may adjust their behaviors to align with

287 their perceived knowledge about environmental issues. For example, suppose someone

288 erroneously believes they know a lot about recycling. In that case, they might adjust their

289 recycling behaviors to match this perceived knowledge, possibly leading to more

290 environmentally friendly behaviors over time as they seek consistency (Gehlbach et al., 2019;

291 Stoutenborough et al., 2016). Therefore, we present the following hypothesis 3.

292 H3: The illusion of knowing in time 1 positively relates to changes in pro-
 13

293 environmental intentions.

294 Following the discussion of our core hypotheses, we will transition in the next section

295 to explore pro-environmental self-identity as a moderating mechanism.

296 2.7 Pro-Environmental Self-Identity as a Grouping Variable

297 Self-identity is forged by the descriptors individuals choose for themselves, shaped by

298 personal drives and social interactions, encompassing the anticipations of others and the roles

299 we undertake (Stryker & Burke, 2000). Pro-environmental self-identity encapsulates an

300 individual’s self-view as someone who engages in environmentally responsible behaviors,

301 and can sway one’s attitudes and actions, propelling them to act accordingly with this self-

302 image (Dermody et al., 2015; van der Werff et al., 2013; Whitmarsh & O’Neill, 2010).

303 The moderating role of pro-environmental self-identity in shaping pro-environmental

304 attitudes and intentions has been acknowledged in various studies (Carfora et al., 2017;

305 Tarinc et al., 2023). For instance, Carfora et al. (2017) found that pro-environmental self-

306 identity significantly moderated the impact of perceived behavioral control on intentions and

307 the influence of past behavior on both intentions and actions across pro-environmental

308 behaviors using a two-wave longitudinal design. The influence of pro-environmental self-

309 identity as a moderating variable is intriguing, given its potential to impact the relationships

310 delineated in our hypotheses (Carfora et al., 2017; Whitmarsh & O'Neill, 2010), and how

311 self-identity interacts with the elements of each hypothesis could vary based on the degree of

312 pro-environmental self-identity an individual embodies.

313 Specifically for Hypothesis 1, which posits that the illusion of knowing in time 1 is

314 positively related to the change of factual knowledge, the mechanism may be that individuals

315 with a high pro-environmental self-identity may be driven by their values and self-concept to

316 seek information and update their climate change knowledge (Dermody et al., 2015;

317 Whitmarsh & O’Neill, 2010). This drive might lead them to correct their illusion of knowing
 14

318 over time as they gather more factual knowledge through exposure and learning. However,

319 individuals with a low pro-environmental self-identity may lack the motivation to amend

320 their understanding, thus exhibiting a weaker relationship between the illusion of knowing

321 and the change in factual knowledge over time.

322 Specifically for Hypothesis 2, which posits that factual knowledge at time 1 is

323 positively related to changes in pro-environmental intentions, the mechanism could work as

324 follows: individuals with a strong pro-environmental self-identity might be more apt to

325 translate their factual knowledge into changes in pro-environmental actions due to an internal

326 sense of responsibility towards environmental protection (van der Werff et al., 2013;

327 Whitmarsh & O’Neill, 2010). Conversely, those with a weaker pro-environmental self-

328 identity may lack the motivation to alter their actions despite possessing similar levels of

329 factual knowledge, as their internal identification with environmentally responsible behaviors

330 is not as pronounced.

331 For Hypothesis 3, which suggests that the illusion of knowing in time 1 is positively

332 related to the change in pro-environmental intentions, the mechanism could be that

333 individuals with a high pro-environmental self-identity may find their illusion of knowing

334 bolstering their confidence regarding climate change understanding, thereby encouraging

335 their engagement in more pro-environmental actions. On the other hand, individuals with a

336 low pro-environmental self-identity may not exhibit a significant change in pro-

337 environmental actions driven by the illusion of knowing, given their inherently lower

338 motivation to address climate change.

339 Following the discussion on the moderating role of pro-environmental self-identity,

340 we present the following research question 1 and Figure 1 for all hypotheses and the research

341 question 1:

342 RQ1: Are there any differences between the Chinese public’s high and low levels of
 15

343 pro-environmental self-identity in the relationships among the illusion of knowing,

344 factual knowledge, and changes of factual knowledge, and pro-environmental

345 intentions?

346 Figure 1

347 Conceptualized Model in the Current Research

348

349 3. Method

350 3.1 Procedures and Participants

351 A longitudinal study design was employed, comprising two waves of data collection.

352 A three-week interval was established before conducting the second wave to mitigate the

353 potential influence of memory on responses and minimize attrition. The data were gathered

354 via an online survey administered through Credamo, which provides access to a diverse panel

355 of more than 1.5 million adults residing in China.

356 The study consisted of two waves of data collection, with the initial wave conducted

357 from November 20 to December 3, 2021, and the follow-up survey from December 22, 2021,

358 to January 3, 2022. A total of 891 participants provided their demographic information and
 16

359 responses in the first wave. The research protocol was approved by the Research Ethics

360 Committee at the institution overseeing the study.

361 In the second wave, 634 participants returned, and after stringent data cleaning

362 procedures, including removing participants who failed attention check questions or provided

363 inconsistent demographic data across the two waves, a final sample of 427 participants was

364 obtained, corresponding to a retention rate of approximately 48%. The final sample

365 comprised 59% female participants, with an average age of 29.30 years (SD = 6.39),

366 predominantly well-educated (71% with bachelor’s degrees), and a median monthly

367 household income between 10,000 to 20,000 CNY (approximately USD 1,572 to USD

368 3,144).

369 3.2 Measures

370 3.2.1 Changes in Intentions to Engage in Pro-environmental Actions

371 Participants’ intentions to undertake pro-environmental actions were assessed using

372 an 8-item scale from Bradley et al. (2020). Participants rated their likelihood to engage in

373 each behavior on a 5-point scale, with 1 representing “very unlikely” and 5 representing

374 “very likely.” The behaviors focused on two main areas: transportation (including walking or

375 cycling, using public transportation, driving electric vehicles, and minimizing unnecessary

376 travel) and waste management (including going paperless, using recyclable bags, avoiding

377 disposable plastic containers, and recycling).

378 Consistent with prior research, the individual scores across the eight behaviors were

379 aggregated to form an overall intention score for pro-environmental actions (Bradley et al.,

380 2020; Shi et al., 2023). Mean intention scores were calculated at both time points (MT1 =

381 34.14, SDT1 = 3.14; MT2 = 33.80, SDT2 = 3.05). To determine changes in pro-environmental

382 behavioral intentions, the scores from time 2 were subtracted from those at time 1, yielding a

383 change score (M = - 0.04, SD = 0.34).

 17

384 3.2.2 Factual Knowledge of Climate Change

385 To measure participants’ factual knowledge about climate change, a set of 5 items was

386 employed (Hart et al., 2015; Yang et al., 2020a, 2020b). Respondents were presented with

387 statements about climate change and asked to judge their accuracy using a 3-point scale: “1 =

388 true”, “2 = false”, “0 = don’t know.” Sample statements included misconceptions and facts

389 such as the role of the ozone layer in global warming, the impact of solar output compared to

390 human-produced greenhouse gases, and the effect of climate change on sea and ocean water

391 levels. The respondents’ scores were averaged across all items to produce a knowledge score,

392 reflecting the proportion of correct responses from the total items presented.

393 3.2.3 Change of Factual Knowledge

394 To calculate the change in factual knowledge about climate change, we subtracted the

395 score of factual knowledge at time 1 from the score at time 2. This difference score reflects

396 the participants’ factual knowledge changes over the study period. A positive score indicates

397 an increase in knowledge, whereas a negative score suggests a decrease. The mean change in

398 factual knowledge across participants was minimal (M = 0.01, SD = 0.19), indicating a slight

399 overall increase in actual knowledge of climate change issues between the two-time points.

400 3.2.4 Illusion of Knowing Regarding Climate Change

401 The illusion of knowing in the context of climate change is identified when an

402 individual’s perceived knowledge exceeds their actual knowledge. To measure this

403 phenomenon, we asked participants to self-assess their knowledge about climate change on a

404 10-point scale, with 0 representing “knowing nothing” and 10 representing “knowing

405 everything.” The perceived knowledge scores were converted to a percentage format to align

406 the factual and perceived knowledge scales for comparative analysis. This format mirrors the

407 factual knowledge measurement, thus allowing a direct comparison between perceived and

408 actual knowledge levels. The illusion of knowing at time 1 was calculated by deducting the
 18

409 factual knowledge score from the perceived knowledge score, yielding a mean illusion of

410 knowing score (M = 0.04, SD = 0.20).

411 3.2.5 Pro-Environmental Self-Identity

412 We utilized a four-item scale adapted by Whitmarsh and O’Neill’s (2010) work to

413 measure pro-environmental self-identity. Responses were gathered on a 7-point Likert scale,

414 including affirmative statements including “I think of myself as an environmentally-friendly

415 consumer” and “I am very concerned with environmental issues,” as well as reversed items

416 such as “I would be embarrassed to be seen as environmentally friendly” and “I don’t want to

417 be viewed by family or friends as someone concerned about environmental issues.” After

418 reversing the necessary items, higher scores corresponded to a stronger identification with

419 pro-environmental behaviors. The average score for our participants was moderately high (M

420 = 4.39, SD = 0.45, α = 0.78). We then categorized this variable into two groups: those scoring

421 above the mean were placed in the high identity group, while those scoring below the mean

422 were classified as having a low pro-environmental self-identity (high group: n = 249; low

423 group: n = 178).

424 3.3 Analytical Procedure

425 The Cross-Lagged Panel Model (CLPM) is a widely used approach for analyzing

426 longitudinal data, particularly when the research aims to examine the directional influences

427 between variables over time (Hamaker et al., 2015; Newsom, 2015). The CLPM allows for

428 the simultaneous estimation of autoregressive effects (the effect of a variable on itself at a

429 later time point) and cross-lagged effects (the effect of one variable on another variable at a

430 later time point) while controlling for the stability of the constructs (Hamaker et al., 2015).

431 However, in the current study, the use of CLPM was not feasible due to the nature of

432 the variables and the research objectives. First, one of the key variables in the study, change

433 in factual knowledge, was computed by subtracting the knowledge scores at time 1 from the
 19

434 scores at time 2. This change score does not exist as a separate variable at each time point,

435 making it impossible to include in a CLPM analysis, which requires all variables to be

436 measured at each wave. Second, the primary outcome variable of interest, change in pro-

437 environmental actions, was only measured at time 2, as it represents the difference in pro-

438 environmental actions between the two-time points. Since this variable does not have a

439 corresponding measure at time 1, it cannot be incorporated into a CLPM model, which

440 assumes that all variables are assessed at each time point (Hamaker et al., 2015).

441 Given these limitations and the study’s focus on predicting changes in pro-

442 environmental actions using variables from time one, a path analysis approach was deemed

443 more appropriate for addressing the research questions. We utilized path analyses in Mplus

444 version 8.3 to address our research question and hypotheses. We first constructed a base

445 model to test the three hypotheses. Then, to explore RQ1, we performed a multi-group path

446 analysis. Notably, since the change of factual knowledge is derived from the knowledge

447 scores, we assumed a correlation between these two variables in our analyses. Also, a Pearson

448 correlation matrix for all involved variables in the study is presented in Table 1.

449 The models’ fits were assessed using several indicators: the χ2 goodness-of-fit

450 statistic, the comparative fit index (CFI), the Tucker–Lewis index (TLI), the standardized root

451 mean square residual (SRMR), and the root mean square error of approximation (RMSEA).

452 Acceptable thresholds for a good model fit were set at RMSEA ≤ .06, CFI and TLI ≥ .95, and

453 SRMR < .08, following Hu and Bentler’s (1999) recommendations.

454 Table 1

455 Correlation Matrix for Study Variables

Variables 1 2 3 4 5 6 7 8
1. Age -
2. Gender -.222 -
3. Education .031 .048 -
 20

4. Income .395 -.09 .137 -

1
5. Factual Knowledge .095 -.119 -.06 .040 -
3
6. Illusion of Knowing .078 .023 .050 .055 -.82 -
0
7. Change of Factual -.049 .064 -.04 .114 -.48 .43 -
Knowledge 5 5 6
8. Changes in Pro- -.004 -.03 .025 -.01 -.04 .02 .118 -
Environmental Behavioral 9 2 2 7
Intentions
9. Pro-Environmental Self- .315 -.05 .002 .159 .148 .13 .01 -.05
Identity Group 6 1 9 3
456

457 4. Results

458 4.1 Results for the Base Model

459 The fit indices reported suggest a perfect fit of the model, with perfect scores for CFI

460 and TLI at 1.000 and an RMSEA and SRMR of 0.000. This is probably due to the model’s

461 simplicity, as it contains a limited number of variables and uses path analysis, often resulting

462 in better-fit indices, consistent with previous research that has a similar number of included

463 variables (Ma et al., 2023c). The Chi-Square Test of Model Fit value is significant, common

464 in large samples, and does not necessarily reflect a poor model fit.

465 As shown in Figure 2, the illusion of knowing was a significant predictor for the

466 change in factual knowledge, with a standardized coefficient of .441, p < .001, indicating that

467 a greater perceived understanding of climate change is associated with increased knowledge

468 over time. Age negatively predicted the change in factual knowledge (β = −.134, p = .005),

469 suggesting younger participants showed more significant changes in factual knowledge.

470 Furthermore, income level was positively related to the change in factual knowledge (β
 21

471 = .159, p = .001), indicating that higher income levels correspond with more substantial

472 changes in knowledge. Education level also significantly influenced the change in factual

473 knowledge (β = −.087, p = .043), albeit to a lesser extent.

474 Figure 2

475 Standardized Coefficients for the Base Model

476

477 Note. *** indicates p ≤ .001.

478 In contrast, changes in pro-environmental behavioral intentions were not significantly

479 predicted by the illusion of knowing (β = −.028, p = .753) or by factual knowledge at time 1

480 (β = −.068, p = .444), suggesting that neither the perceived nor the actual initial knowledge

481 about climate change were influential in changing intentions to engage in pro-environmental

482 behaviors. Additionally, all the demographic variables, including age, gender, education, and
 22

483 income, did not significantly impact changes in pro-environmental behavioral intentions.

484 4.2 Results for the Multi-Group Model

485 Similar to the base model, the reported fit indices for the multi-group model indicate

486 an exceptional fit for the structural model, with perfect scores for CFI and TLI at 1.000 and

487 an RMSEA and SRMR of 0.000. The Chi-Square Test of Model Fit value is significant,

488 common in large samples, and does not necessarily reflect a poor model fit.

489 4.2.1 Results for the Low Pro-Environmental Self-Identity Group

490 As shown in Figure 3, for individuals with a low pro-environmental self-identity, the

491 illusion of knowing is significantly associated with the change in factual knowledge (β

492 = .423, p < .001). However, the illusion of knowing did not significantly predict changes in

493 pro-environmental behavioral intentions (β = -.234, p = .087), showing a negative trend that

494 did not reach statistical significance. Factual knowledge at time 1 was a significant negative

495 predictor of changes in pro-environmental behavioral intentions (β = -.274, p = .043),

496 suggesting that those with lower factual knowledge initially were more likely to change their

497 intentions to engage in pro-environmental behaviors. Demographic factors did not show

498 significant effects on either of the dependent variables for this group.

499 Figure 3

500 Standardized Coefficients for the Low Group Model

 23

501

502 Note. * indicates p < .05. *** indicates p ≤ .001.

503 4.2.3 Results for the High Pro-Environmental Self-Identity Group

504 As shown in Figure 4, for those with a high pro-environmental self-identity, the

505 results indicate that the illusion of knowing had a strong positive relationship with the change

506 in factual knowledge (β = .443, p < .001). This finding suggests that individuals who believed

507 they knew more about climate change were more likely to increase their knowledge over

508 time. Age was negatively associated with the change in factual knowledge (β = -.174, p

509 = .004), with younger individuals showing a more substantial increase in knowledge. Higher-

510 income levels were positively associated with the change in factual knowledge (β = .233, p

511 < .001), indicating that wealthier individuals experienced more significant knowledge

512 changes. Education level also negatively influenced the change in factual knowledge (β =
 24

513 -.183, p = .001), with those with lower education levels showing greater knowledge gains.

514 Figure 4

515 Standardized Coefficients for the High Group Model

516

517 Note. *** indicates p ≤ .001.

518 Regarding changes in pro-environmental behavioral intentions, the illusion of

519 knowing was a significant positive predictor (β = .470, p < .001), indicating that those with a

520 greater illusion of knowing of climate change were more likely to change their intentions

521 towards pro-environmental actions. Similarly, factual knowledge at Time 1 was a significant

522 positive predictor of changes in pro-environmental behavioral intentions (β = .424, p = .001),

523 showing that individuals with higher initial knowledge were more inclined to change their

524 intentions to engage in pro-environmental behaviors. For this group, demographic factors,
 25

525 including age, gender, education, and income, did not significantly predict changes in pro-

526 environmental behavioral intentions.

527 Overall, in our multi-group comparison, the coefficients from the illusion of knowing

528 to change of factual knowledge were almost identical across groups, at .443 for the high

529 identity group and .423 for the low. Conversely, the pathways from the illusion of knowing

530 and factual knowledge to changes in pro-environmental intentions differed significantly, with

531 a positive relationship in the high self-identity group and a negative in the low. This

532 divergence in directionality is so pronounced that formal statistical tests to establish

533 significant differences between these two paths between the groups are unnecessary and

534 redundant.

535 5. Discussion

536 5.1 Illusion of Knowing as a Motivator for Changing the Current Knowledge

537 Based on a longitudinal investigation, our study found a positive correlation between

538 the illusion of knowing about climate change and knowledge change over time, regardless of

539 individuals’ environmental identity. This result aligns with the Positive Illusion Model

540 (Taylor & Armor, 1996) and is consistent with prior cross-sectional evidence with the notion

541 that the illusion of knowing can propel individuals to engage with the focal issue, thus

542 leading to more knowledge (Milfont, 2012; Moser, 2014; van der Linden et al., 2015).

543 Similarly, our research indicates that possessing the illusion of knowing may contribute to

544 individuals’ perceived capacity to confidently engage in related activities such as campaigns,

545 interpersonal discussions, and media exposure, ultimately leading to accumulating more

546 knowledge than before.

547 Previous studies have argued that trust beliefs and enhanced self-efficacy prompt

548 individuals to seek more information, acquire knowledge, and develop skills (Artino, 2012;

549 Ter Huurne & Gutteling, 2009). Additionally, early scholarship has suggested that individuals
 26

550 adjust their informational behaviour to avoid cognitive dissonance and maintain self-

551 perception (Bem, 1967), evidenced by increased learning and acquisition of climate change

552 knowledge resulting from the illusion of having mastered related knowledge. Hence, our

553 evidence longitudinally documents that change in factual knowledge, which was introduced

554 as a novel concept in this study, can be positively predicted by illusory knowledge.

555 Importantly, our findings demonstrate a consistent effect of the illusion of knowing on

556 knowledge change across two groups, elucidating that illusion motivates individuals to align

557 their factual and illusory knowledge about climate change, irrespective of their environmental

558 self-identity. This conclusion enriches the Positive Illusion Model from the perspective of

559 audience segmentation, encouraging environmental campaigners that aims to increase public

560 knowledge to target all audiences equally in their interventions.

561 5.2 Conditional Influence of Illusion of Knowing on Pro-Environmental Behavior

562 According to the results of our base model, the illusion of knowing overall exhibited a

563 null effect on the change of pro-environmental behavioral intention; however, the pattern of

564 this effect significantly differed between low-identity and high-identity groups. While prior

565 literature (Betts et al., 2009; Taylor & Armor, 1996) have identified a positive relationship

566 between illusion and behavior conduction, our findings suggest a nuanced relationship.

567 Unlike learning more about environmental facts, explicit, more pro-environmental behaviors

568 demand higher internal motivations. Individuals with high environmental identity, seeking

569 action-perception consistency (Bem, 1967; Zanna et al., 1981), are likely motivated by

570 environmental responsibilities and values to engage in pro-environmental behaviors.

571 Additionally, in line with the TPB (Ajzen, 1991), perceived behavioral control functions as a

572 crucial antecedent for forming behaviors. In our context, it is reasonable to expect individuals

573 who self-identify as environmentalists to understand correct approaches better and have

574 access to possible solutions (i.e., efficacy) to individually and collectively mitigate
 27

575 environmental problems (Shi et al., 2023). Thus, this study contributes to a deeper

576 understanding of how the illusion of knowing affects individuals’ pro-environmental

577 behaviors, considering their environmental identity.

578 5.3 High Environmental Self-Identity Help Translate Knowledge into Actions

579 Another hypothesis of this study focused on the longitudinal relationship between

580 factual knowledge levels and subsequent behavior change. The findings of our base model

581 revealed an insignificant relationship between factual knowledge at T1 and the subsequent

582 change in pro-environmental behavioral intention at T2. This result supports the notion of

583 ‘cognitive misers’ (Glenberg et al., 1982; Orbell & Dawes, 1991; Yang et al., 2020a, 2020b),

584 suggesting that holding heuristics and mental shortcuts play a significant role in forming

585 opinions and decision-making, rather than science literacy or knowledge deficit (Davies,

586 2008; Miller, 1992).

587 Our multi-group analysis revealed that environmental identity moderated the

588 relationship between knowledge and pro-environmental behavior over time. For individuals

589 with high environmental identity, their knowledge about climate change positively related to

590 pro-environmental behavior change. In contrast, knowledge at T1 negatively associated with

591 subsequent behavior change at T2 among low-identity individuals. This contradictory pattern

592 suggests that environmentally knowledgeable individuals with high identity are more inclined

593 to mitigate environmental hazards due to their awareness of climate change causes and

594 consequences, with knowledge and identity becoming the impetus for pro-environmental

595 actions.

596 However, for knowledgeable individuals indifferent about the environment (low

597 identity), their knowledge might translate into a negative belief in changing the environment,

598 decreasing their pro-environmental actions and viewing climate change as unchangeable.

599 This aligns with prior studies indicating that individuals with greater science literacy and
 28

600 education may hold more polarized beliefs on controversial science topics (Drummond &

601 Fischhoff, 2017).

602 Our findings suggest that factual knowledge can only positively influence pro-

603 environmental action promotion when individuals' self-identity aligns with the issue. This

604 implies that scientific knowledge alone may not sufficiently influence or change beliefs and

605 behavior about climate change; future research and environmental campaigns should consider

606 incorporating factors such as self-identity.

607 5.4 Implications and Limitations

608 As the first study to unveil the longitudinal relationship between the illusion of

609 knowing and changes in factual knowledge, as well as pro-environmental actions over time,

610 our findings carry multiple implications. Theoretically, the Positive Illusion Model (Taylor &

611 Armor, 1996) has been tested and validated in the context of climate change. Our research

612 documents the desirability of the illusion of knowing in promoting factual knowledge about

613 climate change. Additionally, our conclusions demonstrate an interaction between two crucial

614 sound models in science communication—the Cognitive Miser Model and the Knowledge

615 Deficit Model—advancing the debatable evidence of previous studies. Our multi-group

616 analysis, considering individuals’ self-identity on environmentalism, provides a scaffold to

617 understand the conditional effect of illusion and factual knowledge on pro-environmental

618 behavior changes, enriching explanatory power from the perspective of audience

619 segmentation.

620 Practically, our findings suggest that cultivating the illusion of knowing about climate

621 change is conducive to the continuous accumulation of related knowledge through

622 engagement in online and offline interpersonal communication and environmental

623 campaigns. Thus, environmental practitioners are encouraged to cover the climate change

624 issue more frequently and proportionately to elevate the public’s confidence and efficacy
 29

625 derived from the illusion of knowing about climate change through strategic climate

626 communication (Yang et al., 2020a). This positive psychological solution has received much

627 scholarly attention recently (Ojala, 2023). Furthermore, given the vital characteristic of

628 environmental self-identity in the relationships between illusion, factual knowledge, and pro-

629 environmental behavior, future sustainable development interventions can boost public

630 environmental concerns and incorporate more environmentalists to yield the intended effects

631 of environmental campaigns. This includes leveraging their peer influence on social

632 networking sites as well as interpersonal interactions (Dalrymple et al., 2013)

633 In terms of limitations and suggestions for future research, several aspects should be

634 acknowledged. First, as an initial exploratory study, we focused solely on direct effects in the

635 path model. Future research should explore potential mediating and moderating processes

636 using multiple waves of data collection to enhance understanding of the mechanisms between

637 the illusion of knowing and its outcomes. Second, this study only considered intrapersonal

638 illusion (i.e., self-reported illusion of knowing). Researchers have noted that the illusion of

639 knowing in comparison with others also deserves scholarly attention (Yang et al., 2020a).

640 Including an interpersonal perspective might enable future studies to better understand how

641 illusory knowledge operates in a collective context between individuals. This aspect may be

642 related to collective risk perception and efficacy in climate change mitigation on both an

643 individual and societal level (Shi et al., 2023).

644 Third, considering the complexity of extracting pro-environmental behaviors in self-

645 report questionnaires, we only collected participants’ intentions to engage in individual and

646 collective pro-environmental actions. Although scholars argue for a strong link between

647 behavioral intention and actual behavior (Ajzen, 1991; Webb & Sheeran, 2006), future

648 studies are warranted to measure people’s pro-environmental behavior using alternative

649 methods (e.g., field observations, lab observations, or digital traces, Lange & Dewitte, 2019).

650
 30

651 References

652 Ajzen I (1991) The theory of planned behavior. Organizational Behavior and Human

653 Decision Processes 50(2): 179-211.

654 Ajzen I (2015) The theory of planned behaviour is alive and well, and not ready to retire: a

655 commentary on Sniehotta, Presseau, and Araújo-Soares. Health Psychology Review

656 9(2): 131-137.

657 Arbuckle JG, Prokopy LS, Haigh T, Hobbs J, Knoot T, Knutson C, Loy A, Mase AS,

658 McGuire J, Morton LW, Tyndall J and Widhalm M (2013) Climate change beliefs,

659 concerns, and attitudes toward adaptation and mitigation among farmers in the

660 Midwestern United States. Climatic Change 117(4): 943–950.

661 Artino AR Jr (2012) Academic self-efficacy: from educational theory to instructional

662 practice. Perspectives on Medical Education 1(2): 76–85.

663 Avhustiuk MM, Pasichnyk ID and Kalamazh RV (2018) The illusion of knowing in

664 metacognitive monitoring: Effects of the type of information and of personal,

665 cognitive, metacognitive, and individual psychological characteristics. Europe’s

666 Journal of Psychology 14(2): 317.

667 Bem DJ (1967) Self-perception: An alternative interpretation of cognitive dissonance

668 phenomena. Psychological Review 74(3): 183.

669 Betts LR, Rotenberg KJ and Trueman M (2009) An investigation of the impact of young

670 children’s self‐knowledge of trustworthiness on school adjustment: A test of the

671 realistic self‐knowledge and positive illusion models. British Journal of

672 Developmental Psychology 27(2): 405-424.

673 Bostrom A, Hayes AL and Crosman KM (2019) Efficacy, action, and support for reducing

674 climate change risks. Risk Analysis 39(4): 805–828.

675 Bradley GL, Babutsidze Z, Chai A and Reser JP (2020) The role of climate change risk
 31

676 perception, response efficacy, and psychological adaptation in pro-environmental

677 behavior: A two nation study. Journal of Environmental Psychology 68: 101410.

678 Brosowsky NP and Egner T (2021) Appealing to the cognitive miser: Using demand

679 avoidance to modulate cognitive flexibility in cued and voluntary task switching.

680 Journal of Experimental Psychology: Human Perception and Performance 47(10):

681 1329.

682 Carfora V, Caso D, Sparks P and Conner M (2017) Moderating effects of pro-environmental

683 self-identity on pro-environmental intentions and behaviour: A multi-behaviour study.

684 Journal of Environmental Psychology 53: 92-99.

685 Chen L and Yang X (2018) Using EPPM to evaluate the effectiveness of fear appeal messages

686 across different media outlets to increase the intention of breast self-examination

687 among Chinese women. Health Communication 31(11): 1369-1376.

688 Copsey T, Hoijtink L, Shi X and Whitehead S (2013) How the people of China live with

689 climate change and what communication can do. In Whitehead S and Wilson D (eds)

690 pp. 1–61.

691 Corcoran K and Mussweiler T (2010) The cognitive miser’s perspective: Social comparison

692 as a heuristic in self-judgements. European Review of Social Psychology 21(1): 78-

693 113.

694 Dalrymple KE, Shaw BR and Brossard D (2013) Following the leader: Using opinion leaders

695 in environmental strategic communication. Society & Natural Resources 26(12):

696 1438-1453.

697 David P and Rundle-Thiele S (2019) Rethinking behaviour change: a dynamic approach in

698 social marketing. Journal of Social Marketing 9(2): 252-268.

699 David P, Rundle-Thiele S and Pallant JI (2019) (Re) Focussing on behavioural change: an

700 examination of the utility of hidden Markov modelling. Journal of Social Marketing
 32

701 9(2): 130-145.

702 Davies SR (2008) Constructing communication: Talking to scientists about talking to the

703 public. Science Communication 29(4): 413-434.

704 Dermody J, Hanmer-Lloyd S, Koenig-Lewis N and Zhao AL (2015) Advancing sustainable

705 consumption in the UK and China: The mediating effect of pro-environmental self-

706 identity. Journal of Marketing Management 31(13-14): 1472-1502.

707 Diamond E and Urbanski K (2022) The Impact of Message Valence on Climate Change

708 Attitudes: A Longitudinal Experiment. Environmental Communication 16(8): 1046-

709 1058.

710 Drummond C and Fischhoff B (2017) Individuals with greater science literacy and education

711 have more polarized beliefs on controversial science topics. Proceedings of the

712 National Academy of Sciences 114(36): 9587-9592.

713 Fabrigar LR, Petty RE, Smith SM and Crites SL Jr (2006) Understanding knowledge effects

714 on attitude-behavior consistency: the role of relevance, complexity, and amount of

715 knowledge. Journal of Personality and Social Psychology 90(4): 556.

716 Gehlbach H, Robinson CD and Vriesema CC (2019) Leveraging cognitive consistency to

717 nudge conservative climate change beliefs. Journal of Environmental Psychology 61:

718 134-137.

719 Gifford R and Comeau LA (2011) Message framing influences perceived climate change

720 competence, engagement, and behavioral intentions. Global Environmental Change

721 21(4): 1301-1307.

722 Glenberg AM, Wilkinson AC and Epstein W (1982) The illusion of knowing: Failure in the

723 self-assessment of comprehension. Memory & Cognition 10(6): 597-602.

724 Gong X, Zhang J, Zhang H, Cheng M, Wang F and Yu N (2020) Internet use encourages pro-

725 environmental behavior: Evidence from China. Journal of Cleaner Production 256:
 33

726 120725.

727 Gustafson A and Rice RE (2020) A review of the effects of uncertainty in public science

728 communication. Public Understanding of Science 29(6): 614-633.

729 Hamaker EL, Kuiper RM and Grasman RP (2015) A critique of the cross-lagged panel model.

730 Psychological Methods 20(1): 102-116.

731 Hart PS and Nisbet EC (2012) Boomerang effects in science communication: How motivated

732 reasoning and identity cues amplify opinion polarization about climate mitigation

733 policies. Communication Research 39(6): 701-723.

734 Hart PS, Nisbet EC and Myers TA (2015) Public attention to science and political news and

735 support for climate change mitigation. Nature Climate Change 5(6): 541-545.

736 Hetland P (2014) Models in science communication: Formatting public engagement and

737 expertise. Nordic Journal of Science and Technology Studies 2(2): 5-17.

738 Hoekstra AG, Noordzij K, de Koster W, and van der Waal J. (2024). The educational divide

739 in climate change attitudes: Understanding the role of scientific knowledge and

740 subjective social status. Global Environmental Change, 86, 102851.

741 Holland EM (2007) The risks and advantages of framing science. Science 317(5842): 1168-

742 1170.

743 Hu LT and Bentler PM (1999) Cutoff criteria for fit indexes in covariance structure analysis:

744 Conventional criteria versus new alternatives. Structural Equation Modeling: A

745 Multidisciplinary Journal 6(1): 1-55.

746 Jørgensen SL and Termansen M (2016) Linking climate change perceptions to adaptation and

747 mitigation action. Climatic Change 138(1-2): 283–296.

748 Kriznik NM, Kinmonth AL, Ling T and Kelly MP (2018) Moving beyond individual choice

749 in policies to reduce health inequalities: the integration of dynamic with individual

750 explanations. Journal of Public Health 40(4): 764-775.

 34

751 Lange F and Dewitte S (2019) Measuring pro-environmental behavior: Review and

752 recommendations. Journal of Environmental Psychology 63: 92-100.

753 Lu H (2022) The role of repeated exposure and message fatigue in influencing willingness to

754 help polar bears and support climate change mitigation. Science Communication

755 44(4): 475-493.

756 Ma X, Ma L and Liu BF (2023a) Should relationships be at the heart of public relations? A

757 meta-analysis of the antecedents and consequences of organization-public

758 relationships. Public Relations Review 49(5): 102388.

759 Ma X, Yang Y and Chen L (2023b) Promoting behaviors to mitigate the effects of climate

760 change: Using the Extended Parallel Process Model at the personal and collective

761 level in China. Environmental Communication 17(4): 353-369.

762 Ma X, Yang Y, Lin T, Zhang Y and Zheng E (2023c) Loneliness, Purpose in Life, and

763 Protective Behaviors: Examining Cross-Sectional and Longitudinal Relationships in

764 Older Adults Before and During COVID-19. The Journals of Gerontology: Series B

765 gbad117.

766 Metag J and Schäfer MS (2018) Audience segments in environmental and science

767 communication: Recent findings and future perspectives. Environmental

768 Communication 12(8): 995-1004.

769 Michie S, Johnston M, Francis J, Hardeman W and Eccles M (2008) From theory to

770 intervention: mapping theoretically derived behavioural determinants to behaviour

771 change techniques. Applied Psychology 57(4): 660-680.

772 Milfont TL (2012) The interplay between knowledge, perceived efficacy, and concern about

773 global warming and climate change: a one‐year longitudinal study. Risk Analysis: An

774 International Journal 32(6): 1003-1020.

775 Miller JD (1992) Toward a scientific understanding of the public understanding of science
 35

776 and technology. Public Understanding of Science 1(1): 23.

777 Moser SC (2010) Communicating climate change: history, challenges, process and future

778 directions. Wiley Interdisciplinary Reviews: Climate Change 1(1): 31-53.

779 Moser SC (2014) Communicating adaptation to climate change: the art and science of public

780 engagement when climate change comes home. Wiley Interdisciplinary Reviews:

781 Climate Change 5(3): 337-358.

782 Moser SC (2016) Reflections on climate change communication research and practice in the

783 second decade of the 21st century: what more is there to say? Wiley Interdisciplinary

784 Reviews: Climate Change 7(3): 345-369.

785 Newsom JT (2015) Longitudinal Structural Equation Modeling: A Comprehensive

786 Introduction. Routledge.

787 Ojala M (2023) Hope and climate-change engagement from a psychological perspective.

788 Current Opinion in Psychology 49: 101514.

789 Orbell J and Dawes RM (1991) A “cognitive miser” theory of cooperators advantage.

790 American Political Science Review 85(2): 515-528.

791 Otto S and Pensini P. (2017). Nature-based environmental education of children:

792 Environmental knowledge and connectedness to nature, together, are related to

793 ecological behaviour. Global environmental change, 47, 88-94.

794 Pan Y, Opgenhaffen M and Van Gorp B (2021) China’s pathway to climate sustainability: A

795 diachronic framing analysis of people’s daily’s coverage of climate change (1995–

796 2018). Environmental Communication 15(2): 189–202.

797 Pan Y, Xie Y, Jia H and Luo X (2023) Ideologies, conspiracy beliefs, and the Chinese public’s

798 politicized attitudes to climate change. Sustainability 15(1): 131.

799 Pan Y, Xie Y, Jia H, Luo X and Zhang R (2022) Lower carbon, stronger nation: Exploring

800 sociopolitical determinants for the Chinese public’s climate attitudes. International
 36

801 Journal of Environmental Research and Public Health 20(1): 57.

802 Ployhart RE and Vandenberg RJ (2010) Longitudinal research: The theory, design, and

803 analysis of change. Journal of Management 36(1): 94-120.

804 Poortinga W, Spence A, Whitmarsh L, Capstick S and Pidgeon NF (2011) Uncertain climate:

805 An investigation into public scepticism about anthropogenic climate change. Global

806 Environmental Change 21(3): 1015-1024.

807 Priest SH (2001) Misplaced faith: Communication variables as predictors of encouragement

808 for biotechnology development. Science Communication 23(2): 97-110.

809 Rundle-Thiele S, David P, Willmott T, Pang B, Eagle L and Hay R (2019) Social marketing

810 theory development goals: an agenda to drive change. Journal of Marketing

811 Management 35(1-2): 160-181.

812 Sarafino EP (1996) Principles of behavior change: Understanding behavior modification

813 techniques. Wiley.

814 Schäfer S (2020) Illusion of knowledge through Facebook news? Effects of snack news in a

815 news feed on perceived knowledge, attitude strength, and willingness for discussions.

816 Computers in Human Behavior 103: 1-12.

817 Shi J, Li Z, Chen L and Tang H (2023) Individual and Collective Actions Against Climate

818 Change Among Chinese Adults: The Effects of Risk, Efficacy, and Consideration of

819 Future Consequences. Science Communication 45(2): 195-224.

820 Steg L (2023) Psychology of climate change. Annual Review of Psychology 74: 391-421.

821 Stoutenborough JW, Bromley‐Trujillo R and Vedlitz A (2014) Public support for climate

822 change policy: consistency in the influence of values and attitudes over time and

823 across specific policy alternatives. Review of Policy Research 31(6): 555-583.

824 Stryker S and Burke PJ (2000) The past, present, and future of an identity theory. Social

825 Psychology Quarterly 284-297.

 37

826 Sundel M and Sundel SS (2017) Behavior change in the human services: Behavioral and

827 cognitive principles and applications. SAGE publications.

828 Tarinc A, Ergun GS, Aytekin A, Keles A, Ozbek O, Keles H and Yayla O (2023) Effect of

829 Climate Change Belief and the New Environmental Paradigm (NEP) on Eco-Tourism

830 Attitudes of Tourists: Moderator Role of Green Self-Identity. International Journal of

831 Environmental Research and Public Health 20(6): 4967.

832 Taylor SE and Armor DA (1996) Positive illusions and coping with adversity. Journal of

833 Personality 64(4): 873-898.

834 Ter Huurne EF and Gutteling JM (2009) How to trust? The importance of self‐efficacy and

835 social trust in public responses to industrial risks. Journal of Risk Research 12(6):

836 809-824.

837 Topp K, Thai M and Hryciw DH (2019) The role of entertainment in engagement with

838 climate change. Environmental Education Research 25(5): 691-700.

839 Tøsse SE (2013) Aiming for social or political robustness? Media strategies among climate

840 scientists. Science Communication 35(1): 32-55.

841 Trémolière B and Djeriouat H (2021) Exploring the roles of analytic cognitive style, climate

842 science literacy, illusion of knowledge, and political orientation in climate change

843 skepticism. Journal of Environmental Psychology 74: 101561.

844 van der Linden S, Maibach E and Leiserowitz A (2015) Improving public engagement with

845 climate change: Five “best practice” insights from psychological science. Perspectives

846 on Psychological Science 10(6): 758-763.

847 Van der Werff E, Steg L and Keizer K (2013) The value of environmental self-identity: The

848 relationship between biospheric values, environmental self-identity and environmental

849 preferences, intentions and behaviour. Journal of Environmental Psychology 34: 55-

850 63.
 38

851 van Giesen RI, Fischer AR and van Trijp HC (2018) Changes in the influence of affect and

852 cognition over time on consumer attitude formation toward nanotechnology: A

853 longitudinal survey study. Public Understanding of Science 27(2): 168-184.

854 Wang Y, Li Y, Zhang J and Su X (2019) How impacting factors affect Chinese green

855 purchasing behavior based on Fuzzy Cognitive Maps. Journal of Cleaner Production

856 240: 118199.

857 Webb TL and Sheeran P (2006) Does changing behavioral intentions engender behavior

858 change? A meta-analysis of the experimental evidence. Psychological Bulletin 132(2):

859 249.

860 Whitmarsh L and O’Neill S (2010) Green identity, green living? The role of pro-

861 environmental self-identity in determining consistency across diverse pro-

862 environmental behaviours. Journal of Environmental Psychology 30(3): 305-314.

863 Xu T, Tian H, Zhu H and Cai J (2022) China actively promotes CO2 capture, utilization and

864 storage research to achieve carbon peak and carbon neutrality. Advances in Geo-

865 Energy Research 6(1): 1–3.

866 Yang X, Chen L and Ho SS (2020a) Does media exposure relate to the illusion of knowing in

867 the public understanding of climate change? Public Understanding of Science 29(1):

868 94-111.

869 Yang X, Wei L and Su Q (2020b) How is climate change knowledge distributed among the

870 population in Singapore? A demographic analysis of actual knowledge and illusory

871 knowledge. Sustainability 12(9): 3782.

872 Zanna MP, Olson JM and Fazio RH (1981) Self-perception and attitude-behavior consistency.

873 Personality and Social Psychology Bulletin 7(2): 252-256.

874