Consumer Decision-Making Styles Extension to Trust-Based Product Comparison Site Usage Model radoslaw macik Maria Curie-Sktodowska University, Poland radoslaw.macik@umcs.pl The paper describes an implementation of extended consumer decision-making styles concept in explaining consumer choices made in product comparison site environment in the context of trust-based information technology acceptance model. Previous research proved that trust-based acceptance model is useful in explaining purchase intention and anticipated satisfaction in product comparison site environment, as an example of online decision shopping aids. Trust to such aids is important in explaining their usage by consumers. The connections between consumer decision-making styles, product and sellers opinions usage, cognitive and affective trust toward online product comparison site, as well as choice outcomes (purchase intention and brand choice) are explored trough structural equation models using pls-sem approach, using a sample of 461 young consumers. Research confirmed the validity of research model in explaining product comparison usage, and some consumer decision-making styles influenced consumers' choices and purchase intention. Product and sellers reviews usage were partially mediating mentioned relationships. Key words: consumer decision-making styles, online product comparison site usage, cognitive and affective trust, products/sellers reviews, purchase intention, pls-sem Introduction Paper goal is to extend trust-based acceptance model in the context of online product comparison site by including consumer decisionmaking styles concept and brand choice on the example of the simulated choice of an automatic coffee machine in a quasi-experimental setting. The sample of 461 young consumers participated in an online quasi-experimental setting. As mentioned extensions were not previously analysed, research made is exploratory in nature. The main research question is which and how consumer decision-making styles influence trust toward product comparison site usage and product brand choice. Data analysis utilises pls-sem approach, management 11 (3): 191-192 Radoslaw Majcik including pls-mga (multi-group analysis) for main chosen brands, as a way to explore postulated relationships. The paper is organised in ten parts. After the short introduction, the online product comparison sites mechanics and business models are presented, with the description of trust-based technology adoption model and introduction to the concept of consumer decisionmaking styles. Next, conceptual research model with research questions is introduced, followed by detailed description of used sample and measures (with reliability and validity assessment). Results part is organised along main research model, its estimates, and multi-group comparisons regarding groups for main chosen brands. Obtained results are discussed in next part of the paper, ending with research implications, limitations of the study, and conclusion. Contemporary Online Product Comparison Sites Common access to online shopping changed buying habits of many consumers in last 20 years. The share of online retail spending (on goods) increased over the time with 15-18 percentage points growth y-o-y, up to over 10% share in total retail on mature markets as United Kingdom (the leader with about 15% share), United States or Germany (see http://www.retailresearch.org/onlineretailing.php). This involves a large number of decisions to find products and sellers online, with two most common alternative approaches: • the choice of the well-known online store brands (like Amazon, Zalando etc.) or places where someone previously bought with satisfaction (without comparison of sellers), or, • finding the best deal - often with the help of product comparison sites. The second strategy is in the scope of presented research. Contemporary online product comparison sites evolved from simple price comparison engines introduced nearly 20 years ago. They are also working under different business model than their predecessors. Product comparison engines are working as infomediaries typically in business model assuming paid integration (via api and parsing structured xml file with offers data) of particular vendor offers with comparison engine. Solutions using bots crawling the net to seek online store and their assortment to include in comparison engine without payment and direct integration are nowadays rare -even Twenga makes possible shop integration for a fee. In both cases, the mechanics of product comparison site working is to aggregate information from product comparison agent (or 220 management • volume 11 Consumer Decision-Making Styles bot), that is configured to gather product information (such as actual price, product availability, product description etc.) from online vendors and/or product information databases. As consumer interacts with product comparison site, typically having recognizable brand, he/she is not interested about underlying technology (allowing the site to present demanded information on request) and/or nature of commercial agreements between comparison site and online vendor, this suggests that product comparison agent should be transparent to the comparison site user. Aggregated information retrieved on online shopper request is revealed to him/her in the form of ranking. By interacting with product comparison site consumers leave some traces of their behaviour, that are valuable for online vendors and comparison sites for their marketing activities. Figure 1 shows the flows between online vendor, product comparison site, and consumer. Product comparison sites are nowadays enhanced by opinions from consumers about products and sellers (possibly so called 'trusted opinions' of non-anonymous for the site users who bought a particular product). Those opinions are usually presented as average ratings - particularly for sellers' credibility and detailed pieces of text. Young consumers are more innovative toward information technology usage. They also are using online decision shopping aids including mobile tools more often and in the more extensive way (Ma-cik and Nalewajek 2013), so studying this group behaviour can be useful to make predictions by analogy for consumers later accepting new technologies. Previous research also suggests the power of online opinions and reviews for this group of consumers (Nalewajek and Mapk 2013). The influence of online reviews on purchasing behaviour has been confirmed by many studies in the information systems and consumer behaviour fields (e.g., Forman, Ghose, and Wiesenfeld 2008; Kham-mash and Griffiths 2011). Typically the effect of positive and negative reviews for particular e-commerce site have been studied, and product reviews have been left from detailed consideration. Negative reviews are believed to have a stronger effect on consumer decisions than positive ones (Park and Lee 2009), as being more diagnostic and informative (Lee, Park, and Han 2008). Typically the consumer using product comparison site faces with a mix of positive and negative product reviews and seller opinions, this is known as inconsistent reviews setting (Tsang and Prendergast 2009). For this study, both types of opinions have been used: about products and about sellers. number 3 • fall 2016 227 Radoslaw Majcik figure 1 Flows Between Online Vendor, Product Comparison Site and Consumer: Simplified Approach (numbers represent steps of flows between ecosystem members; own elaboration, loosely based on concept of Wan, Menon, and Ramaprasad (2007, 66)) Focus was on declared number of opinions read more or less precisely, leaving out of consideration their negative or positive connotations, under the assumption: the more opinions read, the greater trust to product comparison site. Trust-Based Acceptance Model Numerous research studies show that trust toward online business is a key driver for the success of e-commerce (Cheung and Lee 2006; Hong and Cho 2011), particularly for online retailers (Kim and Park 2013). Many studies researching consumer trust toward e-commerce site are following Komiak and Bensabat (2006) trust-based acceptance model built upon widely used in e-commerce studies theory of reasoned action (tra) (Hoehle, Scornavacca, and Huff 2012; Komiak and Benbasat 2006). According to tra individual's behaviour is pre- 220 management • volume 11 Consumer Decision-Making Styles dieted by his/her behavioural intention, while behavioural intention is formed as an effect of attitude, beliefs, and subjective norms (Fish-bein and Ajzen 1975). Those connections are causal relationships, so they are typically modelled using sem approach. Komiak and Benbasat (2006) developed mentioned trust-based acceptance model for explaining the adoption of online recommendation agents. They examined two types of trust in the model: cognitive trust and affective trust. Cognitive trust is conceptualized as trusting beliefs while affective trust should be considered as a form of trusting attitude. In online environments, consumers often affectively evaluate trusting behaviour. High affective trust suggests having favourable feelings toward performing the behaviour. The trust-based acceptance model highlights that cognitive trust affects emotional trust, which further leads to individuals' adoption intention (Komiak and Benbasat 2006). This is convergent with tra approach when adoption process resembles the following sequence: belief 'attitude' intention, although the subjective norm is the construct dropped in trust-based acceptance model as adoption behaviour is considered as voluntary rather than mandatory according to Komiak and Benbasat (2006). Cognitive trust can be analysed in three main categories: competence, benevolence, and integrity as suggest McKnight, Choudhury, and Kacmar (2002). Trust in competence refers to the extent to which consumers perceive an online retailer or service provider as having skills and abilities to fulfil what they need (Mayer, Davis, and Schoor-man 1995). Trust in benevolence is consumers' perception that the retailer/service provider will act in their interest (Hong and Cho 2011). Trust in integrity refers to consumers' perception about honesty and promise-keeping by online retailer/service provider (McKnight, Choudhury, and Kacmar 2002). Those concepts are used in this research in the context of product comparison engine usage. Consumer Decision-Making Styles A consumer decision-making style concept is defined as 'a mental orientation characterizing a consumer's approach to making choices' (Sproles and Kendall 1986, 268), and consumer decision-making styles can be perceived as 'basic buying-decision making attitudes that consumers adhere to, even when they are applied to different goods, services or purchasing decisions' (Walsh et al. 2001, p. 121). Consumer decision-making styles are connected to consumer personality, and research suggest that they are relatively stable constructs (Sproles and Kendall 1986; Lysonsky, Durvasula and Zotos number 3 • fall 2016 227 Radoslaw Majcik table 1 Description of Consumer Decision-Making Styles: Extended version Style name/short name Description Perfectionistic perf Sensitive to high quality products, prone to spend money and/or time to get the expected quality expecting customer care, thoroughly comparing the available options Brand-Conscious bc Believing that price of branded products is appropriate to their quality, buying well-known and heavily advertised brands, often in shopping malls and specialty stores Novelty Fashion Conscious nfc Willing to put extra effort to obtain a trendy, new products sooner than others; follower of fashion, always in line with current trends, often buys due variety-seeking motives Recreational Shopping Conscious rsc Hedonistic, perceiving shopping environment as pleasant and desirable, spending much time on shopping Price-Value Conscious pvc Prone for getting highest possible 'value for money' - sensitive to price reductions, looking for low prices, often carefully comparing products before purchase, rarely buys cheapest products Impulsive imp Relying on impulse to buy does not plan purchases, not paying much attention to how much is spending, prone for buying on sales Confused by Overchoice co Feels the fatigue of to many products, brands and shopping options, often has trouble in deciding Habitual Brand-Loyal hbl Has strong habits for buying specific brands and/or at the same places Compulsive comp Having tendency to uncontrolled spending, and addiction for shopping (style added by author) Ecologically Aware eco Prone to choose products that are ecologically safe for him/her and for environment (style added by author) notes Own elaboration, including early insights by Sproles and Kendall (1986). 1995). Particular shopping activities and attitudes toward shopping can be perceived as direct outcomes of consumer's decision-making styles (Tai 2005), and tendencies revealed in particular person styles profile are modified in particular shopping process by situational factors. Consumer decision-making concept has been used in several contemporary studies (Walsh et al. 2002; Tai 2005), and proved to be useful to explain outcomes of particular shopping activities and attitudes toward shopping, including usage of online channel (Mapk and Macik 2009). Consumer decision-making styles are measured typically via pcs (Profile of Consumer Style) questionnaire proposed by Sproles and Kendall (1986). In this research extension and reconstruction of pcs has been used, with 2 new styles have been added on the base of pre- 220 management • volume 11 Consumer Decision-Making Styles |rq2 |rq3 | Consumer Decision-Making Styles (extended to 10 styles) \ figure 2 Conceptual Research Model vious author research. In result 10 styles (including original 8) were measured by 30 items scaled as Likert-type scale with five variants of answers (short form of reconstructed by Mapk and Mapk (2015) pcs scale named spdzi4k). Those styles are described in greater detail in table 1. Listed styles are forming personal profile consumer decisionmaking styles - particular person possesses an individual combination of them, when all styles are manifesting itself on different levels, with some styles more intense or prominent (Sproles and Kendall 1986). Conceptual Research Model and Research Questions Mentioned concepts of trust-based adoption model and consumer decision-making styles putted in context of online product comparison sites usage were leading to propose conceptual model (figure 2). In this approach gained with time experience in online product comparison sites usage and opinions about products and sellers are antecedents for cognitive and affective trust for online product comparison site according to trust-based adoption model, where cognitive trust measured in three sub-dimensions (trust in competence, trust in benevolence and trust in integrity) influences affective trust and later purchase intention. Experience with opinions usage and trust-based adoption model constructs are explained by some of consumer decision-making styles measured in ten dimensions (it was assumed that only selected styles will be useful). number 3 • fall 2016 227 Radoslaw Majcik Because of exploratory character of the study three main research questions have been formulated: rqi How previous consumer experience with product comparison site usage and opinions about products and sellers usage are connected with trust toward product comparison site constructs from trust-based adoption model? rq2 Which and how consumer decision-making styles are influencing the level of experience with product comparison site usage and opinions about products and sellers? rq3 Which and how consumer decision-making styles are influencing constructs from trust-based adoption model for product comparison site usage? No exact hypotheses were assumed for this research, particularly the set of consumer decision-making styles included in model was exactly exploratory, and modified during the modelling. Research model derived from conceptual one has been assessed via structural equation modelling approach utilizing pls-sem - recommended for exploratory stages of theory extensions (Hair, Ringle, and Sarstedt 2011) - and later via multi-group analysis using pls-mga algorithms. Sample and Measures sample Data have been collected during March 2015 through cawi questionnaire with e-mail invitation sent to authors students and their peers, that returned 461 usable responses from 575 sent invitations, giving response rate of 80.2%. Students were awarded small increase in course activity grade for participation and recruitment of their peers (this award was less than 4% of total possible grade). In effect, the sample consists of 60.2% women and 39.8% men. The average age of participants is 24.5 years with standard deviation of 5.1 years (range: 18-46 years old, median: 23 years). Each 1/3rd of participants were inhabitants of different level of urbanization areas: rural areas, small towns and larger cities. All participants must be active internet users and make at least one online purchase during a year prior study. Sample structure regarding gender and age is close to population of both full-time and part-time students of public university located in the South-Eastern part of Poland, where the data have been collected. measures Items to measure constructs used in this study were adapted from previously published research or have been developed by the au- 220 management • volume 11 Consumer Decision-Making Styles table 2 Scales Used in Study Construct (1) (2) (3) (4) Consumer experience in product comparison sites usage Consumer Experience Own developmenta N/A 9 Cognitive Trust in Competence ct_Competence McKnight, Choudhury and Kacmar (2002) travestation 4 (3)b Cognitive Trust in Benevolence ct_Benevolence McKnight, Choudhury, and Kacmar (2002) travestation 4 (3)b Cognitive Trust in Integrity ct_Integrity McKnight, Choudhury, and Kacmar (2002) travestation 4 (3)b Affective Trust Affective Trust Komiak and Ben-basat (2006) reconstruction 4 Purchase Intention Purchase Intention Gefen, Kara-hanna and Straub (2003) reconstruction 4 Product Reviews Usage Product Reviews Usage Own developmenta N/A 2 Sellers Reviews Usage Sellers Reviews Usage Own developmenta N/A 2 Brand Conscious Style bc Sproles and Kendall (1986) reconstruction 3 Confused by Overchoice Style co Sproles and Kendall (1986) reconstruction 3 Ecologically Aware Style eco Own developmentc N/A 3 Perfectionistic Style perf Sproles and Kendall (1986) reconstruction 3 Continued on the next page thor. As questionnaire language was Polish, this required to translate and culturally adapt (by authors) scales written originally in English, including reconstruction procedures where needed. In effect, used scales are derived from original measures. Basic data about used scales is provided in table 2. Data analysis for this study has been performed using Smartpls 3.2 software (see www.smartpls.com), as most of the measurement variables were not normally distributed. Bootstrap procedure (resampling with replacement, sample size equal of original sample size - 461 observations) with 10000 repetitions for pls procedure and 5000 repetitions for pls-mga algorithm has been utilised to get inference statistics for measures and evaluated models. number 3 • fall 2016 227 Radoslaw Majcik table 2 Continued from the previous page Construct (1) (2) (3) (4) Price-Value Conscious Style pvc Sproles and Kendall (1986) reconstruction 3 Recreational Shopping Conscious Style rsc Sproles and Kendall (1986) reconstruction 3 notes Column headings are as follows: (1) name in tables and diagrams, (2) items derived from, (3) level of adaptation, (4) number of items. Only consumer decisionmaking styles included in model are shown in table, other four excluded. a Used also in Macik and Macik (2016b). b One item dropped due to low factor loading. cUsed also inMacik and Macik (2016a). Only consumer decision-making styles included in model are shown in table, other four excluded. reliability and validity of measures Reliability of measures in this study has been assessed by two commonly used measures: Cronbach's Alpha coefficient and Composite Reliability (cr) measure, as they represent lower and upper boundaries of true scale reliability respectively (Henseler, Ringle, and Sarstedt 2015). Using both criterions reliability of most constructs meets typical requirements - values of crs are all over suggested value 0.7 (Hair, Ringle, and Sarstedt 2013, 7), with some Alphas for co, perf and pvc lower than desired - tables 3 and 4. table 3 Reliability of Measures: Cronbach's Alpha Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust 0.802 0.801 0.020 39.344 0.000 0.758 0.837 bc 0.719 0.718 0.024 29.446 0.000 0.667 0.762 co 0.618 0.616 0.035 17.797 0.000 0.542 0.679 ct in Benevolence 0.713 0.710 0.029 24.506 0.000 0.649 0.763 ct in Competence 0.732 0.730 0.027 27.218 0.000 0.675 0.779 ct in Integrity 0.777 0.775 0.023 33.363 0.000 0.726 0.817 Consumer Experience 0.928 0.928 0.006 157.430 0.000 0.916 0.938 eco 0.788 0.788 0.020 39.697 0.000 0.746 0.824 perf 0.566 0.565 0.031 18.258 0.000 0.501 0.623 pvc 0.617 0.615 0.036 17.283 0.000 0.541 0.680 Product Reviews Usage 0.788 0.788 0.025 31.112 0.000 0.735 0.834 Purchase Intention 0.797 0.796 0.021 37.739 0.000 0.750 0.833 rsc 0.867 0.866 0.012 74.794 0.000 0.841 0.888 Sellers Reviews Usage 0.835 0.834 0.021 40.181 0.000 0.791 0.872 notes (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) i-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. 220 management • volume 11 Consumer Decision-Making Styles table 4 Reliability of Measures: Composite Reliability (cr) Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust 0.871 0.870 0.012 75.399 0.000 0.846 0.891 bc 0.840 0.829 0.035 23.985 0.000 0.662 0.852 co 0.787 0.772 0.047 16.631 0.000 0.569 0.810 ct in Benevolence 0.839 0.838 0.014 61.789 0.000 0.810 0.863 ct in Competence 0.849 0.848 0.013 66.398 0.000 0.822 0.872 ct in Integrity 0.871 0.870 0.012 74.503 0.000 0.845 0.891 Consumer Experience 0.940 0.940 0.005 198.685 0.000 0.930 0.948 eco 0.870 0.848 0.083 10.531 0.000 0.263 0.885 perf 0.770 0.762 0.024 31.480 0.000 0.678 0.789 pvc 0.794 0.789 0.019 42.068 0.000 0.736 0.816 Product Reviews Usage 0.904 0.904 0.010 87.166 0.000 0.884 0.924 Purchase Intention 0.868 0.867 0.012 72.181 0.000 0.841 0.889 rsc 0.915 0.909 0.036 25.182 0.000 0.850 0.926 Sellers Reviews Usage 0.924 0.923 0.009 104.208 0.000 0.906 0.940 notes (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) i-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. table 5 Convergent Validity of Measures: Average Variance Extracted (ave) Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust 0.628 0.627 0.024 26.461 0.000 0.580 0.672 bc 0.637 0.624 0.036 17.838 0.000 0.444 0.658 co 0.558 0.546 0.038 14.834 0.000 0.390 0.590 ct in Benevolence 0.635 0.634 0.023 27.447 0.000 0.587 0.678 ct in Competence 0.652 0.651 0.022 29.196 0.000 0.607 0.694 ct in Integrity 0.692 0.691 0.022 31.513 0.000 0.645 0.731 Consumer Experience 0.636 0.636 0.019 33.588 0.000 0.598 0.672 eco 0.691 0.667 0.072 9.558 0.000 0.238 0.721 perf 0.534 0.528 0.022 24.125 0.000 0.464 0.558 pvc 0.564 0.560 0.026 22.127 0.000 0.497 0.601 Product Reviews Usage 0.825 0.825 0.017 47.873 0.000 0.792 0.859 Purchase Intention 0.622 0.621 0.024 25.756 0.000 0.572 0.667 rsc 0.782 0.772 0.042 18.623 0.000 0.646 0.807 Sellers Reviews Usage 0.858 0.858 0.015 56.201 0.000 0.827 0.887 notes (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) i-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. Convergent validity for used measures assessed via Average Variance Extracted (ave) is very good - all constructs are meeting the cri- number 3 • fall 2016 227 Radoslaw Majcik table 6 Discriminant Validity of Measures: Fornell Larcker Criterion Const. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (n) (12) (13) (14) (1) 0.792 (2) 0.171 0.798 (3) -0.039 0.150 0.747 (4) 0.609 0.096 -0.026 0.797 (5) 0.747 0.161 -0.022 0.657 0.807 (6) 0.633 0.099 -0.035 0.691 0.700 0.832 (7) 0.246 0.086 0.067 0.100 0.205 0.167 0.798 (8) 0.132 -0.013 0.081 0.047 0.069 0.019 0.210 0.831 (9) 0.070 0.310 -0.020 -0.018 0.111 0.014 0.220 0.125 0.731 (10) 0.188 0.181 0.156 0.152 0.232 0.175 0.162 0.199 0.233 0.751 (11) 0.262 0.110 0.173 0.121 0.185 0.121 0.327 0.145 0.142 0.074 0.908 (12) 0.739 0.191 0.022 0.500 0.597 0.504 0.285 0.148 0.039 0.209 0.223 0.788 (13) 0.188 0.171 0.073 0.143 0.162 0.148 0.040 0.090 0.069 0.336 0.053 0.198 0.884 (14) 0.228 0.143 0.033 0.202 0.209 0.140 0.327 0.057 0.169 0.072 0.571 0.211 0.013 0.926 notes Constructs: (1) Affective Trust, (2) bc, (3) co, (4) ct in Benevolence, (5) ct in Competence, (6) ct in Integrity, (7) Consumer Experience, (8) eco, (9) perf, (10) pvc, (11) Product Reviews Usage, (12) Purchase Intention, (13) rsc, (14) Sellers Reviews Usage. Numbers on matrix diagonal are square roots from ave for constructs; numbers off-diagonal are correlations between them, this is alternative form to report Fornell-Larcker Criterion (Henseler, Ringle, and Sarstedt 2014, 117). terion of ave value higher than 0.5 as suggested by Fornell and Larcker (1981) - table 5. Even for constructs having lower internal consistency in terms of Cronbach's Alpha (co, perf and pvc) the ave values are at least satisfactory. Discriminant validity of used measures is also at very good (table 6). The Fornell-Larcker Criterion stating that ave for each construct should be higher from all squared correlations between particular construct and other measures (Fornell and Larcker 1981) is met for all constructs (see also note for table 6, as in mentioned table this criterion is reported in alternative form). Results whole sample model On the base or conceptual model shown on figure 2 and initial data analysis structural equations model presented on figure 3 has been estimated using SmartPLS 3.2 software. Previous analysis (Ma-cik and Macik 2016) confirmed the validity of trust-based adoption model to explain purchase intention in product comparison site environment. In structural model depicted on figure 3 consumer experience explains both reviews constructs, that are also interconnected - as in virtual channel product choice is typically made earlier than vendor/seller choice, so it was assumed that product review usage should explain sellers review usage, also because of similar factors influencing reviews following as a whole - persons more often using 220 management • volume 11 Consumer Decision-Making Styles figure 3 Research Model with Results Obtained via pls-sem (values on paths are standardized path coefficients with bootstrap obtained p-values reported in parentheses) product reviews inside product comparison engine are more likely more heavily relying on sellers reviews, to establish sellers credibility. Estimated model exhibits reasonable fit - proportion of variance explained, measured with R2 statistics is over 0.5 for main explained variables, particularly 0.612 for Affective Trust and 0.551 for Purchase Intention. The level of coefficients of determination (R2) for all constructs playing roles of dependent variables are presented in table 7. Also low srmr (Square Root of Mean Residuals) value on the level of 0.039 suggests reasonable model fit to the data. Table 8 presents in detail path coefficient values in original sample and inference statistics for paths obtained via bootstrapping. Path coef-ficientsfrom original sample with significance levels are also shown on figure 3. In general, consumer experience with product and sellers reviews number 3 • fall 2016 227 Radoslaw Majcik table 7 Coefficients of Determination for Dependent Variables in Estimated Model (R2 Values) Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust 0.612 0.617 0.034 17.974 0.000 0.559 0.691 ct in Benevolence 0.436 0.438 0.046 9.528 0.000 0.352 0.530 ct in Competence 0.106 0.119 0.029 3.676 0.000 0.086 0.207 ct in Integrity 0.482 0.483 0.044 11.005 0.000 0.401 0.571 Consumer Experience 0.061 0.071 0.025 2.462 0.014 0.038 0.143 Product Reviews Usage 0.130 0.137 0.030 4.267 0.000 0.090 0.215 Purchase Intention 0.551 0.555 0.041 13.311 0.000 0.482 0.640 Sellers Reviews Usage 0.348 0.351 0.040 8.603 0.000 0.279 0.436 notes (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) i-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. table 8 Path Coefficients in Estimated Model Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust ^ Purchase Intention 0.726 0.726 0.032 22.941 0.000 0.663 0.786 bc ct in Competence 0.088 0.097 0.046 1.916 0.055 0.024 0.206 co ^ Product Reviews Usage 0.152 0.159 0.043 3.529 0.000 0.089 0.256 ct in Benevolence ^ Affective Trust 0.146 0.146 0.051 2.866 0.004 0.045 0.245 ct in Benevolence ^ ct in Integrity 0.680 0.678 0.033 20.335 0.000 0.607 0.737 ct in Competence ^ Affective Trust 0.515 0.513 0.049 10.578 0.000 0.411 0.603 ct in Competence ^ ct in Benevolence 0.643 0.642 0.038 16.957 0.000 0.567 0.714 ct in Integrity ^ Affective Trust 0.156 0.157 0.049 3.208 0.001 0.064 0.257 Consumer Experience ^ Product Reviews Usage 0.317 0.317 0.042 7.529 0.000 0.235 0.402 Consumer Experience ^ Sellers Reviews Usage 0.157 0.158 0.044 3.600 0.000 0.074 0.245 eco ^ Affective Trust 0.069 0.072 0.033 2.073 0.038 0.015 0.142 perf ^ Consumer Experience 0.193 0.202 0.050 3.849 0.000 0.120 0.313 pvc ^ ct in Competence 0.174 0.176 0.056 3.107 0.002 0.071 0.288 pvc ^ ct in Integrity 0.072 0.074 0.037 1.967 0.049 0.005 0.149 pvc ^ Consumer Experience 0.117 0.119 0.053 2.222 0.026 0.020 0.227 pvc ^ Purchase Intention 0.072 0.074 0.033 2.218 0.027 0.015 0.143 Product Reviews Usage ^ Affective Trust 0.121 0.121 0.033 3.678 0.000 0.057 0.186 Product Reviews Usage ^ Sellers Reviews Usage 0.519 0.519 0.039 13.342 0.000 0.442 0.594 rsc ^ ct in Competence 0.086 0.091 0.041 2.091 0.037 0.020 0.181 Sellers Reviews Usage ^ ct in Benevolence 0.068 0.068 0.036 1.907 0.057 -0.003 0.136 Sellers Reviews Usage ^ ct in Competence 0.182 0.181 0.044 4.122 0.000 0.090 0.265 notes (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) i-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. usage are loosely connected with trust-based adoption model constructs. Also the direct influence of six selected (on the base of correlation analysis) consumer decision-making styles is not so strong, although those relationships are statistically significant. Magnitude of consumer decision-making styles influence increases when total effects (including indirect effects) are taken into account. As the model is quite complicated, some indirect effects are pres- 220 management • volume 11 Consumer Decision-Making Styles table 9 Total Effects in Estimated Model Constructs (1) (2) (3) (4) (5) (6) (7) Affective Trust — Purchase Intention 0.726 0.726 0.032 22.941 0.000 0.663 0.786 *bc — Affective Trust 0.060 0.066 0.032 1.872 0.061 0.015 0.143 *bc — ct in Benevolence 0.057 0.062 0.030 1.895 0.058 0.014 0.134 bc — ct in Competence 0.088 0.097 0.046 1.916 0.055 0.024 0.206 *bc — ct in Integrity 0.039 0.042 0.021 1.852 0.064 0.009 0.093 *bc — Purchase Intention 0.043 0.048 0.024 1.825 0.068 0.010 0.106 *co — Affective Trust 0.030 0.031 0.010 2.960 0.003 0.014 0.053 *co — ct in Benevolence 0.015 0.015 0.005 2.669 0.008 0.006 0.027 *co — ct in Competence 0.014 0.015 0.005 2.629 0.009 0.006 0.027 *co — ct in Integrity 0.010 0.010 0.004 2.679 0.007 0.004 0.018 co — Product Reviews Usage 0.152 0.159 0.043 3.529 0.000 0.089 0.256 *co — Purchase Intention 0.021 0.022 0.007 2.994 0.003 0.010 0.038 *co — Sellers Reviews Usage 0.079 0.083 0.023 3.465 0.001 0.045 0.134 ct in Benevolence — Affective Trust 0.252 0.252 0.046 5.496 0.000 0.164 0.343 ct in Benevolence — ct in Integrity 0.680 0.678 0.033 20.335 0.000 0.607 0.737 *ct in Benevolence — Purchase Intention 0.183 0.183 0.035 5.286 0.000 0.117 0.252 ct in Competence — Affective Trust 0.676 0.675 0.035 19.395 0.000 0.601 0.739 ct in Competence — ct in Benevolence 0.643 0.642 0.038 16.957 0.000 0.567 0.714 *ct in Competence — ct in Integrity 0.437 0.436 0.042 10.424 0.000 0.352 0.516 *ct in Competence — Purchase Intention 0.491 0.490 0.039 12.588 0.000 0.412 0.564 ct in Integrity — Affective Trust 0.156 0.157 0.049 3.208 0.001 0.064 0.257 *ct in Integrity — Purchase Intention 0.113 0.114 0.035 3.209 0.001 0.046 0.184 *Consumer Experience — Affective Trust 0.084 0.083 0.018 4.678 0.000 0.050 0.119 *Consumer Experience — ct in Benevolence 0.060 0.059 0.017 3.510 0.000 0.027 0.092 *Consumer Experience — ct in Competence 0.059 0.059 0.017 3.396 0.001 0.025 0.093 *Consumer Experience — ct in Integrity 0.041 0.040 0.012 3.501 0.000 0.018 0.062 Consumer Experience — Product Reviews Usage 0.317 0.317 0.042 7.529 0.000 0.235 0.402 *Consumer Experience — Purchase Intention 0.061 0.061 0.013 4.608 0.000 0.036 0.086 Consumer Experience — Sellers Reviews Usage 0.322 0.323 0.045 7.180 0.000 0.236 0.409 eco — Affective Trust 0.069 0.072 0.033 2.073 0.038 0.015 0.142 *eco — Purchase Intention 0.050 0.052 0.024 2.076 0.038 0.010 0.103 *perf — Affective Trust 0.016 0.017 0.005 2.998 0.003 0.008 0.029 *perf — ct in Benevolence 0.012 0.012 0.005 2.500 0.012 0.004 0.022 *perf — ct in Competence 0.011 0.012 0.005 2.429 0.015 0.004 0.022 *perf — ct in Integrity 0.008 0.008 0.003 2.506 0.012 0.003 0.015 perf — Consumer Experience 0.193 0.202 0.050 3.849 0.000 0.120 0.313 *perf — Product Reviews Usage 0.061 0.064 0.018 3.371 0.001 0.034 0.106 *perf — Purchase Intention 0.012 0.012 0.004 2.965 0.003 0.006 0.021 *perf — Sellers Reviews Usage 0.062 0.066 0.020 3.094 0.002 0.034 0.113 *pvc — Affective Trust 0.139 0.140 0.041 3.345 0.001 0.063 0.225 *pvc — ct in Benevolence 0.119 0.120 0.037 3.226 0.001 0.052 0.195 pvc — ct in Competence 0.181 0.183 0.057 3.197 0.001 0.076 0.296 Continued on the next page ent. As total effect is the sum of direct effect and indirect effect(s), only direct and total effects are reported (tables 8 and 9). The indirect effect, in this case, is easy to calculate as the difference between total and direct effects (or as multiplication of particular path coefficients). In the case of lack of direct relationship total effect equals indirect effect - such cases are marked with asterisk table 9. number 3 • fall 2016 227 Radoslaw Majcik table 9 Continued from the previous page *pvc — ct in Integrity 0.153 0.155 0.052 2.939 0.003 0.057 0.262 pvc — Consumer Experience 0.117 0.119 0.053 2.222 0.026 0.020 0.227 *pvc — Product Reviews Usage 0.037 0.038 0.017 2.122 0.034 0.005 0.074 pvc — Purchase Intention 0.173 0.176 0.044 3.942 0.000 0.097 0.269 *pvc — Sellers Reviews Usage 0.038 0.039 0.018 2.064 0.039 0.005 0.077 Product Reviews Usage — Affective Trust 0.194 0.193 0.036 5.334 0.000 0.122 0.263 *Product Reviews Usage — ct in Benevolence 0.096 0.095 0.024 3.957 0.000 0.047 0.142 *Product Reviews Usage — ct in Competence 0.095 0.094 0.025 3.829 0.000 0.045 0.143 *Product Reviews Usage — ct in Integrity 0.065 0.065 0.017 3.936 0.000 0.031 0.095 *Product Reviews Usage — Purchase Intention 0.141 0.140 0.027 5.301 0.000 0.088 0.191 Product Reviews Usage — Sellers Reviews Usage 0.519 0.519 0.039 13.342 0.000 0.442 0.594 *rsc — Affective Trust 0.058 0.061 0.028 2.045 0.041 0.013 0.125 *rsc — ct in Benevolence 0.055 0.058 0.027 2.027 0.043 0.012 0.119 rsc — ct in Competence 0.086 0.091 0.041 2.091 0.037 0.020 0.181 *rsc — ct in Integrity 0.037 0.040 0.019 1.961 0.050 0.007 0.083 *rsc — Purchase Intention 0.042 0.045 0.021 1.989 0.047 0.009 0.093 *Sellers Reviews Usage — Affective Trust 0.140 0.139 0.031 4.540 0.000 0.075 0.197 Sellers Reviews Usage — ct in Benevolence 0.185 0.184 0.045 4.148 0.000 0.094 0.269 Sellers Reviews Usage — ct in Competence 0.182 0.181 0.044 4.122 0.000 0.090 0.265 *Sellers Reviews Usage — ct in Integrity 0.126 0.125 0.031 4.123 0.000 0.061 0.181 *Sellers Reviews Usage — Purchase Intention 0.102 0.101 0.023 4.409 0.000 0.053 0.144 notes Column headings are as follows: (1) original sample (o). Bootstrap estimates: (2) sample mean (M), standard error (sterr), (4) f-statistics (|o/sterr|), (5) p-values. Bootstrap bias corrected 95% confidence interval: (6) low, (7) up. * indirect effect only multi group comparisons regarding chosen brand In this study, consumers were expected to make choice of an automatic coffee machine (as a suggestion for a neighbour buy) in product comparison site environment. This choice has been recorded on the level of particular product recognizable by exact type (described as producer alphanumerical code). To form groups for comparison chosen brand has been used. Study participants can choose any of brands available in product comparison site although better-known brands (of large general table 10 Structure of Brand Choices Made by Research Participants with Size of Groups for pls-mga Groups of brands Brand name Group size (n) Share (%) Included for PLS-MGA analysis Saeco 150 32.5 De Longhi 107 23.2 Krups 75 16.3 Bosch 63 13.7 Siemens 42 9.1 Excluded from PLS-MGA analysis Severin 3 0.7 Zelmer 3 0.7 other 18 3-9 220 management • volume 11 table il Path Coefficients in Five Analysed Brand Groups Paths (direct effects) Coefficient estimates p-values (from bootstraping) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Affective Trust ->■ Purchase Intention 0.707 0.744 0.639 0.773 0.638 0.000 0.000 0.000 0.000 0.000 bc ->■ ct in Competence 0.001 0.052 0.208 0.071 0.246 0.993 0.693 0.172 0.493 0.274 co Product Reviews Usage 0.318 -0.018 0.260 0.183 0.244 0.003 0.872 0.325 0.024 0.152 ct in Benevolence ->■ Affective Trust 0.197 0.304 -0.033 0.211 0.019 0.147 0.001 0.718 0.033 0.939 ct in Benevolence ->■ ct in Integrity 0.680 0.644 0.523 0.722 0.818 0.000 0.000 0.000 0.000 0.000 ct in Competence ->■ Affective Trust 0.511 0.390 0.579 0.562 0.385 0.000 0.000 0.000 0.000 0.023 ct in Competence ->■ ct in Benevolence 0.764 0.565 0.473 0.696 0.614 0.000 0.000 0.000 0.000 0.000 ct in Integrity Affective Trust 0.179 0.102 0.295 0.020 0.359 0.179 0.265 0.007 0.816 0.111 Consumer Experience ->■ Product Reviews Usage 0.114 0.372 0.286 0.398 0.030 0.420 0.000 0.023 0.000 0.878 Consumer Experience ->■ Sellers Reviews Usage 0.229 0.235 0.115 0.199 -0.131 0.043 0.006 0.353 0.009 0.435 eco —* Affective Trust 0.048 0.055 0.093 0.145 0.046 0.567 0.539 0.397 0.010 0.776 peef ->■ Consumer Experience 0.314 0.311 0.073 0.163 0.020 0.002 0.000 0.784 0.085 0.943 pvc ct in Competence 0.411 0.121 0.287 0.124 -0.029 0.004 0.448 0.030 0.325 0.850 pvc ->■ ct in Integrity 0.223 0.000 0.089 0.041 0.199 0.036 0.998 0.395 0.602 0.061 pvc ->■ Consumer Experience 0.260 0.155 0.082 0.037 0.360 0.065 0.226 0.657 0.773 0.037 pvc ->■ Purchase Intention 0.078 0.059 0.261 -0.038 0.150 0.366 0.410 0.003 0.646 0.412 Product Reviews Usage ->■ Affective Trust 0.195 0.215 -0.032 0.079 0.325 0.016 0.001 0.682 0.149 0.002 Product Reviews Usage ->■ Sellers Reviews Usage 0.543 0.411 0.555 0.527 0.538 0.000 0.000 0.000 0.000 0.000 esc ->■ ct in Competence -0.033 0.167 0.120 0.141 -0.212 0.845 0.089 0.366 0.095 0.413 Sellers Reviews Usage ->■ ct in Benevolence 0.006 0.083 0.169 0.029 0.146 0.949 0.289 0.087 0.621 0.303 Sellers Reviews Usage ->■ ct in Competence 0.298 0.222 -0.074 0.305 -0.199 0.006 0.011 0.495 0.000 0.282 notes (1) Bosch, (2) DeLonghi, (3) Krups, (4) Saeco, (5) Siemens. table 12 Significance of Differences Between Groups: pls-mga Non-Parametric Test p-Values Paths (direct effects) Significance of diff. between path coeff. in groups: p-values (from pls-mga test) (1)"(2) (i)"(3) (i)"(4) (i)"(5) (2)"(3) (2)-(4) (2)-(5) (3)-(4) (3)-(5) (4)"(5) Affective Trust ->■ Purchase Intention 0.621 0.281 0.714 0.331 0.102 0.649 0.196 0.944 0.525 0.132 bc ->■ ct in Competence 0.604 0.857 0.664 0.853 0.819 0.544 0.830 0.169 0.618 0.836 co Product Reviews Usage 0.022* 0.544 0.131 0.356 0.779 0.914 0.909 0.292 0.382 0.708 ct in Benevolence ->■ Affective Trust O.752 0.072 0.549 0.257 0.006* 0.244 0.141 0.965 0.567 0.227 ct in Benevolence ->■ ct in Integrity O.349 0.103 0.674 0.915 0.152 0.827 0.967 0.963 0.993 0.864 ct in Competence ->■ Affective Trust 0.227 0.657 0.614 0.283 0.912 0.910 0.522 0.434 0.150 0.168 ct in Competence ->■ ct in Benevolence O.O47* 0.022 0.250 0.174 0.244 0.909 0.658 0.963 0.798 0.318 ct in Integrity ->■ Affective Trust O.312 0.748 0.157 0.769 0.919 0.253 0.858 0.020* 0.631 0.913 Consumer Experience ->■ Product Reviews Usage O.938 0.820 0.964 0.359 0.288 0.596 0.058** 0.788 0.135 0.041* Consumer Experience ->■ Sellers Reviews Usage 0.508 0.248 0.401 0.047 0.215 0.373 0.035* 0.712 0.124 0.048* eco —* Affective Trust O.54I 0.635 0.831 0.480 0.610 0.818 0.458 0.645 0.388 0.263 peef ->■ Consumer Experience O.485 0.220 0.126 0.180 0.219 0.105 0.178 0.527 0.438 0.356 pvc ->■ ct in Competence O.O79** 0.262 0.056** 0.018* 0.799 0.502 0.241 0.168 0.063** 0.213 pvc ->■ ct in Integrity 0.067** 0.183 0.084** 0.441 0.732 0.624 0.909 0.346 0.773 0.886 pvc ->■ Consumer Experience O.29O 0.225 0.125 0.705 0.380 0.256 0.852 0.412 0.874 0.928 pvc ->■ Purchase Intention O.43O 0.933 0.165 0.644 0.960 0.185 0.684 0.012* 0.291 0.830 Product Reviews Usage ->■ Affective Trust 0.580 0.022* 0.115 0.840 0.009* 0.055** 0.820 0.877 0.996 0.982 Product Reviews Usage ->■ Sellers Reviews Usage O.I44 0.540 0.442 0.510 0.875 0.838 0.800 0.402 0.478 0.557 esc ct in Competence O.856 0.777 0.832 0.293 0.382 0.405 0.100** 0.547 0.147 0.118 Sellers Reviews Usage ->■ ct in Benevolence O.734 0.881 0.580 0.793 0.754 0.287 0.648 0.111 0.441 0.779 Sellers Reviews Usage ->■ ct in Competence O.284 0.010* 0.507 0.018* 0.017* 0.769 0.030* 0.998 0.268 0.012* notes Groups numbered as in table 11. *p <0.05, **0.05