The relationship between a firm’s information transparency and consumers’ purchase intentions

(1)

1

The relationship between a firm’s information

transparency and consumers’ purchase intentions

Master Thesis

MSc in Business Administration - Marketing Track

Student

: Janneke Huisingh

Student number

: 10698671

Submission date : 27 January 2014

Version

: Final version

Thesis supervisor : Dr. M. Lee

(2)

2

Abstract

Companies are increasingly disclosing information about their business practices as a result of the growing consumer demand for information transparency. The purpose of this study is to get a deeper understanding of the factors that may influence the relationship between a firm’s transparency and consumers’ purchase intentions. An interpretive analysis, through an experiment in combination with a questionnaire, revealed that no direct relationship could be found between a firm’s transparency and consumer purchase intentions. However, a positive relationship was found between a firm’s transparency and consumers’ perceived transparency. Also a link was found between consumers’ perceived transparency and purchase intentions such that higher perceived transparency leads to higher purchase intentions. There can be concluded that a firm’s transparency itself does not have an impact on purchase intentions, it only impacts through consumers’ perceived transparency. In addition, the type of information was hypothesized to act as a moderator in the relationship between a firm’s transparency and consumers’ perceived transparency. However, findings showed that only information quality, and not information quantity, acted as a moderator such that high quality information has a more positive effect on consumers’ perceived transparency compared to low quality information. Furthermore, consumer trust toward disclosed information does not influence purchase intentions and no support was found for consumer trust acting as a moderator in the relationship between consumers’ perceived transparency and purchase intentions.

This thesis study adds to the existing literature on supply chain information transparency and of consumer behavior research. Both theoretical and managerial implications, as well as limitations, and directions for future research are discussed.

(3)

3

Statement of originality

This document is written by Student Janneke Huisingh who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(5)

5

1. Introduction

Nowadays, the global business environment is complex, turbulent, and ultracompetitive (Dyer & Ha-Brookshire, 2008). Consumer needs are constantly changing so businesses need to compete on multiple competitive bases, such as quality, costs, and customer services, in order to fulfil these fast-changing and diverse consumer needs. Today’s consumer demands also include transparent and sustainable products as people care more and more about their society and environment (Bhaduri & Ha-Brookshire, 2011). This thesis study will focus especially on the trend toward transparency.

From a consumer perspective, transparency can be defined as “the visibility and accessibility of information especially concerning business practices’’ (Merriam-Webster, cited in Bhaduri & Ha-Brookshire, 2011, p. 136). Both the increased awareness of the environment and the advanced technology in communication have highlighted the demand for transparency, especially in the global supply chain. Companies are now faced with demands from consumers for detailed information about production processes. A result of this demand for transparency is that companies need to communicate openly about their business practices. Strutnin (2008) suggested that maintaining a transparent supply chain is essential for a firm’s performance as it plays an important role in building customer loyalty, brand image, and brand reputation.

Several studies focused on the impact of information transparency on consumers’ purchase intentions. Carter & Curry (2010), for example, studied transparent pricing which refers to information revealing the allocation among agents in the supply chain. The authors used experiments in different product categories and their findings showed that transparent prices change consumer choice behavior. Furthermore, Lafferty & Goldsmith (1999) studied corporate transparency based on its community involvement and environment protection efforts. An experimental study was used to study the impact of transparency on purchase

(6)

6 intentions. Findings suggested that consumers were more likely to purchase from brands with a transparent supply chain than from brands that were not transparent. Overall, previous studies indicated that transparent practices have a significant impact on consumers’ purchase intentions. However, little is known so far about possible factors that may influence the relationship between a firm’s supply chain information transparency and consumers’ purchase intentions.

Many companies experience difficulties with being transparent. For example, Van der Cruijsen & Eijffinger (2010) argued that consumers often perceive less transparency compared to the actual transparency provided by the firm. Consumers’ transparency perceptions are driven by many individual as well as psychological characteristics. Also, consumers may have limited knowledge about the actual transparency efforts of the firm which influences how the firm’s transparency is perceived. Firms need to find a way to make the gap between their provided transparency and consumers’ perceived transparency as small as possible. However, a second difficulty firms struggle with is to find the best or most effective strategy for providing information (Salaün & Flores, 2001). Consumers often evaluate (transparent) information based on two criteria, namely the quality and quantity of information. So, the quality and quantity of provided information may be important factors in a firm’s marketing strategy, but which type of information should be used in order to best fulfil consumers’ needs? A third difficulty for firms is that consumers may distrust the information that is disclosed. Cohn & Wolfe (2013) found that consumers often think companies disclose information only for legal reasons, to make more money, or that they only disclose information that portrays the business in a positive way, leaving out negative details. Thus, consumer trust may influence consumers’ purchase intentions as well.

Overall, diverse factors may influence the relationship between a firm’s transparency and consumers’ purchase intentions. However, the three factors discussed above (consumers’

(7)

7 perceived transparency; the type of information; consumer trust toward disclosed information) have not been explored thoroughly. To address these gaps, this thesis study will focus on the following research question: How do the type of information, consumers’ perceived transparency, and consumer trust toward disclosed information influence the relationship between a firm’s transparency and consumers’ purchase intentions?

The research question above will be studied within the food industry. A number of crises in the agribusiness sector as well as the growing customer demand, resulted in the importance of transparency in food supply chains in recent years (Deimel et al., 2008). The purpose of this study is to build a conceptual model and emphasize the factors that are associated with consumers’ purchase intentions with regard to supply chain transparency within the food sector. Additionally, the study adds to the existing literature on supply chain information transparency and of consumer behavior research.

This paper starts with a review of existing literature on the thesis topic. Based on this literature review, hypotheses are formulated. Next, the methodology for testing the hypothesis will be described, followed by the results of the study. The paper will conclude with a discussion of the findings, theoretical and managerial implications, limitations and directions for future research, and the main conclusions will be summarized.

(8)

8

2. Literature review

2.1 The age of transparency

Consumer empowerment has led to a higher demand for information. Within the consumer-firm relationship, power shifted from the marketer to the consumer (Labrecque et al., 2013). An increase in access to information due to social media makes it possible for consumers to amplify their voices to anyone across the world. Furthermore, increased access to information has resulted in more educated consumers with higher information needs. As a result, firms are dealing with more powerful consumers than ever before. Fournier & Avery (2011) refer to the above described phenomenon as “the age of transparency”. “The age of transparency” leads to a demand toward companies to disclose information.

As a result, more and more companies make their business practices transparent and these transparent practices proved to enhance companies’ performances. Several studies argued that companies need to be transparent in order to perform better, to receive a competitive advantage, and to manage their reputations. For example, Bastian & Zentes (2013) conducted research about transparency in Western European agri-food supply chains. These authors discussed the antecedences and consequences of supply chain transparency (SCT) in sustainable agrarian supply chain management. SCT was defined as “the degree to which a supply chain player has access to relevant information about products, processes and flows of capital without loss, noise, delay and distortion” (Beulens et al., cited in Bastian & Zentes, 2013, p. 554). Partial least squares regression was used in an empirical sample with 131 supply chains with lead firms in Germany, Austria, and Switzerland.Findings revealed that transparent supply chains show a better performance in all performance dimensions (i.e. social performance, ecological performance, long-term relationship success, and operational performance). However, the challenge for companies is to disclose information without losing consumers due to possible disclosure of negative information.Nowadays, transparency should

(9)

9 be included in a firm’s competitive strategy in order to manage their performance, competitive advantage, and reputation.

2.2 The influence of transparency on purchase intentions

As mentioned in the introduction, transparency can be defined as “the visibility and accessibility of information especially concerning business practices’’ (Merriam-Webster, cited in Bhaduri & Ha-Brookshire, 2011, p. 136). Several studies focused on the positive impact of transparency on consumer purchase intentions. Consumer purchase intention can be defined as ‘‘the buyer’s self instruction to purchase the brand (or take another relevant purchase related action)’’ (Rossiter & Percy, cited in Bhaduri & Ha-Brookshire, 2011, p.138). Carter & Curry (2010) found that people value transparent pricing, which refers to information revealing the allocation among agents in the supply chain. These authors used experiments in diverse product categories with different sampling frames resulting in the finding that transparent prices systematically change consumer utility functions and stated choice behavior. People were even willing to pay more for the same product or service when the information was communicated via transparent pricing.

Furthermore, Lafferty and Goldsmith (1999) studied corporate credibility or transparency based on its community involvement and environment protection efforts. An experimental study examined this source of credibility to assess its impact on attitude-toward-the-ad, attitude-toward-the-brand, and purchase intentions. Findings suggested that where transparency is high, attitude-toward-the-ad, attitude-toward-the-brand, and purchase intentions will be higher than when it is low. Consumers were more favorable toward and more likely to purchase from brands with transparent business practices than those that were not transparent, which leads to the first hypothesis.

(10)

10 However, Van der Cruijsen & Eijffinger (2010) argued that (imperfect) consumer perceptions have an impact on consumers’ purchase intentions and behavior. Often, consumers perceive less transparency compared to the actual transparency provided by the company, which will be discussed in the next section.

2.3 Firm’s transparency versus consumers’ perceived transparency

In their research toward the European Central Bank, Van der Cruijsen & Eijffinger (2010) looked at the causes and consequences of consumers’ transparency perceptions because these perceptions influence consumers’ behavior. One of their findings showed that consumers often have limited knowledge about the actual transparency efforts of a firm. Most consumers rely on media information which can be obtained with little effort. It is possible that consumers do not know where else to look for the specific information, or they are just not motivated enough to start searching. Often, they are not even aware of the fact that the company also provides information on the website, which is normally more detailed than the limited information disclosed via the media. Furthermore, media information is often biased as media coverage is not random.

A second finding showed that perceptions of transparency are driven by many individual characteristics (e.g. optimism) as well as psychological characteristics. For example, consumers are more likely to grasp information that is in line with their prior beliefs or ideas about a firm’s transparency compared to information that is in contrast with their prior beliefs and ideas (Rabin, cited in Van der Cruijsen & Eijffinger, 2010, p.390). In addition, consumers often use their memory to form transparency perceptions, which biases the perceptions as people will mostly rely on the evidence they remember best (Tversky & Kahneman, cited in Van der Cruijsen & Eijffinger, 2010, p.390). Furthermore, consumers are overconfident about their beliefs and this overconfidence is pervasive (Malmendier & Tate, cited in Van der Cruijsen & Eijffinger, 2010, p.390). Another psychological characteristic of

(11)

11 consumers is that they may view the environment differently which can result in a different way of interpreting the same information (Babcock & Loewenstein, Van der Cruijsen & Eijffinger, 2010, p.390).

Based on the findings above, Van der Cruijsen & Eijffinger (2010) suggested that consumers may perceive less openness than companies actually provide. Thus, the perceived degree of transparency by consumers can be different from the actual degree of firms’ transparency (De Haan, cited in Van der Cruijsen & Eijffinger, 2010, p.389). Actual transparency refers to the actual level of openness by the firm (i.e. the information provided by the firm), while perceived transparency refers to consumers’ feeling or experience of being informed about relevant aspects (i.e. feeling or experience of the firm’s openness) (Deimel et al., 2008).

Additionally, Michaelson & Contracter (cited in Van der Cruijsen & Eijffinger, 2010, p.390) found that people cannot base their purchase intention or actual behavior on something that they do not know, nevertheless, consumers are likely to base their behavior on their own perceptions, even if these perceptions are different from what objective information dictates. Therefore, consumers’ transparency perceptions, rather than firms’ actual transparency, influence consumer purchase intentions. There is argued that the relationship between a firm’s transparency and purchase intention is influenced by consumers’ perceptions about the firm’s practises. To conclude, perceived transparency can be seen as the explanatory variable in the relationship between a firm’s transparency and consumers’ purchase intentions, which leads to a second hypothesis.

Hypothesis 2: The relationship between a firm’s transparency and consumers’ purchase

(12)

12

2.4 The moderating role of the type of information

Firms strive for higher consumer purchase intentions and the aim of this thesis study is to help marketers to improve marketing strategies in order to induce these purchase intentions. As discussed is paragraph 2.2, transparent information may influence purchase intentions. However, firms may struggle to find the best or most effective strategy for providing their information (Salaün & Flores, 2001). Consumers often evaluate (transparent) information based on two criteria, namely the quality and quantity of information. So, the quality and quantity of provided information may be important factors in a firm’s marketing strategy. Moreover, both quality and quantity of a firm’s transparency may be useful in understanding and explaining the relationship between a firm’s transparency, consumers’ perceived transparency, and purchase intentions. Within their study, Salaün & Flores (2001) tried to define the type of strategy that firms should best adopt in order to fulfill consumers’ information needs. They focused especially on consumers’ information needs in the food sector, which includes more than just examining the product and looking at the price, consumers want more information such as about the conditions in which it was manufactured. The higher the amount of information (i.e. quantity of information) firms provide, the more consumer information needs can be fulfilled, which results in higher perceived transparency by consumers. Often, firms disclose as much information as possible in order to fulfill consumer needs for transparency (Salaün & Flores, 2001). The diverse consumer information needs demand for information on many different subjects. However, a great deal of this disclosed information may be irrelevant for consumers. Information may be of no interest if it does not correspond to consumers’ needs or expectations. Moreover, consumers have to spend time to search within the mass of information to find the specific information they need. Thus, massive over-information leads to costs for consumers in terms of time spent looking for the necessary information, as well as boredom or impatience. This may result in

(13)

13 less confidence in and purchase intentions toward the product. But still, even though consumers may be confused by too much information, it is necessary for firms to provide a specific amount of information in order to fulfill consumers’ diverse information needs.

Quality of information leads to other possible types of information (i.e. high versus low quality information). Quality influences consumers’ perceived transparency and purchase intentions as well. Good quality information is becoming a necessary prerequisite in order to set up an active partnership between firm and consumers (Salaün & Flores, 2001). Good quality information is defined as “information which satisfies criteria of appreciation specified by the user, together with a certain standard of requirement (this may vary according to the use that is made of it)” (Salaün & Flores, 2001, p.26). Compared to high quantity information, high quality information is more efficient information and results in a better understanding and fulfillment of consumer needs, which results in higher perceived transparency. Based on the above arguments there can be hypothesized that the type of information moderates the relationship between a firms’ transparency and consumers’ perceived transparency, which in turn effects consumer purchase intentions.

Hypothesis 3: The relationship between a firm’s transparency and consumers’ perceived

transparency is moderated by the type of information provided.

Hypothesis 3a: High quantity information has a more positive effect on consumers’ perceived

transparency compared to low quantity information.

Hypothesis 3b: High quality information has a more positive effect on consumers’ perceived

transparency compared to low quality information.

However, quantity and quality of information are easily combined in the minds of consumers. Often, there is no clear distinction between the two. In addition to the findings of Salaün & Flores (2001), De Pelsmacker & Janssens (2007) also found that the role of quantity

(14)

14 of information is puzzling. These authors developed a model for fair trade buying behaviour using a sample of 615 Belgians. On the one hand, they found that high quantity information has a positive effect on product-specific attitudes and on buying behaviour. On the other hand, they found that the availability of high quantity of information leads to less concern and more scepticism. There is assumed that consumers react negatively to a large quantity of information. Furthermore, De Pelsmacker & Janssens (2007) argued that this negative reaction of consumers can be explained by the strong effect of perceived information quality on perceived information quantity. People often link high quality of information with high quantity because high quality makes them think there is enough information. In contrast, high quantity information makes people think that the company tries to ease consumers' conscience. To summarize, high quality information results in consumers’ perceptions of high quantity information which results in no distinction between the two factors in consumers’ minds. The above information leads to an additional hypothesis.

Hypotheses 3c: There is no distinction in the mind of consumers between quality and

quantity of information.

2.5 The influence of trust toward disclosed information on purchase intention

Even when consumers’ perceived transparency is close to a firm’s actual transparency, it does not always lead to more confidence as consumers may distrust the information provided by the firm. Cohn & Wolfe (2013) found that consumers often think companies disclose information only for legal reasons, to make more money, or that they only disclose information that portrays the business in a positive way, leaving out negative details. Focussing on effects on purchase intentions, Spenner & Freeman (2012) stated that “trust” is not about trust toward companies, it is about trusting the information gathered. In this thesis study, “trust” will refer to the extent to which consumers believe the information provided includes the complete truth (i.e. whether or not the information provided is honest and

(15)

15 complete). One of the reasons for consumer distrust lies with the marketers as they often simply recommend their brand based on product features and benefits (Spenner & Freeman, 2012). Nonetheless, consumers need information about an adviser’s decision criteria and brand usage as well. Moreover, providing information about the advisor may help in building consumer trust.

The focus in this study is that transparency contributes to the trust toward disclosed information which should not be confused with trust toward the company itself. Kirby (2012) mentioned that consumers will not really trust companies unless they are transparent. They suggest that being transparent results in increased consumer trust toward the firm itself. Furthermore, Kanagaretnam et al. (2010) found a relationship between transparency and trust as well while studying the impact of transparency and repeated interactions on the level of trust and trustworthiness (reciprocity) in an investment game setting. In this article, trust was defined as “the positive expectation that another will not - through words, actions, or decisions - act opportunistically” (Robbins & Langton, cited in Kanagaretnam et al., 2010, p. 242). These authors used “trust” to refer to consumer trust toward firms, rather than toward information. The findings present an increase in trusting behavior in one-shot interactions as well as repeated interactions in cases of transparency. In addition, transparency is important in creating trustworthiness (reciprocity) in one-shot interactions. With repeated interactions, trust and reciprocity will increase independent of the presence of transparency. To summarize, firms being transparent leads to increased trust toward the firm which in turn leads to an increase in trusting behavior by consumers.

Nevertheless, little is known so far about how the relationship between consumer trust toward information and transparency plays a role in consumers’ purchase intentions. In a situation of consumer distrust toward disclosed information, firms being transparent may create a different effect on purchase intentions. Arpan & Roskos-Ewoldsen (2005) tested the

(16)

16 proposition that rapid self-disclosure of crisis information would enhance perceptions of organizational credibility, that it would result in perceptions of the crisis as less severe, and that it would ultimately lead to greater intent to purchase the organization’s product. Previous studies suggested that disclosing negative information about oneself improves the credibility of the disclosed information (Williams et al., cited in Arpan & Roskos-Ewoldsen, 2005, p. 427). The findings show that enhanced credibility of the information predicts perceptions of the crisis as less severe. In turn, this results in greater intent to purchase the organization’s product. In conclusion, Arpan & Roskos-Ewoldsen suggest that whether or not consumers think the disclosed information is credible influences purchase intentions.

Hypothesis 4: Consumer trust toward disclosed information influences consumers’ purchase

intentions.

Hypothesis 4a: More consumer trust toward disclosed information positively affects

consumers’ purchase intentions.

Hypothesis 4b: Less consumer trust toward disclosed information negatively affects

consumers’ purchase intentions.

Furthermore, Bhaduri & Ha-Brookshire (2011) did a study to understand the factors that may influence consumer attitude and purchase intention. The study focused on apparel products from businesses that are transparent about their supply chain. After interviewing consumers, the authors found that attitude and purchase intention are affected by price, quality, prior knowledge about the industry, distrust on efforts of the business, and by hedonic values and social responsibility value gained by consumers of transparent products.

One important finding by Bhaduri & Ha-Brookshire includes the effect of trust or distrust on the relationship between consumers’ attitudes toward buying products from firms with transparent business practices and their purchase intentions for these products. In this

(17)

17 context, trust refers to consumer’s trust toward a firm’s disclosed information (i.e. disclosed information perceived as transparent by consumers) and Bhaduri & Ha-Brookshire suggest it acts as a moderator in the relationship between attitudes and purchase intentions. For example, if a consumer would have a positive attitude toward buying a product from a business with transparent practices a positive purchase intention would be expected, however, consumer distrust toward the businesses’ transparency claims might change the consumer’s purchase intention. Thus, Bhaduri & Ha-Brookshire suggest that despite businesses’ transparent supply chains, distrust in the businesses’ policies or claims might still keep consumers from having a positive purchase intention.

To summarize, perceived transparency influences purchase intentions, however, trust toward the disclosed information influences this relationship: more trust toward the disclosed information leads to a more positive influence on purchase intentions, less trust or even distrust toward this information leads to a more negative influence on purchase intentions. So the effect of a transparent supply chain on purchase intentions depends upon the level of trust toward the disclosed information. Moreover, trust toward disclosed information can act as a moderator. Based on these arguments, a fifth hypothesis exists:

Hypothesis 5: The relationship between perceived transparency and consumers’ purchase

intentions is moderated by consumer trust toward disclosed information.

Hypothesis 5a: More consumer trust toward disclosed information positively affects the

relationship between perceived transparency and purchase intention.

Hypothesis 5b: Less consumer trust toward disclosed information negatively affects the

(18)

18

2.6 Conceptual model

This thesis study tries to add knowledge to the literature regarding influences of transparency on consumers’ purchase intentions. From this literature review can be concluded that the type of information (quality versus quantity), consumers’ perceived transparency, and trust toward disclosed information, may all be related to transparency and purchase intentions. However, these three factors have not been explored thoroughly. To address these gaps, this thesis study will focus on the following research question: How do the type of information, consumers’ perceived transparency, and consumer trust toward disclosed information influence the relationship between a firm’s transparency and consumers’ purchase intentions?

The hypothesized relationships between the variables can be clarified in a conceptual model (figure 1). In this hypothesized conceptual model, ‘firm’s transparency’ is the independent variable and ‘consumers’ purchase intentions’ is the dependent variable. The relationship between these two variables is mediated by the variable of ‘consumers’ perceived transparency’ and moderated by ‘the type of information’ and ‘consumer trust toward disclosed information’.

Figure 1. Conceptual model hypothesized for consumer purchase intention regarding products from businesses

(19)

19

3. Methodology

3.1 Context study: supply chain transparency in the food sector

Companies can be transparent about different areas of information, for example, the sources used to produce a product, the ingredients used in a product, the price allocation, the supply chain, etcetera. In this thesis study, the focus is on transparency of information about the supply chain. Supply chain can be defined as ‘‘the key business processes from end-user through original suppliers, that provides products, services, and information that add value for customers and other stakeholders’’ (Lambert et al., cited in Bhaduri & Ha-Brookshire, 2011, p.136). Strutnin (2008) recommended firms to have a transparent and traceable supply chain in order to build customer loyalty, brand image and reputation, and to maintain business legitimacy. A traceable supply chain includes being transparent about the movement of source materials through processors, manufactures, and distribution channel members to the end user.

Furthermore, this thesis study focuses on the food industry. Food supply chains especially have been exposed to increasing requirements for transparency in recent years (Deimel et al., 2008). Consumers often take transparency into account while making their daily purchase decisions about food which makes transparency an important and challenging topic.

3.2 Research strategy and design

The purpose of this thesis study was to explore the relationship between a firm’s transparency and consumers’ purchase intentions as well as the role of the type of information, consumers’ perceived transparency, and consumer trust toward the disclosed information in this relationship. In order to get a broad understanding of the research topic this study was designed to be interpretative in nature. Central in interpretivism is the necessity to understand

(20)

20 differences between humans in their role as social actors in their natural environment (Saunders & Lewis, 2012). Furthermore, the study made use of a quantitative research strategy which is characterized by a deductive approach (Bryman, 2012). A deductive approach refers to “testing a theoretical proposition by using a research strategy specifically designed for the purpose of its testing” (Saunders & Lewis, 2012, p.108). Some key characteristics of deduction include the explanation of causal relationships between variables and the use of clearly structured methods to facilitate replication.

The hypotheses of this study were tested using an experiment in combination with a questionnaire. This strategy enables the collection of data about the same things from a large number of people in a cost-effective manner. Also, it is possible to generate findings that are representative for the whole population and there is a better control of the time schedule compared to more qualitative strategies. This was important given the limited time period of this study. The questions and statements of the questionnaire were clearly formulated in order to minimize misunderstanding by participants.

More specifically, a 2x2 factorial between-subject design was used (high versus low quality and high versus low quantity of disclosed information). A between-subject design includes different participants within each treatment group, which perfectly fits within this thesis study as there were five different groups. Moreover, this research is an experimental vignette study. “Vignette studies use short descriptions of situations or persons (vignettes) that are usually shown to respondents within surveys in order to elicit their judgments about these scenarios” (Atzmüller & Steiner, 2010, p.128). In this thesis study the aim was to use five different texts (vignettes). These texts provided information about the supply chain of a new product on the website from a specific company. Scenario 1 included information of high quality and low quantity; scenario 2 included information of low quality and high quantity; scenario 3 included information of both high quality and quantity; scenario 4 included

(21)

21 information of both low quality and quantity; scenario 5 included no information about the supply chain at all. Participantswere randomly assigned to one of the five scenarios and each participant was asked to fill in a questionnaire after they have read the information.

Furthermore, a non-existing company name was used in order to prevent harming an existing company in any way. The non-existing company, called “Chocolate Lovers”, is a Dutch brand producing 100% slave-free chocolate products that are available in the Netherlands. The chocolate industry was chosen because of the fact that transparency of business practises is a serious concern and a interesting trend within this industry (Gregory, 2013). 70% of all cacao beans comes from Western African countries, mostly Ghana and the Ivory Coast. These beans are sold to many chocolate companies, including the largest in the world. The concern of transparency stems from the fact that slavery today still exists in the global cocoa trade, many cacao farms in Western Africa make use of both slavery and child labor.

The firm’s transparency (independent variable) was expressed in the text on Chocolate Lovers’ website including information on the supply chain. The type of information (quality versus quantity) was manipulated which resulted in the different scenarios of the website text. The different texts used to support the questionnaire, included information about a new product by Chocolate Lovers, a chocolate bar with a new and unique taste. Participants were asked to imagine that Chocolate Lovers introduces this new chocolate bar and that the brand provided relevant information about this bar on their website. The information, as provided on the website, was shown on the page prior to the questionnaire. The type of information about the supply chain of this new product was manipulated. As mentioned before, five different scenarios of the text were used. The topic, style and design of the different scenarios were all the same as well as the first paragraph that introduced and described the new product. However, the scenarios differed in terms of quality and quantity of the disclosed supply chain

(22)

22 information (second paragraph). The English and Dutch versions of the scenarios and questionnaires can be found in Appendix A and B.

High quality information was reflected in a detailed description of four subjects of the supply chain, specific numbers (e.g. wage of farmers), and specific names (e.g. name of tablet producer). The four supply chain subjects included: the source of the cacao beans, the farmers (e.g. working conditions), the price paid to the farmers, and the transportation details. Quality was manipulated in terms of these supply chain details, numbers, and names. Thus, low quality information included less detailed information on the four supply chain subjects and included no specific numbers or names. Quantity of information was manipulated by the length of the supply chain information (i.e. by the number of words). Thus, high quantity information consisted of more words compared to low quantity information. Below, the different versions of the website texts are described more specific:

- Scenario 1, high quality and low quantity of supply chain information: High quality was reflected in a detailed description of the four subjects of the supply chain (e.g. “Our cacao beans come from Ghana (30%) and Ivory Coast (70%)”), by using specific numbers (e.g. “18% of the selling prices of the products is solely their wages”), and by using specific names (e.g. “Our beans go directly from Beansource (importer) to Choco Job (chocolate maker), to Prodoco (tablet producer)”). Low quantity was reflected in the use of only four lines of text.

- Scenario 2, low quality and high quantity of supply chain information: The lower quality was expressed by using less detailed information on the four subjects of the supply chain (e.g. “A few years ago we started buying our cocoa directly from cocoa farmers from good environments in these different countries.”), using no specific numbers (e.g. “Farmers receive an acceptable percentage wage of the price you pay for our products”), and by using no specific names (e.g. “the beans are carefully

(23)

23 transported to our chocolate maker”). The amount of text was three times bigger (12 lines instead of 4) than scenario 1 to express the higher quantity.

- Scenario 3, both high quality and high quantity of supply chain information: High quality was reflected in a detailed description of the four subjects of the supply chain (e.g. “30% of our cacao beans comes from Ghana and 70% from Ivory Coast”), by using specific numbers (e.g. “Farmers receive wage that accounts for 18% of the selling price of our products”), and by using specific names (e.g. “Then, the beans are transported to our chocolate maker, Choco Job”). In order to express the higher quantity the amount of text was three times bigger (12 lines instead of 4) than the scenarios with low quantity.

- Scenario 4, both low quality and low quantity of supply chain information: The lower quality was expressed by using less detailed information on the four subjects of the supply chain (e.g. “Our cacao beans come from farmers in different countries”), using no specific numbers (e.g. “The farmers receive an acceptable wage”), and by using no specific names (e.g. “Our beans are carefully transported to our importer, then to our chocolate maker, and finally to our tablet producer.”). Low quantity was reflected in the use of only four lines of text.

- Version 5 (control group) provided no information about the supply chain at all. Only the first paragraph (the same for every scenario), including general information about the new product, was provided.

3.3 Pre-test

Scenario 5 was the control group as this website text did not include information about the supply chain. A pre-test among 27 participants (a minimum of 5 participants per scenario) was done to check whether participants distinguished between information provided (scenario 1, 2, 3, and 4) and no information provided (scenario 5). An independent sample t-test for

(24)

24 perceived transparency showed a difference between the ‘no information provided’-group (M = 3.300, SD = .209) and the ‘information provided’-group (M = 3.580, SD = .584), however, this difference was not significant (t(25) = -1.041, p = .162). Thus, there can be concluded that participants did not noticed the difference between scenario 5 versus scenarios 1, 2, 3, and 4. In order to solve this problem, the website texts were adapted a bit.

3.4 Sample

The research population of this study includes the Dutch consumer. Participants needed to be consumers of food products who buy products themselves (so not, for example, children from which their parents do the groceries). The used sample consisted of 192 participants with a minimum of 35 participants for all five scenarios. However, there were 9 participants who judged the provided website information as not realistic and were deleted from the data set because of misinterpretation of the website text. A sample of 183 participants was left (31.7% male and 68.3% female) of which 76% is aged between 17 and 25 (95.1% is aged between 17 and 55). Participants were reached via Facebook and email.

3.5 Measures

The different variables that were measured in this study are explained in this section. To come up with relevant measures, a literature search was conducted. When available, existing measures were adapted from other studies, and if necessary, the wording was changed to better fit into this specific study. The complete questionnaire, including all determinants, can be found in Appendix A (English) and B (Dutch).

3.5.1 Consumers’ perceived transparency

‘Customers’ perceived transparency’ was the hypothesized mediator in this thesis study and refers to consumers’ feeling or experience of being informed about relevant aspects (Deimel et al., 2008). The five determinants used to measure perceived transparency were based on

(25)

25 this study by Deimel et al. (2008), and were examined on a five-point Likert scale ranged from strongly disagree (1) to strongly agree (5). An example determinant used in the questionnaire was: “The text includes relevant information for me as a consumer.”

3.5.2 Consumer trust toward disclosed information

The hypothesized moderator in the relationship between consumers’ perceived transparency and purchase intentions was ‘consumer trust toward disclosed information’. In this thesis study, ‘trust’ referred to the extent to which consumers believe the information provided includes the complete truth (i.e. whether or not the information provided is honest and complete). The determinants used to measure this variable were based on both this definition and on measures used by Kaplan & Nieschwietz (2003). A total of seven determinants was used and these were examined using a five-point Likert scale ranged from strongly disagree (1) to strongly agree (5). An example determinant used in the questionnaire is: “I believe that the information provided is true.”

3.5.3 Consumers’ purchase intentions

The dependent variable, consumers’ purchase intentions, was also examined on a five-point Likert scale ranged from strongly disagree (1) to strongly agree (5) that reflected consumers’ purchase intentions after reading the provided supply chain information. The five determinants used to measure consumer purchase intentions were based on the articles by Cook et al. (2002), Bai et al. (2008), and Bhaduri & Ha-Brookshire (2011). One of the statements used was: “I have a strong intention to purchase products from Chocolate Lovers.”

3.5.4 Quantity and quality check

Questions were added to be able to check participants’ perceptions of the manipulated type of information (quality and quantity). Two questions were used asking the participants to evaluate the quantity and quality of the provided website text on a five-point Likert scale

(26)

26 ranged from strongly disagree (1) to strongly agree (5). Participants evaluated the following two determinants: “I think the information provided is of good quality” and “The website text offers a lot of information about the supply chain.”

3.5.5 Reality check

Two questions were added to check whether participants perceived the company’s activities as realistic. Although a non-existing company is used in this study, it is very important that participants perceive the setting as realistic. An unrealistic setting may tend participants to give random answers as they do not believe the scenario described. The setting is made as realistic as possible by acting in a way that the company really exists, for example, by mentioning existing (real-life) details of the cacao trade and cacao supply chains. In order to check participants’ perceived reality of the story, existing measures from Ribbens & Malliet (2010) were adapted. Some changes in wording were done to fit the target context. The two final items included: “The scenario described in the survey is realistic” and “I can imagine the scenario described in the survey is a real life scenario”. The two statements were measured on a five-point Likert scale ranged from strongly disagree (1) to strongly agree (5).

3.5.6 Control variables

A few questions concerning demographics were asked at the end of the questionnaire in order to determine whether or not these demographics influenced the participants’ answers. The demographics included were: gender, age, and education. The age categories used were 0-16, 17-25, 26-35, 36-45, 46-55, and 55+. With selecting their highest level of education completed, participants could choose out of the following options: Elementary school, High school, Intermediate vocational education, Bachelor, Master, or “other, namely…”

Issue involvement was included as control variable as well. This variable was measured through asking to what extent the participants care about supply chain transparency, through what extent they think information about a product’s supply chain is important, and

(27)

27 whether or not a product’s supply chain usually influences their purchase decisions. These measures were based on measures used by Gill et al. (1988). The three statements were measured on a five-point Likert scale ranged from strongly disagree (1) to strongly agree (5).

3.5.7 Reliability check

In order to be sure that the statements used in the questionnaire reflected the constructs, it was useful to check the reliability of the scales. Cronbach’s alpha is the most common measure of scale reliability and a value of .7 is acceptable (Field, 2009). Reliability of the scales was tested in a pre-test among 27 participants (a minimum of five participants per scenario). Cronbach’s alpha for consumer trust was found to be .872 and for issue involvement .811 which indicated reliable scales. Cronbach’s alpha for perceived transparency was found to be .655 and for purchase intentions .634. In order to increase reliability of these two scales, a few new statements were added to the questionnaire.

3.6 Analytical methods

In order to prepare the conducted data for further analysis, names and labels were created in SPSS for all different measures of the variables. Next, different analytical methods were used to test the hypotheses. To test hypothesis 1, a linear regression analysis was conducted to investigate the ability of a firm’s transparency to predict consumers’ purchase intentions, while controlling for age, gender, education, and issue involvement. The mediating effect of hypothesis 2 was tested by using a Sobel test for mediation. In order to conduct this Sobel test, two linear regression analyses were done to compute the raw regression coefficient and its standard error for the association between a firm’s transparency and consumers’ perceived transparency, and for the association between consumers’ perceived transparency and consumers’ purchase intentions. Control variables were included in this analysis. The moderating effect of hypothesis 3 was tested using an one-way ANOVA analysis to check whether there is a difference in perceived transparency by consumers between the five groups

(28)

28 exposed to one of the five different information types (scenarios). In order to find out how the five groups differ from each other, it was necessary to carry out further analysis using planned comparisons with the contrasts option in SPSS. Next, hypothesis 3c suggested that there is no distinction in the mind of consumers between quality and quantity of information. This hypothesis was tested by conducting a univariate ANOVA in order to find the interaction effect between quality and quantity. Furthermore, hypothesis 4 was tested by conducting a (2-step) linear regression analysis to investigate the ability of consumer trust to predict consumers’ purchase intentions, after controlling for perceived transparency, age, gender, education, and issue involvement. Finally, the moderating effect of hypothesis 5 was tested using a regression analysis. As the hypothesis wanted to test a moderating relationship, the moderator (consumer trust toward disclosed information) was entered into the regression analysis with the program Process by Andres F. Hayes, using conceptual model 1. Age, gender, education, and issue involvement were included as control variables (covariates). The exact steps of all analytical methods will be discussed in the next chapter.

(29)

29

4. Results

This section includes the results of the study. The data preparation will be discussed, followed by the reality check, manipulation check, and correlations. Then, the hypotheses will be tested.

4.1 Data preparation

Results were conducted from 192 participants. In order to prepare the data for further analysis, names and labels were created in SPSS for all different measures of the variables, including: the type of information (TypeInfo: 5 scenarios), information (Information: provided(1) versus not provided(0)), degree of quality (Quality: high(1) versus low (0)), degree of quantity (Quantity: high(1) versus low (0)), consumers’ perceived transparency (PercTrans), consumer trust toward disclosed information (ConsTrust), consumers’ purchase intentions (PI), issue involvement (IssueInvol), perceived quality of the information (PercQual), perceived quantity of the information (PercQuant), age, gender, and education. For example, purchase intentions (PI) was measured using five measures (statements) in the questionnaire, so five variables were created in SPSS with the names PI1, PI2, PI3, PI4, and PI5. Furthermore all of these variables were coded whereby counter‐ indicative items were recoded (only ConsTrust6 was recoded).

Furthermore, all respondents answered all questions so there were no missing data. If there would have been missing data, these could have been handled by just leaving the SPSS cells blank.

4.2 Reality check and manipulation check

A reality check was done to check whether or not participants viewed the given website-information as realistic. First, the scale mean (RealityTOT) was computed out of the two measures for reality (Reality1 and Reality2). By conducting an one sample t-test, the mean of

(30)

30 the reality scale mean was found to be 3.66 (M = 3.66, SE = .048, SD = .660). However, there were 9 participants who judged the provided website information as not realistic. These participants answered one of the two or both questions about reality with 1 (strongly disagree) or 2 (disagree), and were deleted from the data set because of misinterpretation of the website text. Still more than 95% of all participants (183 participants) was left and the reality check was done again. A test value of 3 was used in the one sample t-test because a score of 3 or higher indicates that participants perceive reality as high. Descriptive statistics of this second one sample t-test showed that all participants perceived the provided scenarios as realistic and they can imagine that the scenario happens in real life (M = 3.75, SE = .041, SD = .559). This mean was found to be significantly larger than the test value 3 (mean difference = .75, p = .000).

Scenario 5 was the control group as this website text did not include information about the supply chain. A manipulation check was done to check whether participants distinguished between information provided (scenario 1, 2, 3, and 4) and no information provided (scenario 5). An independent sample t-test for perceived transparency showed a significant difference between the ‘no information provided’-group (M = 3.19, SD = .638) and the ‘information provided’-group (M = 3.68, SD = .606) (t(181) = -4.306, p = .000). There can be concluded that participants distinguish between information provided (scenario 1, 2, 3, and 4) and no information provided (scenario 5).

Furthermore, the type (quality and quantity) of information was manipulated in scenarios 1, 2 ,3 , and 4. A manipulation check was done to check whether participants distinguished between high and low quality and whether they distinguished between high and low quantity. An independent sample t-test for perceived quality showed a significant difference between the low quality group (M = 3.21, SD = .848) and the high quality group (M = 3.61, SD = .836) (t(146) = -2.871, p = .005). There can be concluded that the

(31)

31 participants distinguished between high and low quality of the provided information. An independent sample t-test was done for perceived quantity as well. Looking at the 10% significance level, this test showed a significant difference between the low quantity group (M = 3.16, SD = .895) and high quantity group (M = 3.40, SD = .843) (t(146) = -1.682, p = .095). However, the difference was not significant at the 1% and 5% significance level, so the manipulation of information quantity seems not to be very good. Other reasons for the low significant difference might be that participants do not have another quantity of text to compare it with, or that quantity goes together with quality in the mind of consumers (will be tested later on in hypothesis 3c). Overall, the significant difference at the 10% level, is enough reason to include quantity of information in further analyses.

4.3 Reliability

In order to be sure that the statements used in the questionnaire reflect the constructs, the reliability of the scale was checked. Cronbach’s alpha is the most common measure of scale reliability and a value of .7 would be acceptable (Field, 2009). Cronbach’s alphas for perceived transparency (.801), consumer trust (.852), purchase intentions (.796), and issue involvement (.811) were all above .7, which indicated a reliable scale.

4.4 Correlations

The results of the correlation matrix are shown in table 1 and provide the means, standard deviations, and the correlation coefficients.

The first noteworthy significant correlation found was between perceived transparency and consumer trust (r = .472, p < .01), which indicates that a higher perceived transparency goes together with higher consumer trust. Perceived transparency has a significant positive correlation with purchase intentions as well (r = .429, p < .01). This is in line with hypothesis 2 that suggests that the relationship between a firm’s transparency and consumers’ purchase intentions is mediated by consumers’ perceived transparency. In addition, consumer trust is

(32)

32 significantly correlated with purchase intentions (r = .314, p < .01) which is in line with hypothesis 4 that hypothesizes that consumer trust influences purchase intentions.

Table 1 further showed that the control variables involved some significant correlations as well. Perceived transparency is negatively correlated with education (r = -152, p < .05). Purchase intentions is positively correlated with issue involvement (r = .381, p < .01), gender (r = .171, p < .05), and age (r = .147, p < .05). Therefore, these control variables will be taken into account when conducting regression analysis.

Variable M SD 1. 2. 3. 4. 5. 6. 7. 1. Perceived transparency 3.59 .641 (.801) 2. Consumer trust 3.22 .594 .472** (.852) 3. Purchase intentions 3.33 .677 .429** .314** (.796) 4. Issue involvement 3.70 .680 .045 .040 .381** (.811) 5. Gender 1.68 .467 .143 .114 .171* .058 - 6. Age 2.57 1.17 .114 .139 .147* -.032 -.088 - 7. Education 3.97 .794 -.152* -.132 -.099 .042 -.127 -.201** -

Table 1: Correlation matrix.

M = Mean, SD = Standard deviation.

**Correlation is significant at the .01 level (2-tailed). *Correlation is significant at the .05 level (2-tailed).

Note: scores on the diagonal represent the Cronbach’s alphas.

4.5 Hypotheses testing

4.5.1 Dependent variable: consumers’ purchase intentions

Hypothesis 1 (H1) suggested that a firm’s transparency has a positive effect on consumers’ purchase intentions. A linear regression analysis was conducted to investigate the ability of a firm’s transparency to predict consumers’ purchase intentions, while controlling for age, gender, education, and issue involvement.

Five predictors were entered in the regression analysis, namely: firm’s transparency (information provided or not), and the control variables: age, gender, education, and issue involvement. Table 2 presents the results for this analysis. The model was found to be

(33)

33 statistically significant (F(177) = 9.263, p = .000) and explained 20.7% (R2 = .207) of variance in purchase intentions. However, the prediction by firms’ transparency was not significant (p = .230). This indicated that a firm’s transparency did not have a significant effect on purchase intentions. Thus, hypothesis 1 is not supported. Three out of the five predictors were statistically significant, with issue involvement recording a higher Beta value (β = .390, p < .01) than gender (β = .157, p < .05) and age (β = .148, p < .05). Table 2 summarizes the results of the regression analysis.

R R2 F Sig B SE β t Sig Regression .455a .207 9.263 .000a Age .086 .040 .148 2.137 .034 Gender .228 .099 .157 2.308 .022 Education -.060 .059 -.071 -1.017 .311 Issue involvement .388 .067 .390 5.763 .000 Firm’s transparency .141 .117 .082 1.203 .230

Table 2: Regression analysis for hypothesis 1.

a. Predictors: age, gender, education, issue involvement, firm’s transparency. Dependent variable: purchase intentions.

4.5.2 Mediating effect: consumers’ perceived transparency

Hypothesis 2 (H2) suggested that the relationship between a firm’s transparency and consumers’ purchase intentions is mediated by consumers’ perceived transparency. To test this hypothesis, a Sobel test for mediation was done. In order to conduct this Sobel test, two linear regression analyses were done to compute the raw regression coefficient and its standard error for the association between a firm’s transparency and consumers’ perceived transparency, and for the association between consumers’ perceived transparency and consumers’ purchase intentions.

In the first regression analysis, perceived transparency was entered in SPSS as the dependent variable and a firm’s transparency as the independent variable together with the control variables ages, gender, education, and issue involvement (B = 2.905, SE = .418). In the second regression analysis, purchase intentions was entered as the dependent variable and

(34)

34 both perceived transparency and a firm’s transparency as independent variables together with the control variables age, gender, education, and issue involvement (B = .419, SE = .070). The results for these two analyses are presented in table 3. Based on these results, the Sobel test statistic was found to be 4.535 with an associated p-value of .000. The fact that the observed p-value does fall below .05 would indicate that the association between a firm’s transparency and consumers’ purchase intentions is influenced significantly by the inclusion of consumers’ perceived transparency. However, hypothesis 1 was not supported (see section 4.5.1) which means that there is no relationship at all between a firm’s transparency and purchase intentions. Combining the results from H1 and H2, a link is found between a firm’s transparency and perceived transparency, and then, from perceived transparency to purchase intentions. Overall, there can be concluded that a firm’s transparency only indirectly influences purchase intention. A firm’s transparency itself does not have an impact on purchase intentions, it only impacts through the perceived transparency.

B SE β t Sig. Analysis 1a (Constant) _2,905 _,418 _6,955 _,000 Firm’s transparency _,527 _,115 _,325 _4,594 _,000 Age ,031 ,039 ,058 ,802 ,424 Gender ,199 ,097 ,145 2,058 ,041 Education -,115 ,058 -,143 -1,988 ,048 Issue involvement ,081 ,066 ,086 1,225 ,222 Analysis 2b (Constant) _,202 _,440 _,460 _,646 Firm’s transparency _-,080 _,113 _-,047 _-,706 _,481 Perceived transparency ,419 ,070 ,397 5,973 ,000 Age ,072 ,037 ,126 1,974 ,050 Gender ,144 ,091 ,100 1,581 ,116 Education -,012 ,055 -,014 -,217 ,828 Issue involvement _,354 _,062 _,356 _5,728 _,000

Table 3: Regression analyses for hypothesis 2.

a. Dependent variable: perceived transparency. b. Dependent variable: purchase intentions.

(35)

35 Additionally, in the final analysis three out of the six predictors were found to be statistically significant, with perceived transparency recording a higher Beta value (β = .397, p = .000) than issue involvement (β = .356, p = .000) and age (β = .126, p = .050).

4.5.3 Moderating effect: type of information

Hypothesis 3 (H3) suggested that the relationship between a firm’s transparency and consumers’ perceived transparency is moderated by the type of information provided such that high quantity information has a more positive effect on consumers’ perceived transparency compared to low quantity information (H3a), and such that high quality information has a more positive effect on consumers’ perceived transparency compared to low quality information (H3b).

An one-way ANOVA analysis was done to check whether there is a difference in perceived transparency by consumers between the five groups exposed to one of the five different information types (scenarios). Perceived transparency was used as the dependent variable and the type of information as the independent factor. The analysis showed a statistically significant F-ratio (F(4, 178) = 6.393, p = .000), which indicated that the difference in perceived transparency between the five groups was statistically significant.

In order to find out how the five groups differ from each other, it was necessary to carry out further analysis using planned comparisons with the contrasts option in SPSS. Three contrasts were used. The first contrast compared group 5 with the other four groups (no information provided versus information provided) and revealed that participants’ perceived transparency for group 5 was significantly lower compared to group 1, 2, 3, and 4 (p = .000). Thus, indeed, participants perceived more transparency when supply chain information was provided. Next, the second contrast compared group 1 and 3 with group 2 and 4 (high quality versus low quality information). A significant difference was found between group 1 and 3 compared to group 2 and 4 (p = .019), showing a higher perceived transparency when the

(36)

36 quality of information was high, supporting H3b. The third contrast compared group 2 and 3 with group 1 and 4 (high quantity versus low quantity information). No significant difference was found between these groups (p = .355) meaning that participants exposed to high quantity information did not perceive the transparency different from participants exposed to low quantity information. Thus, H3a is not supported. The means of consumers’ perceived transparency per group are provided in table 4. Table 5shows the results for the contrast test.

Group N Mean PercTrans Standard deviation Standard error

1. High quality, low quantity 35 3,86 ,511 ,086

2. Low quality, high quantity 42 3,53 ,626 ,097

3. High quality, high quantity 36 3,76 ,701 ,117

4. Low quality, low quantity 35 3,61 ,525 ,089

5. No supply chain information provided 35 3,19 ,638 ,108

Table 4: Means of consumers’ perceived transparency per group.

Contrast Value of contrast Standard error t df Sig. (2-tailed) 1* 6,03 1,367 4,409 178 ,000 2* 1,42 ,599 2,371 178 ,019 3* -,56 ,599 -,928 178 ,355

Table 5: Contrast test with consumers’ perceived transparency (equal variances assumed).

* Contrast 1: Group 5 compared to group 1, 2, 3, and 4. Contrast 2: Group 1 and 3 compared to group 2 and 4. Contrast 3: Group 2 and 3 compared to group 1 and 4.

Overall, there can be concluded that H3 is partly supported. H3b can be supported and H3a cannot, which means that the quality of information does play a moderating role in the relationship between a firm’s transparency and consumers’ perceived transparency, and quantity of information does not. However, H3c suggested that there is no distinction in the mind of consumers between quality and quantity of information. In the questionnaire, one question was asked about how participants perceived the quality of the information (perceived quality), and one question was asked about how participants perceived the quantity of

The relationship between a firm’s information transparency and consumers’ purchase intentions