Showing a tree to sell the forest: The impact of attribute- and alternative-based information presentation on consumers’ choices

Journal of Economic Psychology xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Journal of Economic Psychology

journal homepage: www.elsevier .com/ locate/ joep

Showing a tree to sell the forest: The impact of attribute- andalternative-based information presentationon consumers’ choices

0167-4870/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.joep.2013.12.001

⇑ Corresponding author. Address: Via capo di Lucca 34, Bologna 40126, Italy. Tel.: +39 051 2098113; fax: +39 051 246411.E-mail addresses: [email protected] (G. Pizzi), [email protected] (D. Scarpi), [email protected] (G.L. Marzocchi).

Please cite this article in press as: Pizzi, G., et al. Showing a tree to sell the forest: The impact of attribute- and alternative-based intion presentation on consumers’ choices. Journal of Economic Psychology (2013), http://dx.doi.org/10.1016/j.joep.2013.12.001

Gabriele Pizzi ⇑, Daniele Scarpi, Gian Luca MarzocchiDepartment of Management, University of Bologna, Italy

a r t i c l e i n f o

Article history:Received 20 December 2012Received in revised form 11 October 2013Accepted 4 December 2013Available online xxxx

JEL classification:M31

PsycINFO classification:390039203940

Keywords:Information presentationConstrual-levelConsumer choiceAttribute-based informationAlternative-based information

a b s t r a c t

Consumers can be provided with information in either an attribute- or an alternative-basedway. We consider the literature on information presentation through the theoretical lensesof the Construal Level Theory. We propose and find that providing product-relatedinformation in an attribute- rather than an alternative-based way shifts choices. The attri-bute-based pattern leads to high construal levels and choices driven by desirability-related,high-level attributes (e.g., design). But when the same information is acquired followingthe alternative-based pattern, it leads to low construal levels and choices driven by feasi-bility-related, low-level attributes (e.g., price). As a consequence, choice shares for productswhose strength lies in convenience and other feasibility-related features are boosted by thepresentation of alternative-based information. Conversely, choice shares for productswhose strength lies in design and other desirability-related features are increased by thepresentation of attribute-based information. We further find that consumers acquiringinformation in an alternative-based way envision consumption much closer in the futurethan those acquiring information in an attribute-based way. Finally, we find that attri-bute-based information leads to more clicking.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Imagine buying a ticket for a flight to your favorite holiday location. Like many people nowadays, you go to the Internetto seek information about ticket prices, flight-cancellation insurance, and other considerations. You might find a websitethat shows the different available flights one after the next, like the pages of a book. Or you might find a website thatshows you the same flights simultaneously but lists them according to certain characteristics (departure time, price, seat-ing, etc.).

More formally, these two options represent an alternative- vs. an attribute-based information presentation. People whoare searching for information can in fact experience two different patterns of information presentation: one operatesattribute by attribute; the other, product by product (Chernev, 2003; Payne, Bettman, & Johnson, 1988). In a nutshell, alter-native-based models postulate that consumers compute an overall utility measure for each option and then make pairwisecomparisons across the available options. Attribute-based models, on the other hand, postulate that consumers weight the

forma-

http://dx.doi.org/10.1016/j.joep.2013.12.001

mailto:[email protected]




http://www.sciencedirect.com/science/journal/01674870

http://www.elsevier.com/locate/joep


2 G. Pizzi et al. / Journal of Economic Psychology xxx (2013) xxx–xxx

value of each attribute across all the available options, focus on the attributes that are not identical between competing op-tions, and then rank the attributes in decreasing order of importance.

Consumers are increasingly searching for information before making purchases (Jepsen, 2007; Moon, 2004). Accordingly,many retailers now provide consumers with the opportunity to obtain information about a product or a service by learningfrom other consumers’ reviews and recommendations in the form of word-of-mouth (Bickart & Schindler, 2001), as well astools for comparing products or services in an assortment, as such comparisons are usually considered a key driver of pur-chase intention. Thus, it is common to find in the practice examples of attribute- or alternative- based information presen-tation. A typical example of attribute-based information presentation is the printed catalog, the structure of which forcesconsumers to analyze products sequentially. Many websites, on the other hand, provide a plethora of tools to help individ-uals gather information attribute by attribute, for instance, tools that group products by price (Jepsen, 2007; Moon, 2004). Topresent information based on attributes rather than alternatives is not a constraint of the communication channel or theproduct category, but a strategic choice on the retailer’s part. For instance, on the Internet, retailers can work either oneway or the other: the Rolex website provides information about watches using an alternative-based scheme, whereas theAmazon.com website uses an attribute-based scheme to present product information on watches. It therefore becomes a rel-evant research question whether such different mechanisms lead to differences in the choices consumers make.

Nonetheless, empirical research on the effect of information-acquisition strategies on consumers’ choices is scarce. In lit-erature that spans nearly half a century, several articles have addressed the issue of information-acquisition strategies the-oretically, debating whence those patterns come, at the expense of understanding where they lead to. To the best of ourknowledge, no previous study has addressed information acquisition as an independent variable that affects choice. We feelthat the time has come to investigate empirically the effects of alternative- vs. attribute-based information acquisition onchoice.

In this research, we aim to demonstrate empirically that the pattern used to present information does affect consumers’choice shares for the products in the assortment. In doing so, we are the first to document that the effect of informationacquisition transcends variations in individuals’ mental representations and can translate to economically relevant behav-iors. In particular, building on the Construal Level Theory (CLT; Trope & Liberman, 2003) we propose and show in two studiesthat the two different information-presentation strategies trigger a reversal of preferences: choice in the attribute-basedcondition is driven by desirability-related features, and by feasibility-related features otherwise. Study 1 is purposely explor-atory and aims to set the boundaries of our investigation. In Study 1 we manipulate the information-presentation strategyand the product attributes, showing a consequent reversal of preferences. We deepen the scope of the analysis in Study 2,ruling out alternative explanations for the results observed in Study 1 and relaxing some of the constraints due to Study 1’sdesign. Specifically, in Study 2 we investigate the consequences for choice when respondents are allowed to click on theavailable information, controlling for actual differences in construal levels, and adding a control group to serve as a baselineto assess the actual size of the effects. Finally, in Study 3 we add a measure of construal level as a dependent measure, toshow unambiguously that CLT is responsible for the effects reported in Study 1 and Study 2.

2. Theoretical background

2.1. Information presentation

The literature on information presentation distinguishes among attribute- and alternative-based strategies (Chernev,2003; Payne et al., 1988), and the great majority of the theoretical models developed on information-acquisition strategiesin the extant literature fall into either the former or the latter category. For instance, the additive model (Tversky, 1969) pos-tulates that consumers compute an overall utility measure associated with each product as the sum of attribute utilities.Analogously, the majority of confirming dimensions model (Russo & Dosher, 1983) assumes that consumers base their choiceson pairwise comparisons across all the available options, whereas the satisficing rule model advances that consumers makealternative-based comparisons with a cut-off point that determines a satisficing choice (Simon, 1955).

On the contrary, the expectancy-value model (Fishbein, 1963) suggests that consumers engage in attribute-based evalu-ations when they choose from a set of options by weighting the value of each attribute across the available options with thesubjective belief strength. Other attribute-based models have been proposed in the literature suggesting that consumers usea lexicographic rule to rank the attributes in decreasing importance order, and compare the products accordingly until a pre-ferred option is identified (Tversky, 1969). Similarly, according to the additive difference and the elimination by aspect models(Tversky, 1969), consumers make attribute-based comparisons by confronting each attribute level across competing prod-ucts in order to identify the best option by eliminating one option at a time from the choice set.

While the single models differ on some aspects, they share the idea that consumers can acquire information either attri-bute by attribute, or product by product. These models also have in common that information acquisition is a dependentvariable, and they overall consider information acquisition as a consequence of how the assortment is arranged. Specifically,the information-acquisition pattern is a consequence of the choice task (Bettman, 1979); of the number of options (Bettman,Luce, & Payne, 1998); and of the degree of preference articulation (Chernev, 2003). However, we posit that the information-acquisition strategy could be an antecedent of choice, and different ways of presenting information to consumers could leadto different choices from the same assortment.

Please cite this article in press as: Pizzi, G., et al. Showing a tree to sell the forest: The impact of attribute- and alternative-based informa-tion presentation on consumers’ choices. Journal of Economic Psychology (2013), http://dx.doi.org/10.1016/j.joep.2013.12.001


G. Pizzi et al. / Journal of Economic Psychology xxx (2013) xxx–xxx 3

Previous literature has shown that there is a reversal of choices between joint (simultaneous) evaluations and separateevaluations when some attributes are more difficult to evaluate than others (Hsee, Loewenstein, Blount, & Bazerman, 1999).The trade-offs underlying the options are differently emphasized under different evaluation modes. We build on these con-siderations and focus on the trade-offs that individuals engage in when evaluating different options. However, we comparethe trade-offs that emerge as a consequence of different ways of information presentation (attribute- vs. alternative-based)rather than as a consequence of different evaluation modes (joint vs. separate). Attribute- vs. alternative-based informationpresentation is a complementary perspective to joint vs. separate evaluation modes: the latter pertains to the difference be-tween acquiring all available information simultaneously rather than stepwise; the former pertains to the kind of the infor-mation provided rather than to sequentiality.

2.2. Construal Level Theory

CLT posits that individuals use concrete, low-level construals to represent psychologically proximal events, and abstract,high-level construals to represent psychologically distal events (Liberman & Trope, 1998; Trope & Liberman, 2000, 2003).

Recent contributions have proved CLT’s capability of accounting for a wide range of economic behaviors, from saving forretirement to opening a bank account (see Leiser, Azar, & Hadar, 2008, for a review). Our analysis is in line with that streamof research and shows that CLT can suggest novel predictions as to how different strategies for information presentation caninfluence consumers’ choices.

Attributes are general features that can be abstracted from objects, whereas products are more tangible and harder toabstract from their context (Liberman & Trope, 2010). This consideration offers a new perspective to our research goal, asit suggests that—accordingly—attribute-based and alternative-based information should lead to different levels of construalof the same object (Borovoi, Liberman, & Trope, 2010), with attribute-based information being represented using abstract,high-level construals, and alternative-based information being represented using concrete, low-level construals. Previous lit-erature has shown that different construal levels lead to buying different products, even within the same choice set (Lynch &Zauberman, 2007). When consumers adopt different construal levels, they weight product-related information differently(Jung-Grant & Tybout, 2008; Malkoc, Zauberman, & Ulu, 2005; Soman, 2004b). Thus, we expect that providing consumerswith alternative- rather than attribute-based information should affect the choice shares among products that differ in desir-ability and feasibility. Specifically, providing alternative-based information should increase the choice share for productsthat are seen as more feasible, whereas attribute-based information should increase the choice share for products thatare seen as more desirable.

Despite these considerations, no previous study has investigated the relationship between information presentation,choice, feasibility, and desirability. In the following, we present three studies: Study 1 and Study 2 address the relationshipbetween the kind of information presentation and the choice share; Study 3 ensures that construal level is responsible forthe effects observed in Study 1 and Study 2.

3. Study 1

3.1. Sample

Participants in this study were undergraduate students enrolled in a consumer behavior course who participated in ex-change for course credit. Students who agreed to participate in the study were randomly assigned a workspace with a com-puter, onto which the questionnaire and the experimental stimuli had been loaded. A total of 83 participants were gathered;their median age was 21 years, and 62% were females.

3.2. Experimental stimuli

Participants were told that their task was to choose a desk for their room from a set of two. The two desks differed on fourattributes (price, delivery plus assembly, design, and size; Soman, 2004a).

The attributes were identified on the basis of a pretest with a pilot sample of 15 different undergraduate students to iden-tify the most relevant attributes driving the choice of a desk by asking them to rank a list of eight attributes in order of impor-tance. The list of eight attributes was the output of both brainstorming and an analysis of the information most commonlyprovided in furniture catalogs. Price, delivery plus assembly, design, and size ranked as the most relevant attributes for desks.Then, similar to Strahilevitz and Myers (1998) and Dhar and Wertenbroch (2000), we conducted a test on a different pilotsample of 25 undergraduate students (median age 22 years, 54% females) with the aim of characterizing those attributesas being either ‘‘feasible’’ or ‘‘desirable.’’ Participants received the list with the attributes along with the definitions of feasi-bility and desirability, just as in Sagristano, Trope, and Liberman (2002, p. 365). Subjects were instructed to classify each of theattributes into one of four categories: feasible, desirable, both feasible and desirable, or neither feasible nor desirable. Priceand delivery plus assembly were placed into the feasible category by respectively 84% (v2(1; N = 25) = 11.5600, p < .001)and 88% of the respondents (v2(2; N = 25) = 33.680, p < .001); design and size emerged to be categorized as desirability-re-lated by respectively 88% (v2(2; N = 25) = 33.680, p < .001) and 84% of the respondents (v2(2; N = 25) = 29.120, p < .001).



Table 1Study 1: Experimental stimuli.

Feature Feasibility-positive desk Desirability-positive desk

Price 79 € 119 €

Delivery and assembly Included for a minimum purchase of 75 € Included for a minimum purchase of 150 €

Design Mass production Designer madeSize Not modifiable Adaptable to room dimensions


In summary, these four attributes ranked as the most relevant, and there was a clear separation between the extent towhich they were perceived to be related to feasibility or to desirability. This procedure ensured that the experimental stimuliused in Study 1 would be perceived as either feasibility- or desirability-positive. Experimental stimuli are summarized inTable 1.

Respondents taking part in the pretests did not take part in the main studies.

3.3. Experimental design

Information presentation was manipulated between subjects, and target-product description was manipulated withinsubjects, resulting in a 3 (attribute- vs. alternative-based vs. control) � 2 (feasibility-positive vs. desirability-positive) mixedexperimental design. We included a control group to provide a baseline for how consumers chose when information acqui-sition was not constrained.

Participants were told that the information on the four product features had been organized in a grid, and that they wouldsee the grid only once and for a limited time. Next, they were randomly assigned either to the attribute- or alternative-basedcondition or to the control group. The product information was organized in a 4-row, 2-column grid; each column describeda product, and each row described an attribute. All respondents saw the same grid: what changed was how they saw it.

Participants assigned to the alternative-based condition saw one product at time (see Appendix A for example screenshot).In other words, they saw only one of the two columns of the grid at time (the other was temporarily blanked); after 5 s theblanked column was made visible and the visible column was blanked1 (the presentation order of the columns was randomized).

Participants assigned to the attribute-based condition saw one attribute at time (see Appendix B for example screenshot).In other words, they saw only one of the four rows of the grid at a time (the others were temporarily blanked); after 5 s1 oneof the blanked rows was made visible and the previously visible row was blanked (the presentation order of the rows wasrandomized). This procedure continued until all four rows had been displayed.

Participants assigned to the control group saw the complete grid (all rows and all columns simultaneously visible) with-out any blank field.

Target-product description was manipulated within subjects. In line with previous research (Kim, Park, & Wyer, 2009),the feasibility-positive product description had two feasibility-related attributes (price; delivery) that were positive andtwo desirability-related attributes (size; design) that were negative. Accordingly, the feasibility-positive product descriptionpresented a low-priced desk whose delivery and assembly were provided by the retailer, but which was poorly designed anduncomfortably small.

Conversely, as in Kim et al. (2009), the desirability-positive product description had two desirability-related attributes(size; design) that were positive and two feasibility-related attributes (price; delivery) that were negative. Accordingly,the desirability-positive product description presented a desk that was finely designed and of a comfortable size, but whichwas expensive and required delivery and assembly by the customer.

3.4. Results

The desirability-positive option was more attractive when the comparison was attribute-based, and the feasibility-posi-tive option was more attractive when the comparison was alternative-based. The choice share for the desirability-positiveproduct was significantly affected by the type of information presentation: choice shares for the available products shiftedsignificantly between the alternative- and the attribute-based condition (v2(2, N = 83) = 13.67, p < .001). Results are summa-rized in Table 2 and illustrated in Fig. 1.

An analysis of variance (ANOVA) was performed on the difference in choice proportions reported in Table 2. The maineffect of information-presentation strategy was significant (F(1,82) = 7.885, p < .001, g2 = .17), suggesting that at least oneof the three experimental treatments affected choices. Post hoc analyses using the Scheffé criterion for significance indicatedthat the proportion of respondents opting for the desirability-positive alternative was higher in the attribute- than in thealternative-based condition (F(1,61) = 16.237, p < .001, g2 = .21). Moreover, the desirability-positive alternative was alsopreferred more in the attribute-based condition than in the control group, where respondents were not driven throughthe information- presentation process (F(1,51) = 5.457, p = .02, g2 = .10). There were no differences, however, between the

1 The 5-s time interval was calibrated empirically on the pilot sample and found to be fully sufficient to evaluate the information provided, be it in rows orcolumns.



Table 2Choice shares for the feasibility-positive and desirability-positive products, by information-presentation strategy.

Information-presentation strategy Choice share

Feasibility-positive product Desirability-positive product

No. % No. %

Attribute-based (n = 33) 8a 24.2a 25a 75.8a

Alternative-based (n = 30) 21a 70.0a 9 30.0Control group (n = 20) 11 48.2 9 45.0

Note. Total sample = 83; Total desirability-positive choosers = 43; total feasibility-positive choosers = 40.a Significant differences with the control group (p < .05).

Fig. 1. Choices for the feasibility-positive and desirability-positive products, by informationpresentation strategy.


alternative-based condition and the control group (F(1,48) = 1.152, p > .05, g2 = .02). That is, alternative-based informationpresentation led to choice shares similar to those in the unconstrained setting. Attribute-based information presentation,on the other hand, led to significant shifts in choice share, and favored the desirability-positive product.

We acknowledge some limitations of Study 1. One is that we forced respondents to evaluate information within a giventime. Although the time span was carefully pretested, the external validity of the experiment would nonetheless be height-ened if we were to allow each subject to determine autonomously the processing time needed. A second threat to the exter-nal validity of Study 1 might be posed by the impossibility of stepping backward, that is, seeing a row/column again after ithad been blanked out. Accordingly, respondents in the following Study 2 will be allowed to see the rows (columns) for aslong and as many times as they want. Study 2 also includes a neutral product description that balanced feasibility- and desir-ability-related features.

4. Study 2

4.1. Sample

Participants were recruited through an invitation emailed to a list of 200 consumers randomly drawn from a panel ownedby the authors. Participants were told that they were being invited to test a new Web-based service for purchasing airplanetickets online, and were asked to take part in a simulated purchase and to make decisions as if they intended to actually buythe ticket. Of these, 133 agreed to participate (response rate 66.5%); their median age was 26, and 56% were females. Theinvitation email included a link to the webpage with the simulation.

4.2. Experimental stimuli

Participants were told that their task was to choose an airplane ticket from a set of three. The two tickets differed on thefour attributes identified from the pretest (price, departure time, seat reservation, and insurance for lost luggage/flightcancellation).

Similar to the procedure followed in Study 1, we used a pretest with a pilot sample of 40 undergraduate students to iden-tify the most relevant attributes driving the choice of an airplane ticket by asking them to rank a list of 14 attributes in orderof importance. The list of 14 attributes was the output of both brainstorming and an analysis of the information most com-monly provided on websites for price comparison. Price, departure time, seat reservation, and insurance for lost luggage/flight cancellation were ranked as the most relevant attributes for airplane tickets. Then, as for the pretest of Study 1, we



Table 3Study 2: Experimental stimuli.

Feature Airline ticket product

Feasibility-positive Neutral Desirability-positive

Price 43 € 180 € 358 €

Departure time 6 a.m. 5 p.m. 10 a.m.Seat reservation No Assigned at check-in Advance online selection permittedInsurance Not included Only for luggage loss Flight cancellation and luggage


instructed 25 undergraduate students (median age 22 years, 54% females) to classify each of the attributes into one offour categories: feasible, desirable, both feasible and desirable, or neither feasible nor desirable. Price and departure timewere placed into the feasibility-related category by respectively 88% (v2(2; N = 25) = 33.680, p < .001) and 84% of therespondents (v2(3; N = 25) = 46.520, p < .001); seat reservation and insurance for lost luggage/flight cancellation werecategorized as desirability-related by respectively 84% (v2(2; N = 25) = 29.120, p < .001) and 92% of the respondents(v2(2; N = 25) = 38.720, p < .001).

In summary, these four attributes ranked as the most relevant, and there was a clear separation between the extent towhich they were perceived to be related to feasibility or to desirability. This procedure ensured that the experimental stimuliused in Study 2 would be perceived as either feasibility- or desirability-positive. Experimental stimuli are summarized inTable 3.


Information presentation was manipulated between subjects, and target-product description was manipulated withinsubjects, resulting in a 2 (attribute- vs. alternative-based) � 3 (feasibility-positive vs. desirability-positive vs. neutral productdescription) mixed experimental design. The neutral description balanced feasibility and desirability, and served to measurea baseline effect (Kim et al., 2009).

Participants were randomly assigned to either the attribute- or alternative-based conditions and redirected to a webpagecontaining a grid. The grid displayed information about tickets for three different flights to a city on the same continent. As inChernev (2003), participants could click on the table to reveal and view a set of information for as long as they wanted beforemaking a final choice. Thus, unlike in Study 1, respondents in Study 2 had no time constraint. However, as in Study 1, respon-dents were allowed to see only one attribute (or one product) at a time, and switched between attributes by clicking onrespectively only the row (or the column) heading of the grid. Specifically, participants in the attribute-based condition couldclick only on the row headings to display information about each attribute; they could not click on column headings. Par-ticipants in the alternative-based condition, on the other hand, could click only on the column headings to display informa-tion about each alternative; they could not click on row headings. Participants in the control group saw the full grid and werethen asked to make their choice.

Target-product description was manipulated within subjects. In line with previous research (Kim et al., 2009), in the fea-sibility-positive description the two feasibility-related attributes (price; departure time) were positive and the two desirabil-ity-related attributes (seat reservation; insurance) were negative. Accordingly, the feasibility-positive product descriptionpresented a low-priced flight with a convenient departure time (i.e., in the morning2), but it offered neither advanced seatreservation nor insurance for lost luggage/flight cancellation.

In the desirability-positive condition, desirability features were positive and feasibility features were negative. Accord-ingly, the desirability-positive product description presented a high-priced flight with an inconvenient departure time(i.e., at night), but it offered advanced seat reservation and insurance for lost luggage/flight cancellation.

The neutral product description combined features; that is, the features were balanced so as to always offer one feasibil-ity- and one desirability-positive attribute, with the other feasibility and desirability attributes being coded negatively.

Finally, participants were asked to indicate when in the future they would have been willing to use the ticket, using a13-point scale (0 = less than 1 month; 12 = more than 1 year). Participants were then thanked and debriefed.

The website automatically recorded how many times a respondent clicked on a row/column. Clearly, as we manipulatedinformation presentation between subjects, whoever clicked on rows could not click on columns, and vice versa.

4.4. Results

4.4.1. Information-presentation strategy and preference sharesThe information-presentation strategy influenced choice shares for both the feasibility-positive and the desirability-

positive alternatives. In line with the results of Study 1, when respondents acquired information following an

2 All attributes and their levels were pretested on a pilot sample extracted from the same population as in the main study. The average time travel to reachthe airport can vary from city to city, so that external validity could be threatened by an early morning departure time. However, based on our pretest, theinternal validity of our experimental manipulation is high.



Table 4Choice shares for the feasibility-positive, desirability-positive, and neutral products, by information-presentation strategy.

Information-presentation strategy Product

Feasibility-positive product Desirability-positive product Neutral product

No. % No. % No. %

Alternative-based (n = 68) 36 53 5 7 27 40Attribute-based (n = 65) 9 14 30 46 26 40

Note. Total sample = 133.

Fig. 2. Choices for the feasibility-positive, desirability-positive, and neutral products, by information-presentation strategy.


alternative-based strategy, they preferred the feasibility-positive product most often (53%) and the desirability-positiveproduct least often (7%); 40% chose the neutral option (v2(2, N = 68) = 22.44, p < .001). However, when the presentationstrategy was attribute-based, the choice shares were reversed: the desirability-positive product was preferred most often(46%), and the feasibility-positive product was preferred least often (14%); 40% chose the neutral option(v2(2, N = 65) = 11.48, p = .003). This evidence is summarized in Table 4 and illustrated in Fig. 2.

4.4.2. Comparison of choice shares in the experimental conditionsUntil now, we have read Table 4 by rows, comparing the choice shares for the different products within the same infor-

mation-presentation strategy. If we instead read Table 4 by columns, we can compare the choice shares for the same productacross the different information-presentation strategies (attribute- vs. alternative-based).

In doing so, we find a reversal of effects: the proportion of respondents choosing the feasibility-positive product was sig-nificantly greater in the alternative- than in the attribute-based condition (F(1,131) = 33.792, p < .001, g2 = .21). However,the proportion of respondents choosing the desirability-positive product was significantly higher in the attribute- than inthe alternative-based condition (F(1,131) = 31.532, p < .001, g2 = .19). There was no difference between the choice sharesfor the neutral option in the attribute- and alternative-based experimental conditions (F(1,131) = .310, p > .05, g2 = .002).This evidence further corroborates the idea that information presentation affects choices.

4.4.3. Ruling out alternative explanationsFirst, one might ask whether respondents in the two experimental conditions were actually adopting two different con-

strual levels. In order to test this, we asked respondents to pinpoint, on a 13-point scale ranging from 0 to 12 (0 = within lessthan 1 month; 12 = in 12 months), when exactly in the future they would have been willing to use the airline ticket. Respon-dents in the alternative-based condition specified a departure significantly nearer in the future (M = 1.45, SD = 1.05) thanrespondents in the attribute-based condition (M = 4.28, SD = 2.98; F(1,130) = 63.68, p < .001, g2 = .29). This result is in linewith the predictions of CLT, according to which consumers in a low-construal-level mindset collocate events nearer inthe future than consumers in a high-construal-level mindset (Trope & Liberman, 2000).

Second, respondents clicked on the headings of the grid more often when the information presentation was attribute-based (M = 10.02, SD = 6.97) than when it was alternative-based (M = 5.98, SD = 4.58; F(1,131) = 16.35, p < .001, g2 = .11).One might argue that the differences in choice shares can be ascribed to the number of clicks rather than to the informa-tion-presentation strategy, in other words, that respondents had more trouble choosing a preferred alternative in the attri-bute- than in the alternative-based condition. Thus, according to this alternative explanation, the increasing levels ofdifficulty would have affected choice shares. However, there were no differences in choice shares between those respondents



Presentation mode

Construal level

Choice

Flight

4.88* .57*

-1.61* (-.29)

Presentation mode

Construal level

Choice

Desk

2.27* .25*

-1.45* (-1.17)

Fig. 3. Testing the mediation of construal level between information presentation mode and choice. Note. The regression coefficient between presentationmode and choice controlling for construal level is in parentheses. *p < .05.


who made a number of clicks above vs. below the median number of clicks (median = 7; Wilks = .979, F(2,130) = 1.399,p =.250, g2

p ¼ :02).

5. Study 33

The aim of Study 3 is to rule out the possibility that differences in choice shares observed in Study 1 and Study 2 are dueto mechanisms other than construal level. Accordingly, we used the same experimental stimuli as in Study 1 and Study 2 andmeasured construal level as the dependent variable.

5.1. Sample

Participants were recruited through an invitation emailed to a list of 200 consumers randomly drawn from the samepanel used in Study 2, excluding those who had already received the invitation to participate in previous study. Participantswere told that they were being invited to test a new Web-based service for product comparison, and were asked to engage inproduct comparison. Of these, 130 agreed to participate (response rate 65%); their median age was 26, and 64% were females.The invitation email included a link to the webpage with the simulation.


Participants were told that their task was to compare products. Participants saw tables containing information abouteither the desks or the airplane tickets used in the previous studies. Which product would appear was randomized by thesoftware.

Participants were randomly assigned to either the attribute- or alternative-based condition and redirected to the web-page containing the grids used as experimental stimuli in Study 1 (desks) and Study 2 (airplane tickets). Participants firstsaw the table with the information about the products displayed either in an attribute- or alternative-based way, as in Study2. Then, in order to measure construal level as a dependent variable, we administered an adapted version of the BehavioralIdentification Form (BIF; Vallacher & Wegner, 1989) as in Liberman and Trope (1998). Thus, participants read a list of 19neutral actions, each followed by two alternative restatements corresponding to high and low levels of action identification(e.g., neutral phrase: ‘‘paying the rent’’; low-level restatement: ‘‘to give money to the landlord’’; high-level restatement: ‘‘tomaintain a place to live’’). To avoid sequence effects, the presentation order of the 19 items was randomized by the software.Furthermore, the high- and low-level descriptions of each action were randomized, so that for some actions, respondentsread the high-level restatement before reading the low-level description, and vice versa.

3 We thank the anonymous reviewer for suggesting that we add this study to infer unambiguously that construal level is responsible for the effects of Study 1and Study 2.




5.3. Results

In line with Liberman and Trope (1998), we computed an overall construal-level score for each respondent by summingthe scores on the 19 items, which were recoded as 0 if a respondent chose the low-level restatement and as 1 if (s)he chosethe high-level restatement. As such, each respondent’s overall construal-level score ranges from 0 to 19, and the higher thescore, the higher the respondent’s overall construal level.

A one-way ANOVA with construal level as dependent variable and information presentation as independent variableyielded a significant difference in the mean overall construal-level score between those who acquired information on theflight tickets in the alternative-based condition (M = 10.19, SD = 5.17) and those who acquired it in the attribute-based con-dition (M = 14.61, SD = 3.53; F(1,42) = 11.13, p = .002, g2 = .21).

A one-way ANOVA with construal level as dependent variable and information presentation as independent variableyielded a significant difference in the mean overall construal-level score between those who acquired information on thedesks in the alternative-based condition (M = 11.59, SD = 4.43) and those who acquired it in the attribute-based condition(M = 13.86, SD = 3.15; F(1,53) = 4.80, p = .03, g2 = .08).

That is, respondents displayed higher levels of construal when acquiring information in an attribute- rather than an alter-native-based way.

In line with the methodology proposed by Iacobucci (2012) for mediation analysis with categorical variables, we testedthe potential mediating role of construal level on the relationship between information presentation mode and choice. Theresults are summarized in Fig. 3 below and support full mediation (Baron and Kenny, 2012).

The results of Study 3 offer straightforward evidence supporting on the one side the effects of presentation mode on con-strual levels, on the other side that construal level moderates the relationship between presentation mode and choice ob-served in the previous studies.

6. Discussion and conclusion

We interpreted the literature on information presentation through the theoretical lenses of CLT, which, since its devel-opment, has been found to account for a number of economically relevant behaviors, whether in terms of intertemporalchoices, predictions on the choices of others, advice-giving, or saving for retirement (Leiser et al., 2008). We intuited thatdifferent methods of presenting information would foster different decision-making processes, with the alternative-basedcondition favoring feasibility-related features and the attribute-based condition favoring desirability-related features.

The main result in this analysis is that providing product-related information in an attribute- rather than alternative-based way shifts consumers’ choices. More specifically, we found a reversal of preferences that was triggered by the infor-mation-presentation strategy: as we expected, choice in the attribute-based condition was driven by desirability-related,high-level attributes (such as comfort). When the same information was presented following an alternative-based pattern,choice was driven by feasibility-related, low-level attributes (such as price).

In addition, we proposed and found that the different methods for presenting information would lead consumers to envi-sion a different timing for consumption. Individuals acquiring information in an attribute-based way projected actual con-sumption in a distant future, whereas those acquiring information in an alternative-based way envisioned consumptionmuch sooner.

We conducted a series of three studies confronting subjects with different patterns of information presentation. InStudy 1 we manipulated the information-presentation strategy and the product attributes, and found a reversal of pref-erences between feasibility- and desirability-related options. In Study 2, by adding a neutral product description to serveas baseline, we were able to shed further light on the actual size of the effects we found. Taken together, the control groupand the neutral option showed that the effects of attribute- vs. alternative-based information presentation occurred aslong as the products were presented as actually different with regard to feasibility and desirability. Once differences infeasibility and desirability were smoothed and balanced, different strategies for information presentation no longer ex-erted differential impacts on consumers’ choices. To assess whether construal level is responsible for the mechanism thatunderlies the main effects explored by this paper, in Study 3 we added a measure of construal level as dependent mea-sure; we also show that information presentation not only is responsible for shifts in choice shares but also affects indi-viduals’ construal levels. That is, different information-acquisition strategies lead individuals to adopt different construallevels: low construal levels for alternative-based information and high construal levels for attribute-based information.Strengthening the link among information acquisition, CLT, and choice allows us to infer that different information-pre-sentation strategies produce different construal levels that in turn, as posited by CLT, affect choice shares for desirablevs. feasible alternatives, thus providing an unambiguous explanation for the effects of information acquisition on choiceobserved in Study 1 and Study 2.

Our evidence suggests that if a retailer aims at boosting sales for a product that is highly desirable but of low feasibility,she or he should present and compare the available alternatives by following an attribute-based pattern. On the other hand,an alternative-based pattern would be more adequate for boosting sales of products that are highly feasible but of low desir-ability. Accordingly, online retailers could structure their website to be only attribute-based or only alternative-based, orthey could allow consumers to switch between the two styles of information presentation with a simple click.




This study is not meant to be conclusive; nevertheless, we believe our results can be relevant for practitioners and canstimulate future research that could include a broader set of information- and decision-processing contexts. For the aimsof our analysis, we forced respondents to follow either an alternative- or an attribute-based pattern, similar to what manyretailers actually do on their websites. Future research could relax this experimental condition and observe the natural wayconsumers acquire alternative- and attribute-based information, under the influence of an extended set of environmentalcues such as page orientation and attribute numbering/bulleting. Finally, unrelated to our main research question, we havenoticed that people clicked more often when information was attribute-based. On the one hand, this finding can be of man-agerial relevance for click-and-pay in-page advertisers. On the other hand, it offers a different perspective than the elimina-tion-by-aspect model (EBA hereafter), which predicts that elimination by attribute leads to less browsing. While weacknowledge that only future research can shed light on the differences between these two analytical frameworks, a wordof caution is needed here. The two frameworks do not completely overlap. For instance, EBA implies that individuals see allavailable information, whereas in our case no scenario displayed the full information matrix (individuals could never see allrows and all columns simultaneously). Furthermore, EBA distinguishes between unique and shared attributes, where sharedattributes are discarded (Tversky, 1972, p. 468). Instead, in our research, all attributes are shared by the choice options anddiffer in their levels. This is to say, the unique part of the information in our case is the level of each attribute, not the attri-bute itself. Future research could extend our findings and address the effects of information acquisition on browsing behav-ior, also considering the case of unique attributes.

Appendix A. Study 1: alternative-based information presentation

Pti

lease cite this article in press as: Pizzi, G., et al. Shon presentation on consumers’ choices. Journal of

Desk Delta

owing a tree to sell the forest: The impact of attribute- and alternativEconomic Psychology (2013), http://dx.doi.org/10.1016/j.joep.2013.1

Desk Gamma

Price
€ 79 Delivery and assembling Included for a minimum purchase of 75 €
Design
Mass production Size Not modifiable
Appendix B. Study 1: attribute-based information presentation

Desk Delta
Desk Gamma
Price
€ 79 € 119 Delivery and assembling Design Size
References

Baron, R. M., & Kenny, D. A. (2012). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statisticalconsiderations. Journal of personality and social psychology, 51(6), 1173–1182.

Bettman, J. R. (1979). An information processing theory of consumer choice. Reading, MA: Addison-Wesley.Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25, 187–217.Bickart, B., & Schindler, R. M. (2001). Internet forums as influential sources of consumer information. Journal of Interactive Marketing, 15, 31–40.Borovoi, L., Liberman, L., & Trope, Y. (2010). The effect of attractive but unavailable alternatives on the attractiveness of near and distant menus. Judgment

and Decision Making, 5, 102–109.Chernev, A. (2003). Product assortment and individual decision processes. Journal of Personality and Social Psychology, 85, 151–162.Dhar, R., & Wertenbroch, K. (2000). Consumer choice between hedonic and utilitarian goods. Journal of Marketing Research, 37, 60–71.Fishbein, M. (1963). An investigation of relationships between beliefs about an object and the attitude toward that object. Human Relations, 16, 233–240.Hsee, C., Loewenstein, G., Blount, S., & Bazerman, M. (1999). Preference reversals between joint and separate evaluations of options: A review and

theoretical analysis. Psychological Bulletin, 125(5), 06–14.Iacobucci, D. (2012). Mediation analysis and categorical variables: The final frontier. Journal of Consumer Psychology, 22(4), 582–594.Jepsen, A. L. (2007). Factors affecting consumer use of the Internet for information search. Journal of Interactive Marketing, 21, 21–34.Jung-Grant, S., & Tybout, A. M. (2008). The effect of temporal frame on information considered in new product evaluation: The role of uncertainty. Journal of

Consumer Research, 34, 897–913.Kim, Y.-J., Park, J., & Wyer, R. (2009). Effects of temporal distance and memory on consumer judgments. Journal of Consumer Research, 36, 634–645.Leiser, D., Azar, O. H., & Hadar, L. (2008). Psychological construal of economic behavior. Journal of Economic Psychology, 29, 762–776.Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal

theory. Journal of Personality and Social Psychology, 75, 5–18.Liberman, N., & Trope, Y. (2010). Construal-level theory of psychological distance. Psychological Review, 117, 440–463.Lynch, J. G., & Zauberman, G. (2007). Construing consumer decision making. Journal of Consumer Psychology, 17, 107–112.

e-based informa-2.001

http://refhub.elsevier.com/S0167-4870(13)00137-2/h9005
























Malkoc, S. A., Zauberman, G., & Ulu, C. (2005). Consuming now or later? The interactive effect of timing and attribute alignability. Psychological Science, 16,411–417.

Moon, B. J. (2004). Consumer adoption of the Internet as an information search and product purchase channel: Some research hypotheses. InternationalJournal of Internet Marketing and Advertising, 1, 104–118.

Payne, J. W., Bettman, J. R., & Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory andCognition, 14, 534–552.

Russo, J. E., & Dosher, B. A. (1983). Strategies for multiattribute binary choice. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9,676–696.

Sagristano, M. D., Trope, Y., & Liberman, N. (2002). Time-dependent gambling: Odds now, money later. Journal of Experimental Psychology: General, 131,364–376.

Simon, H. A. (1955). A behavioral model of rational choice. Quarterly Journal of Economics, 69, 99–118.Soman, D. (2004a). The effect of time delay on multi-attribute choice. Journal of Economic Psychology, 25, 153–175.Soman, D. (2004b). Framing, loss aversion, and mental accounting. In D. J. Koehler & N. Harvey (Eds.), Handbook of judgment and decision making

(pp. 379–398). Malden, MA: Blackwell.Strahilevitz, M., & Myers, J. G. (1998). Donations to charity as purchase incentives: How well they work may depend on what you are trying to sell. Journal of

Consumer Research, 24, 434–446.Trope, Y., & Liberman, N. (2000). Temporal construal and time-dependent changes in preference. Journal of Personality and Social Psychology, 79, 876–889.Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological Review, 110, 403–421.Tversky, A. (1969). Intransitivity of preferences. Psychological Review, 76, 31–48.Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79, 281–299.Vallacher, R. R., & Wegner, D. M. (1989). Levels of personal agency: Individual variation in action identification. Journal of Personality and Social Psychology,

57, 660–671.

























Documents

Showing a tree to sell the forest: The impact of attribute- and alternative-based information presentation on consumers’ choices