“When” Does It Pay to Be Good? Attributions Mediate the Way CSR Elements Impact on Consumer Responses, and Are Controllable

Abstract

Responding to the appeal for more research on the contingencies that shape the relationship between CSR and corporate performance, this paper incorporates environmental CSR, sets up an experimental survey and employs multiple mediation analysis with the aim to test the mediating role of consumer attributions on the CSR elements–consumer responses relationship; and further to examine the degree to which attributions are controllable, i.e., specific CSR elements activate specific type of attributions. Results support that attributions have a strong predicting power on consumer outcomes. The right time of appearance and the appropriate amount of resources committed to a CSR campaign, through the dual type of attributions they activate (more positive, i.e., values-driven and less negative, i.e., egoistic), impact positively on consumer reactions. In this respect, the study adds to past research showing that attributions are controllable, i.e., specific CSR initiative characteristics of a impact on the dimensionality of attributions and, through that, on specific target-types of consumer responses. This study thus shows that the activation of a dual-level attributions’ system is ambivalent, dependent on the character of the CSR campaign. The fact that specific CSR elements (i.e., CSR Timing) activate dual-level CSR motives that act complementarily indicates that managers should be clear about the capabilities of the elements of their CSR initiatives and how much impact they expect those elements to have on consumer response.

1. Introduction

Past review studies on the topic of Corporate Social Responsibility (CSR) (e.g., [1,2,3,4]) point towards the existence of a positive relationship between CSR and corporate performance [5,6], since stakeholders typically reward ‘good’ companies and punish ‘bad’ ones [7]. In this respect, positive consumer response to responsible companies as a corporate performance indicator has gained increased attention early on. Most studies in the area systematically support that a firm’s CSR behavior influences a range of customer responses, such as company reputation, corporate/product attitude, purchase intention, willingness to pay, consumer-company relationship, perceived purchase risk, etc. [8,9,10,11,12,13,14,15,16,17]. Results of the 2018 CSR RepTrak^® 100 study show that as the CSR index score of a company moves from ‘strong’ to ‘excellent’, so does the percentage of the 230,000 consumers surveyed globally who claim willing to purchase this company’s products, i.e., from 63.5 percent to 83.9 percent respectively [18].

Recently, a growing body of literature suggests shifting research attention from whether it pays to be good (i.e., from verifying the relationship between CSR and any corporate performance metric) to when (or why) it pays to be good (i.e., to the contingencies or ‘contextual ground’ that affect the aforementioned relationship) [19,20,21,22,23]. One of the most influential contingencies identified in extant literature is consumer attributions, namely the motives that consumers attribute to a company’s CSR behavior. The basic idea of attributions theory is that people tend to act as ‘naïve psychologists’ [24], who seek to understand the behaviors of others by making inferences about others’ motives and intentions. Attributions are complex and with strong power to predict consumer outcomes.

In the CSR context in particular, the pioneering work by Ellen et al. (2006) [25] indicates that consumers recognize a multitude of motives that may drive the firm’s decision to engage in CSR. The same authors classify those motives into company-internal and company-external, and differentiate each type further into components of different valence. In particular, company-internal motives (‘business-serving’) can be egoistic (‘instrumental’, e.g., reduce costs) or strategic (e.g., retain customer basis); while company-external motives (‘altruistic or public-serving’) can be values-driven (‘moral’, e.g., the firm believes it is the right thing to do) or stakeholder-driven (e.g., satisfy shareholders’ expectations). Internal and external motives are seen as the two extremes on a ‘motivational continuum’ of mixed motives [24,26]. Resulting from a complex inferential procedure, consumers locate along this continuum a point of equilibrium between ‘doing well’ and ‘doing good’ behind companies’ CSR engagement [27], with relevant implications on consumer reactions. Ellen et al. (2006) [25] claim that consumers react positively to firms whose motives are perceived as values-driven and strategic.

Furthermore, Ellen et al. (2006) [25] establish that attributions are controllable. For instance, specific characteristics of the company running a CSR campaign (e.g., reputation, expertise and perceived corporate involvement in the cause [24]; or corporate honesty [28,29], when perceived by consumers favorably, lead to positive reactions on their behalf. Besides company-related perceptions, CSR campaign-related perceptions also play a role in the direction attributions may grow. In this respect, understanding the true nature of CSR becomes directly contingent upon the CSR-relevant information available. If information about a CSR campaign is not abundant or clear, CSR attributions are not directly inferred, consumers may thus look for additional information to evaluate company’s motives in a more explicit manner. Given the information asymmetry that exists between companies and the public [27], stakeholders are likely to seek information on specific characteristics of the CSR campaign to interpret its true nature, and CSR campaign-related characteristics typically constitute a very salient set of information cues that consumers can assess directly.

The present paper responds to the quest for more research effort to be directed towards the investigation of the degree to which specific CSR characteristics can change the attributions consumer make about CSR and thus trigger ‘controllable’ consumer reactions [30]. Accordingly, this paper aims to test the mediating role of consumer attributions on the relationship between specific CSR characteristics (hereafter CSR ‘elements’) and consumer responses and examine the degree to which the activation of specific consumer attributions is controllable. To fulfil this aim, our paper draws on Attributions theory using consumer attributions as a contingency type; attitude and purchase intention as consumer responses; and specific CSR elements as determinants of consumer responses. More specifically the paper involves an experimental study in the context of environmental sustainability that investigates: (a) if specific CSR elements (i.e., Timing: proactive vs. reactive launch of the CSR campaign; and Intensity: high vs. low time and financial commitment to CSR) have an effect on consumer attitude towards the company and purchase intention towards the company’s product; (b) if consumer attributions (i.e., egoistic, strategic, value-driven and stakeholder-driven) about a company’s CSR involvement mediate the way CSR Timing and Intensity impact on customers’ attitude and purchase intention; and (c) if CSR Timing and Intensity activate attributions that act complementarily in a direction positive for the firm (e.g., more value-driven/less egoistic).

2. Conceptual Framework and Research Hypotheses

Figure 1 shows the conceptual model of the study. The events in the model follow two paths that imply a mediation effect: (a) Timing and (resource) Intensity have a direct effect on consumer attitude towards the company that initiates CSR (path 1) and consumer intention to purchase the company products (path 2); and (b) Timing and Intensity have an indirect effect on consumer attitudes and purchase intention, through consumer attributions (paths 3). The following paragraphs provide a theoretical justification for the model relationships.

Specific CSR campaign elements have an impact on consumer responses, among which CSR Timing and Intensity have received attention early on in the literature (e.g., [31]). A firm can choose to deploy a CSR initiative either voluntarily (i.e., proactively), prior to the reporting of any negative information [32]; or as a response to some negative event (i.e., reactively or ‘post-crisis’) [21,33], in an attempt to mitigate the negative impact on the corporate brand or legitimize itself in the eyes of consumers. Moreover, a firm can choose the amount of effort to put into a CSR program. For instance, the time-length of deployment of an advertising campaign with a social message associates with higher perceived success and is an indication of higher effort [34] and greater sacrifice on behalf of the firm [35].

2.1. Relationship between CSR Timing/CSR Intensity and Consumer Reactions

Typically, consumers react positively to proactive and high-intensity CSR. Expectancy Value theory [36] supports that in these cases consumers attach new, positive associations to the company. If these new associations exceed consumers’ own expectation, an improvement of attitude will come because of the proactive/intensive CSR [9,37]. On the other hand, Prospect theory [38] suggests that subjects pay closer attention to negative information preceding a reactive CSR campaign, because this type of information is scarcer. Overweighing of negative information has a strongly negative impact on consumer attitudes. Consumers become skeptical of the attempt to legitimize the company and do not “reward” it with strong purchase intention. In that case, only proactive, investment-heavy strategies can prove effective in generating positive reactions from consumers, such as purchase and recommendation intentions [39], or repeat patronage [40]. Therefore:

Hypothesis 1a (H1a).

Proactive (reactive) CSR Timing has a positive (negative) effect on consumers’ attitudes towards the company;

Hypothesis 1b (H1b).

Proactive (reactive) CSR Timing has a positive (negative) effect on consumers’ purchase intention;

Hypothesis 1c (H1c).

High (low) CSR Intensity has a positive (negative) effect on consumers’ attitudes towards the company; and

Hypothesis 1d (H1d).

High (low) CSR Intensity has a positive (negative) effect on consumers’ purchase intention.

2.2. Relationship between CSR Timing/CSR Intensity and Consumer Attributions

Attribution theory is an umbrella term for multiple theories that investigate the ‘ways in which individuals arrive at causal inferences, what sort of inferences they make, and what the consequences of these inferences are’ [41]. Attribution theory addresses the processes by which individuals evaluate the motives of others and explains how these perceived motives influence subsequent attitudes and behavior [42]. The main idea of attribution theory is that individuals interpret behaviors and events based on their perceived cause, in lack of trustworthy or clear information [43]. When trust is limited due to lack of information, and there is suspicion that external conditions (i.e., hidden motives) are the cause of the behavior, the observer will most likely conclude that the actor is not predisposed to that behavior [44]. In other words, consumers may care less about what firms are doing than about when and why they are doing it [44,45].

One premise of Attribution theory is that the context influences the way consumers assign motives to the events they witness [46]. In a CSR context, researchers draw on Signaling theory [47,48], which argues that organizations (‘signalers’) use CSR (‘signals’) to communicate positive qualities to stakeholders (‘receivers’), such as company’s genuine concern about social welfare. In turn, stakeholders use signal information to make important decisions, i.e., product purchase-related [27,49]. Since CSR campaigns include positive aspects for both the company and the society, specific elements of CSR, for instance CSR timing and investment effort, when communicated to consumers, can impact positively on the attributions stakeholders make, since a mismatch may otherwise occur between a company’s intent to send a positive signal (i.e., doing good to the society) and the receiver’s ultimate interpretation of it [27].

In the proactive CSR Timing condition, the ‘signal’ is positive, making consumers draw positive conclusions about the firm’s intentions, leading to attributions that are more values-driven (i.e., CSR is the right thing to do) or strategic (i.e., CSR helps the company get more customers) [39]. In the reactive CSR condition, the ’signal’ is negative due to the prior irresponsible event, which results in negative conclusions about the firm’s intentions [33]. Responding to a negative event with a CSR initiative poses a character-driven problem that generates skepticism [50], leading to attributions that are more egoistic (i.e., SCR helps the company to get positive publicity) or stakeholder-driven (i.e., CSR responds to expectations by stockholders) [51].

In addition, the intensity of the effort in terms of (financial and time) resources that a company commits to the CSR target can be one major factor determining whether the company has a true interest in giving back to society, or the company is just exploiting the problem for its own benefit. Investing more resources in a CSR target signals a higher level of involvement in the cause. Higher investment also indicates that the company is generous and caring, while consumers most likely infer that the company’s motives are pure [35], leading to more values-driven or strategic attributions. On the other hand, lower investment may signal that the company is just using the cause in order to gain some benefits (e.g., legitimacy, brand image, etc.), doing the acceptable minimum to support the cause. In this case, consumers will most likely infer that the company behaves selfishly, leading to more stakeholder-driven and egoistic attributions. Therefore:

Hypothesis 2a (H2a).

Proactive (reactive) CSR Timing increases (decreases) (a) values-driven and (b) strategic attributions, and decreases (increases) (c) egoistic and (d) stakeholder-driven attributions; and

Hypothesis 2b (H2b).

High (low) CSR Intensity increases (decreases) (a) values-driven and (b) strategic attributions, and decreases (increases) (c) egoistic and (d) stakeholder-driven attributions.

2.3. Relationship between Consumer Attributions and Consumer Responses

Kelley (1973) [43] suggests that individuals minimize the effect of an attribution when an alternative attribution seems more convincing in driving behavior (the Discounting principle). Since CSR campaigns involve actions that benefit both the company and the society, the attribution of motivations is not always clear. Consumers will then apply a more extensive elaboration that may generate multiple, often rival hypotheses (i.e., both business- and public-serving) about why an activity (the CSR campaign) occurs [20,26]. Fein (1996) [45] shows that suspicions of ulterior motives drive consumers to hold plausible rival hypotheses about a company’s behavior. Most likely, consumers will interpret its actions as self-interested, because of the firm’s need to protect its legitimacy in the face of accusations. In this case, observers may conclude that the firm does not really care for the issue at hand, but it rather tries to portray itself as an ethical company, leading to more egoistic attributions. Jayachandran et al. (2013) [52] support that consumers punish firms that engage in actions perceived as opportunistic or self-serving; while Ellen et al. (2006) [25] claim that consumers react positively to firms whose motives are perceived as values-driven and strategic. The above provide the theoretical support for the link between consumer attributions and reactions. As Figure 1 shows, this link completes the two paths that postulate the mediating role of attributions. Accordingly (see Figure 1).

Hypothesis 3a (H3a).

Consumer attributions mediate the relationship between CSR Timing and consumers’ attitudes towards the company;

Hypothesis 3b (H3b).

Consumer attributions mediate the relationship between CSR Timing and consumers’ purchase intention;

Hypothesis 3c (H3c).

Consumer attributions mediate the relationship between CSR Intensity and consumers’ attitudes towards the company; and

Hypothesis 3d (H3d).

Consumer attributions mediate the relationship between CSR Intensity and consumers’ purchase intention.

3. Materials and Methods

3.1. Development of Experimental Stimuli

The CSR scenario of the current study involves environmental sustainability behavior of a hypothetical food company. The food industry is especially prone to sustainability problems (e.g., reduction of biodiversity, water usage and contamination, greenhouse gases emission, etc.), given its high impact and dependence on natural resources [53]. Moreover, all food products carry credence and experience attributes, thus building good reputation strongly affects consumer reactions to corporate and product brands [54]. Consequently, food companies engage heavily in CSR initiatives [55,56]. Typical examples of CSR applications in the food industry concern issues such as consumer attitudes towards CSR in general (e.g., [57]), as well as towards specific CSR targets such as animal welfare (e.g., [58,59], environmental protection (e.g., [60]) and the production of environmentally friendly or organic products (e.g., [61]).

In the current CSR scenario, a hypothetical meat processing company («Robinson Foods») in USA, a typical individualistic culture, target of many CSR-related surveys (see Choi et al., 2016), initiates an environmental CSR activity (i.e., donation of part of its profits to an environmental organization). The scenario manipulates CSR Timing (i.e., proactive: voluntary donation, without any prior environmental incidence or governmental sanctions on the company vs. reactive: initiation of the CSR activity after media reports that the company received a fine for environmental damage); and CSR Intensity (i.e., high: commitment of more financial resources (10 percent of annual profits) than the industry average (5 percent), and time duration of commitment for several years (seven) vs. low: commitment of financial resources at the industry average, and time duration of commitment for one year). Screenshots from Robinson Foods’ fictitious website present the four (2 timing × 2 intensity) CSR conditions (see an example in the Appendix A).

Moreover, a brief written description of the company accompanied each CSR condition: ‘Robinson Foods is one of the largest processors and marketers of meat products. The company operates production facilities across the country, supplying its customers with a wide range of quality meat products. Due to its size, Robinson Foods sells its products at competitive prices, making their brands quite popular among meat consumers’. Besides the company description, a brief news bulletin about the company’s environmentally irresponsible behavior accompanied the reactive CSR Timing scenario: ‘Four weeks ago, media reports showed that Robinson Foods has been fined by the government for deliberately dumping its wastewater, resulted from processing operations, directly into local water streams. The National Environmental Agency has explained that the water streams will need several years of decontamination work because of the irresponsible dumping’.

In order to develop the most appropriate stimuli, an informal focus group with seven meat consumers tested the logic and comprehensibility of each scenario. Focus group members first read the company description and news bulleting, and then saw the stimuli on paper. Then, they had to recall facts about the company (name, span of operations, etc.) and whether it harmed the environment; to identify the chronology of events (in the case of reactive timing); and the company’s actions to solve the problem. Modifications in the stimuli took place upon responses to the above. Similar to Wagner et al. (2009) [51], specific time markers accompanied the final text in the reactive CSR Timing condition so that respondents understand better the chronology of events and statements (i.e., reporting of the negative behavior takes place four weeks prior to the announcement of the CSR campaign).

3.2. Data Collection

Data collection took place by means of a 2 (proactive vs. reactive) × 2 (low vs. high intensity) between-subjects on-line experiment with four consumer groups of equal size. Each group first saw the company description and then the CSR section of the corporate website, while the two proactive CSR Timing groups also saw before the corporate website the news bulleting about the company’s irresponsible behavior. Then, all participants filled out a questionnaire comprising the manipulation checks (4 items), the attributions (13 items), corporate attitude (4 items), and purchase intention (3 items).

Table 1. Source, phrasing and descriptive statistics of the items in the questionnaire.

Group	Description		Mean Score (SD)
			Group 1 (P, H 1) N1 = 65	Group 2 (P, L) N2 = 64	Group 3 (R, H) N3 = 62	Group 4 (R, L) N4 = 69
Attributions (Ellen et al., 2006) (2)
Values-driven	1	The company feels obliged to help.	5.0 (1.64)	4.5 (1.52)	4.5 (2.10)	4.6 (1.96)
	2	The company has a long-term interest in the community.	5.5 (1.34)	4.7 (1.67)	4.0 (2.07)	4.0 (1.99)
	3	The company’s owners or employees believe in this cause.	5.1 (1.42)	4.9 (1.23)	4.3 (1.67)	3.9 (1.78)
	4	The company wants to make it easier for consumers who care about the cause to support it.	5.2 (1.45)	4.9 (1.42)	4.3 (1.84)	4.1 (1.76)
	5	The company is trying to give something back to the community.	5.4 (1.47)	5.2 (1.25)	4.7 (1.88)	4.3 (1.81)
		Cronbach alpha	0.901	0.843	0.931	0.934
Stake-holder -driven	6	The company feels its customers expect it.	4.9 (1.42)	4.9 (1.21)	5.4 (1.49)	5.4 (1.37)
	7	The company feels society in general expects it.	5.1 (1.42)	5.2 (1.16)	5.3 (1.62)	5.4 (1.43)
	8	The company feels their stockholders expect it.	4.5 (1.56)	4.7 (1.09)	5.0 (1.60)	5.1 (1.48)
	9	The company feels their employees expect it.	4.5 (1.50)	4.6 (1.20)	4.5 (1.52)	4.6 (1.60)
		Cronbach alpha	0.936	0.868	0.899	0.876
Egoistic	10	The company wants to get positive publicity.	5.9 (1.22)	5.6 (1.42)	5.9 (1.51)	6.4 (0.92)
	11	The company is taking advantage of the initiative to help its own business.	5.5 (1.33)	4.9 (1.60)	5.0 (1.73)	5.8 (1.20)
		Cronbach alpha	0.849	0.735	0.736	0.671
Strategic	12	The company will get more customers by supporting this initiative.	5.0 (1.33)	5.1 (1.35)	4.5 (1.58)	4.6 (1.51)
	13	The company will keep more of its customers by supporting this initiative.	5.2 (1.32)	5.1 (1.47)	5.1 (1.54)	5.1 (1.40)
		Cronbach alpha	0.928	0.957	0.886	0.883
Attitude towards the company (Spears & Sigh, 2004) (3)
Attitude	14	How appealing is Robinson Foods to you?	4.9 (1.36)	4.5 (1.51)	3.4 (1.60)	3.3 (1.51)
	15	How good is Robinson Foods in your opinion?	5.0 (1.39)	4.8 (1.27)	3.9 (1.63)	3.5 (1.54)
	16	How likeable do you find Robinson Foods?	4.9 (1.37)	6.7 (1.32)	3.6 (1.66)	3.2 (1.59)
	17	How favorable are your feelings towards Robinson Foods?	5.0 (1.44)	4.7 (1.47)	3.6 (1.72)	3.2 (1.58)
		Cronbach alpha	0.967	0.974	0.978	0.977
Purchase intention (Yi, 1990) (4)
Purchase intention	18	How likely is it that you will buy a product from Robinson Foods?	4.6 (1.52)	4.5 (1.69)	3.5 (1.74)	3.3 (1.69)
	19	How probable is it that you will buy a product from Robinson Foods?	4.7 (1.61)	4.6 (1.60)	3.6 (1.76)	3.4 (1.72)
	20	How possible is it that you will buy a product from Robinson Foods?	4.9 (1.42)	4.8 (1.68)	3.8 (1.79)	3.6 (1.59)
		Cronbach alpha	0.982	0.988	0.987	0.977

Note: ¹: P,H = proactive timing, high intensity; P,L = proactive timing, low intensity; R,H = reactive timing, high intensity; R,L = reactive timing, low intensity; ²: 7-point agreement Likert scales with end-points 1 = strongly disagree and 7 = strongly agree ³: 7-point semantic scales with end-points 1 = completely unappealing/bad/non-likeable/unfavorable and 7 = completely appealing/good/likeable/favorable; ⁴: 7-point semantic scales with end-points 1 = completely unlikely/improbable/impossible and 7 = completely likely/probable/possible.

shows the source, exact phrasing and descriptive statistics of all items.

A small-scale pre-test by means of a convenience sample on Facebook with two groups of n = 10 individuals each took place prior to launching the experiment. The pre-test contained open-ended questions with the intention to generate more insights regarding the stimuli and the questionnaire. Relevant answers led to several small additions to make the text more realistic, as well as the addition at the end of the CSR page of the corporate website of a paragraph containing information on how the environment will benefit from the company’s initiative. As a final step, an on-line research platform randomly assigned 260 respondents to each of the four experimental conditions. Average age of the sample is 43 years, while 51 percent of respondents are females (

Table 2. Profile of the four experimental groups.

Group	Size	Mean Age (Years)	Male	Female	Education (Univ. Graduates)
Group 1	65	44.1	52%	48%	56%
Group 2	64	41.1	46%	54%	64%
Group 3	62	42.5	45%	55%	62%
Group 4	69	45.3	54%	46%	58%

Note: no statistically significant differences exist among the groups.

).

4. Analysis and Results

All constructs have satisfactory to very high reliability (see Table 1). The manipulation of CSR Timing (i.e., ‘Τhe company launched the water conservation initiative as a reaction to negative publicity’; 1=‘strongly disagree’) has a significant effect on respondents (M_reactive = 5.4, M_proactive = 2.9, t = −10.75, p < 0.01). In addition, the manipulation of CSR Intensity (i.e., ‘Robinson Foods puts a lot of effort into its environmental initiative’, ‘Robinson Foods invests a lot to its environmental initiative’, ‘Robinson Foods has a long-lasting commitment to its environmental initiative’; 1 = ‘strongly disagree’) also has a significant effect on respondents (M_{high intensity} = 4.6, M_{low intensity} = 3.73, t = −4.19, p < 0.01).

Moreover, testing the dimensionality of the attribution scale against its initial configuration led to the omission of two items. First, the item ‘The company wants to save money’ of the egoistic attribution factor correlated very weakly with the other two items in the same dimension (i.e., items 10 and 11, Table 1), which led to its omission as irrelevant to the current CSR scenario. Second, the item ‘The company hopes to increase profits by supporting this initiative’ correlated with egoistic instead of strategic attributions. The idea that the company may have some financial benefits to gain from the specific CSR initiative might have prompted respondents to make purely company-centered attributions. The confusing nature of this item led to its omission too.

4.1. Effects of CSR Timing

Univariate ANOVAs on consumer responses (

Table 3. Effect of CSR Timing and CSR Intensity on Attributions, Attitude and Purchase Intention.

Predicted Variable:	CSR Timing		F Value	CSR Intensity
	Proactive Mean (SD)	Reactive Mean (SD)		High Mean (SD)	Low Mean (SD)	F Value
Values-driven	25.3 (5.95)	21.3 (8.16)	19.314 **	24.0 (7.52)	22.4 (7.31)	3.079
Stakeholder-driven	19.3 (4.35)	20.8 (4.60)	6.421 *	20.0 (4.76)	20.1 (4.30)	0.029
Egoistic	11.1 (2.33)	11.7 (2.16)	4.125 *	11.4 (2.30)	11.5 (2.22)	0.272
Strategic	10.2 (2.62)	9.6 (2.82)	3.999 *	9.9 (2.76)	9.9 (2.73)	0.046
Attitude towards the company	19.3 (5.33)	13.8 (6.28)	55.261 **	17.3 (6.42)	15.5 (6.41)	4.936 *
Purchase intention	14.0 (4.59)	10.6 (4.99)	31.784 **	12.6 (5.04)	11.9 (5.13)	1.026

Note: ** p < 0.01; * p < 0.05.

) indicate that CSR Timing impacts on attitude towards the company and purchase intention significantly (F_{1, 248} = 55.26, p < 0.01; and F_{1, 248} = 31.78, p < 0.01, respectively). Participants in the proactive CSR Timing condition give the company stronger attitude rating (M = 4.82) and show stronger inclination to purchase the company’s products (M = 4.66) compared to their counterparts in the reactive CSR Timing condition (M = 3.45 and M = 3.53, respectively). These results support Hypotheses 1a and 1b.

The overall effect of CSR Timing on attributions is significant at the entire sample level (F_{4, 2380} = 11.95, p < 0.01, Wilks’ λ = 0.833; F_{Values-driven} = 18.70, p < 0.01; F_{Stakeholder-driven} = 7.42, p < 0.01; F_Egoistic = 3.90; p < 0.05; and F_Strategic = 2.29; p < 0.05). Univariate ANOVAs on each of the attribution dimensions (see Table 3) indicate that CSR Timing has a significant impact on values-driven attributions (F_{1, 248} = 19.31, p < 0.01), stakeholder-driven attributions (F_{1, 244} = 6.42, p < 0.05), egoistic attributions (F_{1, 244} = 4.12, p < 0.05), and strategic attributions (F_{1, 248} = 3.99, p < 0.05). Compared to the reactive CSR Timing condition, respondents in the proactive CSR Timing condition hold significantly stronger values-driven attributions (M_proactive = 5.06, M_reactive = 4.26); weaker stakeholder-driven attributions (M_proactive = 4.82, M_reactive = 5.2); weaker egoistic attributions (M_proactive = 5.55, M_reactive = 5.85); and stronger strategic attributions (M_proactive = 5.1, M_reactive = 4.8). These results support Hypothesis 2a.

4.2. Effects of CSR Intensity

Univariate ANOVAs on consumer responses (Table 3) indicate that CSR Intensity has a significant impact on attitude towards the company (F_{1, 248} = 4.93, p < 0.05). Respondents in the high CSR Intensity condition give the company stronger attitude ratings (M = 4.32) compared to respondents in the low CSR condition (M = 3.87). CSR Intensity does not have a significant effect on purchase intention (F_{1, 248} = 1.02, p > 0.1). These results support Hypothesis 1c but fail to support Hypothesis 1d, even though the effect is in the postulated direction: respondents in the high CSR Intensity condition show stronger intention to purchase the company’s products (M = 4.2) compared to the low CSR Intensity condition (M = 3.96).

CSR Intensity does not have a significant overall effect on attributions at the entire sample level (F_{4, 23} = 1.078; p > 0.1; Wilks’ λ = 0.982). Univariate ANOVAs indicate that CSR Intensity has no significant effect on any of the individual attribution types, even though the effects are in the hypothesized direction (see Table 3). These results fail to support Hypothesis 2b.

4.3. Mediation Analysis

The Preacher and Hayes (2008) [62] indirect macro was considered appropriate to run mediation analysis, due to its ease of use and ability to classify various, previously undetected mediation types [63]. In the current context, the procedure includes two series of two multiple mediation models, one for each of the two depended variables (Figure 2).

Table 4. Multiple mediation results (CSR Timing and CSR Intensity).

Variables	Coefficients				Bootstrapping 1
	a	b	c	c’	a × b	95% Conf. Interval Lower-Upper
CSR Timing
Model 1: Attitude towards the company as DV (F5,237 = 65.691 **; Adj. R2 = 0.572)
			5.24 (0.74) **	2.94 (0.58) **	2.299	1.217–3.311
Values	3.77 (0.87) **	0.45 (0.05) **	-	-
Stakeholder	−1.48 (0.55) **	0.03 (0.07)	-	-
Egoistic	−0.56 (0.32) *	−0.66 (0.13) **	-	-
Strategic	0.49 (0.32)	0.54 (0.13) **	-	-
Model 2: Purchase intention as DV (F5,237 = 29.179 **; Adj. R2 = 0.368)
			3.17 (0.61) **	1.82 (0.56) **	1.347	0.526–2.124
Values	3.77 (0.87) **	0.23 (0.05) **	-	-
Stakeholder	−1.48 (0.55) **	0.03 (0.07)	-	-
Egoistic	−0.56 (0.28) *	−0.42 (0.12) *	-	-
Strategic	0.49 (0.32)	0.52 (0.13) **	-	-
CSR Intensity
Model 1: Attitude towards the company as DV (F5,237 = 55.432 **; Adj. R2 = 0.529)
			1.95 (0.80) *	0.67 (0.56)	1.282	0.184–2.436
Values	2.01 (0.89) *	0.52 (0.05) **	-	-
Stakeholder	−0.07 (0.56)	0.06 (0.07)	-	-
Egoistic	−0.17 (0.29)	−0.68 (0.13) **	-	-
Strategic	0.18 (0.33)	0.54 (0.14) **	-	-
Model 2: Purchase intention as DV (F5,237 = 25.882 **; Adj. R2 = 0.339)
			0.77 (0.64)	0.02 (0.52)	0.756	0.076–1.438
Values	2.01 (0.89) *	0.28 (0.04) **	-	-
Stakeholder	−0.07 (0.56)	0.03 (0.07)	-	-
Egoistic	−0.17 (0.29)	−0.44 (0.12) *	-	-
Strategy	0.18 (0.33)	0.52 (0.33) **	-	-

Note: ** p < 0.01; * p < 0.05; ¹: 1000 bootstrap resamples.

shows the multiple mediation analysis’ results and

Table 5. Individual indirect effects.

				Bootstrapping 1
Mediation		Sobel z-Score	Std. Error	a × b Mean	95% Conf. Int. Lower-Upper
CSR Timing
CSR Timing → Values	→ Attitude	3.870 **	0.439	1.7012	0.9709 to 2.5697
CSR Timing → Stakeholder	→ Attitude	−0.477	0.236	−0.0534	−0.3239 to 0.1799
CSR Timing → Egoistic	→ Attitude	1.844 *	0.231	−0.3795	0.0347 to 0.8070
CSR Timing → Strategic	→ Attitude	1.420	0.191	0.2716	−0.0571 to 0.7158
CSR Timing → Values	→ Purchase intention	3.206 **	0.279	0.8934	0.3962 to 1.5056
CSR Timing → Stakeholder	→ Purchase intention	−0.497	0.106	−0.0532	−0.3110 to 0.1060
CSR Timing → Egoistic	→ Purchase intention	1.712 *	0.142	0.2442	0.0237 to 0.5573
CSR Timing → Strategic	→ Purchase intention	1.423	0.182	0.2633	−0.0500 to 0.6728
CSR Intensity
CSR Intensity → Values	→ Attitude	2.193 *	0.482	1.0572	0.1991 to 1.9934
CSR Intensity → Stakeholder	→ Attitude	−0.132	0.037	0.0050	−0.0946 to 0.1722
CSR Intensity → Egoistic	→ Attitude	0.604	0.201	0.1215	−0.2479 to 0.5334
CSR Intensity → Strategic	→ Attitude	0.542	0.182	0.0990	−0.2320 to 0.5380
CSR Intensity → Values	→ Purchase intention	2.102 *	0.277	1.0513	0.0003 to 1.1521
CSR Intensity → Stakeholder	→ Purchase intention	−0.127	0.017	0.0023	−0.0628 to 0.1329
CSR Intensity → Egoistic	→ Purchase intention	−0.601	0.130	0.0783	−0.1547 to 0.3985
CSR Intensity → Strategic	→ Purchase intention	0.542	0.183	0.0950	−0.2471 to 0.4965

Note: ** p < 0.01; * p < 0.05; ¹:1000 bootstrap resamples.

the individual indirect effects.

CSR Timing is the depended variable in the first two models. Both models explain a good portion of the total variance (adj. R²: 0.57 and 0.37, respectively. Attributions mediate the relationship of CSR Timing with attitude and intention. Values-driven and egoistic attributions have a complementary mediation effect (i.e., a × b × c positive, see Table 4; [63]) on the relationship of CSR Timing with both attitude and purchase intention, while stakeholder-driven and strategic attributions present no mediation effect on either relationships. These results offer partial support to Hypotheses 3a and 3b, i.e., with regard to values-driven and egoistic attributions.

CSR Intensity is the independent variable in the remaining two models. Both models explain a good portion of the total variance (adj. R²: 0.53 and 0.34, respectively). Attributions mediate the relationship of CSR Intensity with attitude and intention. Values-driven attributions have a complementary mediation effect on the relationship of CSR Intensity with attitude, and an indirect-only mediation effect (i.e., c path non-significant [63] on the relationship of CSR Intensity with purchase intention. On the other hand, egoistic, stakeholder-driven and strategic attributions present no mediation effect on either relationship. These results offer partial support to Hypotheses 3c and 3d, i.e., with regard to only value-driven attributions.

5. Discussion, Implications and Future Research

5.1. General Discussion

Responding to the appeal for more research on the contingencies that may shape the relationship between CSR and corporate performance [3], this paper aims at testing the mediating role of consumer attributions on the relationship between specific CSR elements and consumer responses, and examine the degree to which attributions are controllable, i.e., specific CSR elements activate specific attributions. Results support that attributions play an important role in consumer responses to CSR, are complex, and with strong predicting power on consumer outcomes, as past research postulates (i.e., [50,64]).

Overall, results indicate that CSR Timing affects consumers’ attitude and purchase (H2a holds, H3a,b partially hold). Proactive launch of a CSR initiative makes consumers see the company as having purer, values-driven motives, but also pursuing more own-centered, strategic goals. These perceptions then lead to more positive attitudes towards the company and greater intention to purchase the company’s products (see Table 4). On the other hand, CSR Timing does not alter consumer perceptions about the degree of company’s responsiveness to pressure for achieving strategic or stakeholder targets. Moreover, CSR Intensity has a positive effect on consumer attitude, both directly (H1c holds) and indirectly i.e., through attributions (H3c,d partially hold), while it affects purchase intention only indirectly, i.e., through attributions (H1d and H2b do not hold).

A long-lasting and capital-intensive engagement to CSR makes consumers see the company as more virtue-driven. This perception leads to more positive attitude and greater purchase intention. On the other hand, time and capital commitments to CSR do not alter consumer perceptions about the degree of company’s responsiveness to pressure for achieving egoistic, strategic or stakeholder goals.

This study thus maintains that the right time of appearance of, and the appropriate amount of resources committed to a CSR campaign have an impact on how consumers perceive corporate motives. Proactive and high-intensity CSR, through the type of attributions they activate (more positive-external, i.e., values-driven, and less negative-internal, i.e., egoistic), are likely to offset the effect of negative antecedents to consumer responses, such as skepticism [50], suspicion [45], mistrust [65], or willingness to punish the immorally behaving company [66]. On the other hand, reactive and low-intensity CSR might raise perceptions of Corporate Social Irresponsible (CSI) behavior (i.e., [67,68], leveraging more self-relevant (or less virtue-driven) attributions.

5.2. Contribution and Managerial Implications

Attributions are psychological motives that directly influence behavior, i.e., consumer attitudes, loyalty, purchase intention, or word-of-mouth [19]; the exploration of how attributions work remains of great interest to scholars and practitioners. Findings show the importance of attributions in determining CSR effectiveness through activation of positive consumer reactions, and the resulting need to investigate the complex way attributions work in association with specific CSR elements. In this respect, the present study emphasizes on stakeholders’ roles (i.e., consumer attributions) as important external mediators [3]. Moreover, the study adds to past research on attributions and the on-going discussion about their multi-level nature (i.e., [39,40], showing that specific characteristics of the CSR initiative impact on the dimensionality of attributions and, through that, on specific target-types of consumer responses.

This study thus shows that the activation of a dual-level attributions’ system (i.e., company-intrinsic and -extrinsic [25]) is ambivalent, dependent on the character of the CSR campaign. Interestingly, the two CSR elements of this study (i.e., CSR Timing and CSR Intensity) lead to attributions of different dimensionality (i.e., dual, (more) values-driven and (less) egoistic in the Timing condition; and uni-dimensional, values-driven, in the Intensity condition). This indicates that it is harder (easier) to stimulate specific attributional types of a dual configuration in the absence (presence) of specific CSR characteristics. For instance, proactive CSR Timing activates less egoistic attributions, but high CSR Intensity does not.

On the other hand, respondents in the CSR Timing condition, who make both intrinsic and extrinsic attributions, show more positive reactions than respondents in the CSR Intensity condition, whose attributions are purely extrinsic. Attitude and purchase intention increase more between the proactive and the reactive timing conditions (attitude: from 3.45 to 4.82; and intention: from 3.53 to 4.66) than between the low and the high Intensity conditions (attitude: from 3.87 to 4.32; and intention: from 3.96 to 4.2; see Table 3). These findings corroborate past research, showing that when stakeholders assign company-extrinsic motives to CSR initiatives, stakeholders are more willing to accept the simultaneous existence of company-intrinsic motives too [69]. Presence of mixed motives (as activated in this research by Timing) demonstrates that consumers are pragmatic, holding more realistic perceptions towards companies’ objectives. As these consumers get more involved in CSR and its underlying corporate motives, they become more prone to recognize that honest engagement to a CSR initiative must lead to a beneficial outcome that satisfies both the aims of the society and the business objectives of profit seeking organizations [32]. These consumers reconcile inner with outer motives, which together lead to more positive consumer responses.

The above conclusions have important managerial implications. Dual-level CSR attributions can offset negative perceptions of CSR behavior more effectively compared to uni-dimensional CSR attributions. The above conclusions also highlight the importance of designing and implementing corporate-level CSR strategy with elements that have the power to activate dual-level attributions. On the other hand, managers must be cautious in developing and implementing CSR programs with obvious egoistic objectives or high risk of consumers perceiving them as such (for instance, as an immediate reaction to a negative incident).

5.3. Future Research Directions

This study shows that the nature of attributions depends on the type of CSR elements that prevail in consumer attributions’ activation. Responding to the appeal for more research that will investigate further how the mixed motives of CSR programs can lead to favorable outcomes, this work shows that proactive CSR Timing can contribute in this respect by shifting consumer attributions towards a favorable combination (i.e., less egoistic, more value driven) and resulting to more positive outcomes in the context of environmental CSR.

One point of concern is the proper operationalization of CSR target-elements. This study sees CSR Intensity as function of time duration and capital investment. Both parameters, together with investment’s consistency, make what Dwyer, Schurr and Oh (1987) [70] termed CSR commitment. Past research also deals with individual dimensions of CSR commitment. For instance, Ellen et al. (2006) [25] use time duration (i.e., durability, in Dwyer et al., 1987 [70]) they, however, fail to find a significant relationship between this CSR commitment dimension and consumer response (i.e., purchase intent). The same authors assign this failure to the fact that they have treated commitment as a unidimensional construct, since it is possible that ‘consumers may examine the three dimensions (of Dwyer’s et al. (1987) [70]) holistically rather than as individual pieces of information’ (Ellen et al., 2006 [25], p.155). An obvious step for future research then is to incorporate all dimensions of company CSR commitment and explain their exact interplay with attributions and response outcomes. Moreover, future research needs to establish the exact direction and magnitude of this mixed mediational role in relation to an inclusive list of both CSR elements and types of consumer response across a range of CSR contexts.

The use of a fictitious company to avoid the impact of preexisting attitudes towards a real corporate brand is common practice in relevant research; however, this does not remove the potential impact of preexisting beliefs about the wider industry or sector studied. This work deals with CSR in the meat industry, for which the sustainability challenge is among the greatest of all food sectors. The meat industry often suffers from scandals of consumption safety, environmental degradation, animal torture, inadequate labor conditions, fraud and mislabeling [71]. These negative incidences make global consumers skeptical about the ethical basis of meat production [72]. CSR has gained importance in the meat industry in particular, as a way to protect and repair corporate reputation from the destructive effect of bad publicity and product-harmful crises, with the environment being one of the most relevant CSR strategy targets [73]. It is thus possible that participants’ preexisting beliefs towards the meat industry influenced their answers. Future research needs to control for the effect of preexisting beliefs on the mediation results across industries.

Lastly, mediation type and the statistical significance of the direct effect point towards additional theorizing. Zhao et al. (2010) [63] maintain that complementary mediation typically implies that even though empirical evidence for the hypothesized mediator (i.e., attributions) is satisfactory, omission of an additional moderator is quite likely. In the current context (a) proactive CSR Timing makes consumers assign more values-driven motives, which in turn enhance attitude and purchase intention; (b) proactive CSR Timing makes consumer assign less egoistic motives, which also enhance attitude and purchase intension; and (c) high CSR Intensity makes consumers assign more values-driven motives, which in turn enhance attitude. The a × b × c path effects in all the above mediations are positive, while all direct effects are significant (see Table 4). On the contrary, high CSR Intensity shows only an indirect mediation, whereas the direct effect gets insignificant in the presence of the mediator (the concept of full mediation in Baron and Kenny (1986) [74]). This latter is the only mediation effect where the hypothesized mediator (i.e., attributions) is fully consistent with theory. Despite progress this work has made in relation to the mediations (a)–(c) above, there still is room for further work accounting for the direct effect. Future research should thus define additional mediators with an inverse effect (i.e., negative) on the direct relationship [63]. Antecedent constructs with negative effect on behavior, such as skepticism or suspicion, may capture the part of the CSR elements - attitude effects that attributions currently fail to do.