The effects of input modality and story-based knowledge on users' game experience

Computers in Human Behavior 68 (2017) 180e189

Contents lists available at ScienceDirect

Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh

Full length article

The effects of input modality and story-based knowledge on users' game experience Dong-Hee Shin a, *, 1, Kyung-mi Chung b, 1 a b

School of Media and Communication, #1411, bldg. 303, Heukseok-ro 84, Dongjak-gu, Seoul, 06974, South Korea Department of Interaction Science, Sungkyunkwan University, Seoul, South Korea

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 September 2016 Received in revised form 18 November 2016 Accepted 19 November 2016

This study examines how input modality (finger touch vs. stylus) and story-based knowledge (eWOM only vs. eWOM þ visual game-metaphor-based knowledge) affect assessments of the dynamics of gameplay experience over time: ergonomic and hedonic quality, judgment of appeal, and perceived enjoyment. In the immediate evaluation, we asked two groups with different story knowledge to complete close-ended questionnaires after playing a paid mobile game with different input modalities. After a four-week delay, eighty participants returned and completed the same close-ended and new open-ended questionnaires to measure their impressions of the game and their remembered experiences across the input methods for game interaction. The results show significant main and interaction effects only for the input modality and time in the evaluation of the complete gaming experience, but the results of the univariate analysis show significant differences across story-knowledge groups in the assessment of hedonic quality. Although it was difficult to draw a clear conclusion about which input modality was better, the stylus has good potential as an alternative game controller. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Paid game app eWOM Story-based knowledge Input modality Time distance Gameplay experience

When people have a wide variety of choices for both free and paid mobile game apps, how do they judge the quality of the apps, and which information sources do they consider? According to the 2015 Essential Facts report by the Entertainment Software Association (ESA), graphics are no longer the top factor that influences customers' buying intentions. Instead, the report shows, “Interesting story/premise” is the top factor, accounting for 22% of the overall influence, followed by “price” (15%), “word-of-mouth” (11%), and “quality of graphics” (7%). As low-cost and powerful distribution channels for third-party apps, Google Play and the Apple App Store have introduced countless mobile apps to their users and have allowed those users to publicly rate and review the apps they install, thereby empowering them to spontaneously serve as both evangelists and critics. Compared to a free version of game apps, most people find it difficult to determine whether to purchase the paid game app on the basis of only electronic wordof-mouth (eWOM) reviews (Cheung & Thadani, 2012). eWOM communications can function as both an input to

* Corresponding author. E-mail address: [email protected] (D.-H. Shin). 1 Both authors contributed equally to this study and thus both authors are the co-first authors. http://dx.doi.org/10.1016/j.chb.2016.11.030 0747-5632/© 2016 Elsevier Ltd. All rights reserved.

consumer decision-making (Chen, Nguyen, Klaus, & Wu, 2015) and an outcome of the purchase process (See-To & Ho, 2014). eWOM plays a key role in creating a virtuous cycle of both mechanisms in that the product or service quality affects customer satisfaction (Erkan & Evans, 2016), and customer satisfaction in turn influences eWOM (See-To & Ho, 2014). Given the highly competitive landscape of mobile game apps, it is important to understand exactly what users are saying in their reviews, and that understanding yields important insights into new business models. Compared to eWOM, indirect information closely linked with the qualities of goods such as paid game apps could motivate potential users to acquire hands-on experience. An advertising effort for experience goods might not be easily observable, or viewers might not be aware of the indirect messages that advertising intends to convey. Despite those limitations of indirect information, people exposed to any type of indirect game-related information are more likely to engage with the game than those who are not so exposed. In contrast, eWOM offers relatively direct information about products or services. Although previous studies have investigated the effects of the different characteristics of eWOM, such as valence (positivity and negativity: Sweeney, Soutar, & Mazzarol, 2014; Tsao & Hsieh, 2012), presentation form (textual content and numerical star ratings: Mudambi & Schuff, 2010; Vasa, Hoon,

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

Mouzakis, & Noguchi, 2012), type (user reviews and third-party reviews: Chen & Xie, 2008), and readability (length and number of reviews: Mudambi & Schuff, 2010; Vasa et al., 2012), little interest has been shown in the associations among eWOM, different game-related story knowledge, and other gameplay-related factors that could lead users to perceive certain quality aspects and dynamics of the gaming experience. In terms of the complete user experience of story-based mobile games with a single ending, some players stop playing the game but leave it behind on their devices, and others completely remove it from their devices (Ahn & Shin, 2015). If the game is not updated anymore, most players eventually choose the latter option. Therefore, it is necessary to focus on changes in the overall evaluations of game quality and player experience for hedonic game apps whose use rarely continues in everyday life after completing the gameplay. However, the measurement of gaming experience dynamics for such hedonic game apps has received relatively little attention from researchers, compared with studies about consistently used utilitarian software (Diefenbach & Hassenzahl, 2011) and electronic gadgets such as mobile phones (Shin & Shin, 2011) and pointing devices (Karapanos, Martens, & Hassenzahl, 2012). To understand the impression evaluation process in game players' minds, we consider a time-based data analysis of the gaming experience and elicit experience narratives related to two different input modalities. WOM is a form of person-to-person communication in which one person tells another a story as a structured, coherent retelling of an experience or as a fictional account of an experience (Erkan & Evans, 2016). We analyze how this characteristic of WOM as a form of storytelling manifests in participants' responses in detail in our qualitative data analysis. We are particularly concerned with the question of how the impression of a mobile game that participants played was formed and stored in their memories a month after the initial gameplay. Furthermore, we are interested in how differently participants evaluated and remembered the gaming experiences they had with different input modalities (finger touch vs. stylus). In this study, we extend the concept of input modality to modality interactivity by providing an interactive motor dimension to the story experience while playing a game. We expect that user engagement with the interaction tools will reflect their engagement with the game content (Shin, Choi, Kim, & Lee, 2016). Thus, with this study we contribute to the literature on computer games that focuses on user experience from the viewpoint of aesthetics, game content, and underling game stories by additionally addressing the issue of input modality in relation to the players' engagement. To keep track of both internal stories (created from their firsthand experiences) and external stories (delivered by outside sources) in their memories (See-To & Ho, 2014), we asked people to tell stories of their experience with respect to the two major concerns in a delayed evaluation. Of all possible concerns, the most important point here is that two external stories, eWOM and the narrative in the game metaphor, could be reconstructed as the participants' own internal stories as soon as they started playing the game. Therefore, the following research questions have guided our study. RQ1: What are the effects of input modality and story-based knowledge on the assessment of gameplay experience over time? RQ2: How do different game users evaluate and remember their gaming experience with different input modalities (finger touch vs. stylus)? Through those questions, our eventual goal is to explore the effects of narrative game-metaphor-based knowledge in eWOM communications and the effects of the different input modalities on players' overall evaluations of their experiences: the subjective

181

perceptions of the ergonomic quality (EQ) and hedonic quality (HQ) of a mobile game, the judgment of its appeal, and their enjoyment of it. Our secondary goal in this study is to examine the immediate (immediately after initial gameplay) and delayed (a month later) effects of two independent variables on the same overall evaluation of their experience and their final impression of the game itself, along with their remembered experiences between the input methods for game interaction. Taken together, the findings of the present study provide marketers, developers, and designers with both theoretical and practical implications for designing, promoting, evaluating, and improving mobile game apps, thereby effectively managing the complete user experience. Furthermore, this study guides readers in interpreting the quantitative findings of a controlled experiment using links with qualitative findings from open-ended responses. 1. Literature review 1.1. Electronic word-of-mouth An extremely negative attribute can be perceived as highly informative and diagnostic when many positive attributes are  zquez-Casielles, Sua rez-Alvarez, presented (Va & Río-Lanza, 2013). According to the findings of Tsao and Hsieh (2012), positive WOM (PWOM) is more frequent than negative WOM (NWOM); PWOM appears roughly three times as often as NWOM. From this viewpoint, NWOM tends to be more informative and diagnostic than zquez-Casielles et al., 2013), thus weighting the negative PWOM (Va aspects of products or services more heavily than the positive aspects. The rare negative information is more likely to alert customers to facts they do not know than the relatively common positive information (Tsao & Hsieh, 2012). Negative attribute information is weighted heavily in product judgment contexts, providing direct support for the effects of rarity and extremity. However, most user-generated reviews on products or services are neither purely positive nor purely negative; arguments about the valence and extremity of eWOM have been disputed thus far. In an effort to reflect those arguments, we take both characteristics of eWOM into account, controlling for and manipulating them in our experiment. 1.2. Ergonomic/hedonic quality, judgment of appeal, and perceived enjoyment Irrespective of the sufficiency of available product information, the entire process of experiencing a product with certain quality aspects and perceiving different consequences is repeated in everyday life (Diefenbach & Hassenzahl, 2011). Both the intended and perceived quality of a product have two different quality aspects: (1) EQ and (2) HQ. EQ describes a task-oriented quality aspect addressing the underlying human need for security and control, and HQ refers to a non-task-oriented quality aspect (i.e., originality, innovativeness, or interestingness) that addresses the human needs for novelty induced by visual/sound design or interaction techniques. Both quality aspects are independently perceived by users and appear to contribute equally to the overall judgment of a product's appeal (Diefenbach & Hassenzahl, 2011; Shin, 2015). Determining whether a software system can be regarded as appealing relies heavily on users' subjective perceptions of its EQ and HQ (Diefenbach & Hassenzahl, 2011). A cognitive appraisal, with its distinction between perception and evaluation, has behavioral (i.e., increased/decreased usage frequency and learning time) and emotional consequences (i.e., perceived fun/enjoyment and (dis)satisfaction) (Shin, 2016). In the present study, we offer game designers and developers a

182

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

better understanding of the role that the subjectively perceived EQ and HQ of a mobile game app plays in how users judge appeal and evaluate emotional outcomes. In fact, measuring behavioral outcomes, such as level completion time and the number of completed levels, is not the focus of this study because performance can be strongly influenced by individual differences in cognitive abilities and information processing. To identify additional key factors that can stimulate the perception of HQ over time, we consider story knowledge, input modality, and time. Diefenbach and Hassenzahl (2011) measured the EQ, HQ, and APPEAL values both pre- and post-usage; however, they did not consider “enjoyment” as an emotional outcome of the complete gaming experience. Furthermore, that previous study did not consider how the participants' perceptions and evaluations post-usage persisted or changed after a four-week delay. 2. Method For this study, we evaluated games using a few criteria: fast responses, immediate interaction, time limit in gameplay, and exciting gameplay. Based on those criteria, we chose Monument Valley, a paid game available for download. The game does not require immediate interaction and has no time limit, but a setting function enables players to play on a fast track with direct interaction. Also, for the purpose of our experiment, we enforced a timelimit. We divided the game category into stylus pen and finger because for some games it is inefficient to use just the stylus or just a finger. We implemented the experiment using a smartphone with the 4-inch display set to a 640  1136 native resolution at 326 ppi (pixel per inch), which is too small to easily play Monument Valley via finger-touch. 2.1. Procedure For the immediate evaluation, we conducted the experiment in 50 min using three sessions: a welcoming and screening session (5 min), a priming session (5e10 min; eWOM-only for 5 min or eWOM-Visual storytelling for 10 min), and a main session (35 min), including a 2-min break between the two input modality conditions and the questionnaire completion time. After being seated, participants received consent forms and written instructions. During the screening session, they completed an online version of the Edinburgh Handedness Inventory and a paper-based screening questionnaire with five items and demographics. In accordance with the responses to the items, participants were assigned to one of the two different storyknowledge groups, considering the presentation order of the input modalities as well as the gender and age distribution. In the priming session, one between-subjects group was exposed to a 14-page slideshow of real app reviews from the App Store, and the other between-subjects group was exposed to a 16page slideshow of Escher's works and a famous saying, followed by the slideshow of the app reviews. The experimenter told both groups that all the stimuli were an edited version of real artworks and app reviews. Both groups were instructed to view the slide contents with no definite time limit. After finishing each slideshow, the experimenter asked participants what the artist was trying to convey through his works and message, how a wide variety of app reviews could be summarized, or both across conditions to maintain their attention to the visual stimuli. In the main session, when the experimenter gave them a start signal, participants were asked to start and play the game freely until the experimenter gave them the stop signal, thus enabling them to be immersed in the gameplay without any disturbance. For the game task, in a home-like laboratory setting, participants were

asked to play the game as usual in a comfortable position to prevent their experience from being influenced by controlled settings. While participants in previous research were asked to take as long as necessary to complete a specific level (Zaman, Natapov, & Teather, 2010), those in this experiment were allowed to play the game as far as they could within 12-min and 15-min time limits for two trials. In each trial, the experimenter told participants to use one of the two input modalities, the index finger of their preferred hand or a stylus. Because the level of task difficulty increases steadily in the gameplay, the time limit for the second trial was longer than that for the first trial. By counterbalancing the order of presentation of the interaction tools, participants could concentrate more on evaluating their experiences across the different input methods instead of across task difficulty. After completing as many stages as possible via the different input methods, the participants filled out each online questionnaire. As soon as they had completed the first questionnaire for one input method condition, participants were given a 2-min break to relax. If the next input method was the stylus, the experimenter made sure the stylus worked and then explained how to use it during the break. After the break, participants resumed the game from the stage where they had left off. If the stylus did not work, the participants were told to raise their hand; they received extra time to make up for that used to solve the problem and resume the gameplay. Following confirmation that all the data had been recorded on the answer sheets, the experimenter told the participants to return after four weeks. Participants were not debriefed to prevent them from discerning the actual purpose of the research. In the delayed evaluation, participants filled out a set of paper-based questionnaires about their final impressions of the game and their gaming experience across different input modalities; they were not asked to recall nor were they shown the contents of the visual stimuli or their initial responses to the same questionnaires. 2.2. Participants For the immediate time condition, we recruited 80 undergraduate and graduate students (32 males and 48 females) from a large comprehensive university in Seoul, South Korea. After four weeks, 78 of those participants (31 males and 47 females), ranging in age from 17 to 32 years (Mtotal ¼ 22.19, SDtotal ¼ 2.70), returned for the delayed time condition. The data from two participants assigned to the eWOM-Visual storytelling condition were discarded from the analysis. The participants were equally assigned in terms of gender and age to the two different story-knowledge groups: eWOM-only (16 males and 24 females; aged 17e32 years; MeWOM ¼ 22.20, SDeWOM ¼ 3.15) and eWOM-Visual storytelling (15 males and 23 females; aged 19e27 years; MeWOMeVisual storytelling ¼ 22.18, SDeWOMeVisual storytelling ¼ 2.17). We used an independent-samples t-test to determine whether the gender and age assignment to the two story-knowledge groups was successful, and it was: two-tailed t (76) ¼ 0.026, p ¼ 0.980. To screen and assign volunteers to one of the two storyknowledge groups, we asked the following screening questions: (1) Do you have either a capacitive touchscreen-based smartphone or a tablet PC? (2) Have you ever heard of the Dutch artist M.C. Escher (Maurits Cornelis Escher) or seen his artwork before? (3) Have you ever read the reviews on the game Monument Valley in the Apple App Store? (4) Have you ever purchased or played the game Monument Valley? We accepted participants who had used capacitive touchscreen-based devices and had no prior experience of reading the app reviews or playing the paid game. On the basis of their responses to question 2, both those who might have heard of the artist or seen his works and those who knew the artist for

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

certain were assigned to the eWOM-Visual storytelling group. 2.3. Experimental design To explore how the relationship between the two storyknowledge conditions and the two direct touch methods for a touchscreen-based game individually and interactively influenced the assessment of the gaming experience over time, we constructed our experiment as a 2 (Story knowledge: eWOM only vs. eWOMVisual storytelling)  2 (Input modality: Fingertip vs. Stylus)  2 (Time: Immediate vs. Delayed) mixed randomized-repeated design, with one randomized-group independent variable (IV; story knowledge) and two repeated-measures IVs (input modality and time). In the immediate evaluation, the dependent variables (DVs) were the participants' overall evaluations of their experience with the game: EQ/HQ, judgment of appeal (APPEAL), and perceived enjoyment (PE). In the delayed evaluation, we added the participants' final judgments of the game itself and their remembered experiences of the game interaction across the input methods to the DVs. To control for order effects, we used complete counterbalancing in presenting the two input modalities in all possible orders for the different story-knowledge groups. Some participants played the game via finger touch first, whereas the others did so via the stylus first. In addition, we controlled for participants' gameplay expertise in this experiment. All participants were casual players who play mobile games for 1 h per day or less to try to concentrate on their studies or work, not those who dislike playing a game itself or those who prefer certain genres of mobile games. The experiment was implemented using a smartphone with a 4inch display set to a 640  1136 native resolution at 326 ppi; the screen was not big enough to easily play Monument Valley via finger touch. To compare the players' gaming experiences across two direct touch inputs for the same game, we chose the index finger of their preferred hand as a baseline and the “Adonit Jot Script 1.” The Adonit Jot Script 1 is a 12-mm-wide and 142-mm-long capacitive stylus for iOS with a fine-tuned 1.9-mm-diameter metal tip; the tip is 70% smaller than the 6.0-mm-diameter rubber tip of a competitor stylus. Jot Script is expected to provide its users with the same experience of using touchscreen-based devices as with their index finger, such as tapping, tapping and holding, flicking, dragging, and rotating. 2.4. Visual stimuli 2.4.1. Word-of-mouth The eWOM stimulus was intended to simulate word-of-mouth information for participants. We selected fourteen slides because we determined that number to be the most effective and efficient. Expert interviews using Delphi-methods revealed that more than ten slides would be effective for the stimulus. Pilot tests of twelve respondents show that 14 and 16 were appropriate numbers of slides. Considering time and budget, we decided to use 14 slides (including the front and back covers). Therefore, we built a 14-page slideshow about Monument Valley, using Microsoft PowerPoint (MS PPT). Participants were allowed to go to the next page by clicking on the mouse. To motivate them to be not an onlooker but a true observer, we designed the eWOM visual stimuli to mimic users' real-life experiences of searching for apps to download. We captured screenshots of real app reviews in the Apple App Store, using some of them selectively to control for the unexpected effects of valence, extremity, and sidedness in eWOM. Because positive eWOM is three times as common as negative eWOM, we manipulated the ratio of positive eWOM to negative eWOM in our slideshow to be approximately 3:1.

183

2.4.2. Visual storytelling We also created a 16-page slideshow in MS PPT showing Monument Valley as an interactive version of M.C. Escher prints. Making a game can be the same as turning one's imagination into reality, and therefore, players might not differentiate between their imagination and the real but impossible game world. To reflect on those points, we chose Escher images and one of his famous sayings that seemed to be closely linked with the elements of Monument Valley. To help the participants focus on the details of figures and structures, 13 of the 16 slides were black and white images; only 3 of them were colored images to strengthen the impression of the use of color. 2.5. Dependent measures All items for the dependent measures were translated into Korean to present the items to the Korean native speakers and were then translated back into English for this paper. We used all 27 items about EQ/HQ, APPEAL, and PE repeatedly to measure various aspects of the complete gaming experience under both time conditions: immediate and delayed assessments. 2.5.1. Perceptions of ergonomic and hedonic quality To address different perceptions of Monument Valley's quality aspects, we measured both EQ and HQ for the mobile game using 15 pairs of adjectives on the 7-point semantic differential scales developed by Diefenbach and Hassenzahl (2011): (1) 8 pairs of EQ items (comprehensibleeincomprehensible, supportingeobstructing, simpleecomplex, predictableeunpredictable, cleareconfusing, trustworthyeshady, controllableeuncontrollable, and familiarestrange) and (2) 7 pairs of HQ items (boringeinteresting, cheapecostly, dulleexciting, standardeexclusive, nondescripteimpressive, ordinaryeoriginal, and conservativeeinnovative). Both the EQ and the HQ scales were presented in a random order. 2.5.2. Judgment of appeal To determine whether users considered Monument Valley appealing, we used 8 pairs of APPEAL items constructed by Diefenbach and Hassenzahl (2011) using a 7-point semantic differential scale: unpleasantepleasant, badegood, unaestheticeaesthetic, rejectingeinviting, unattractiveeattractive, unsympatheticesympathetic, discouragingemotivating, and undesirableedesirable. 2.5.3. Perceived enjoyment As a part of the basic nature of games (Shin & Ahn, 2013), we adopted our PE measure from Liu and Li (2011) and used a 7-point Likert scale anchored from strongly disagree to strongly agree: (1) “I think it is fun to play mobile games,” (2) “I think the process of playing a mobile game would be unpleasant,” (reverse-coded) (3) “I think playing a mobile game would bring me pleasure,” and (4) “I enjoy playing mobile games.” 2.5.4. Final impressions of the game and the remembered gaming experience across different input modalities We constructed the measures for the four-week delayed impressions of the game and the gaming experience across different direct input methods with a “required” open-ended question and an “optional” open-ended question: (1) Please write down more than three things that come to mind when thinking of the game “Monument Valley” (words, phrases, or clauses available) and (2) What do you remember about your gaming experience with your finger and the stylus now, four weeks after playing the game? Please write down your thoughts (2e4 lines of text).

184

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

3. Results

Table 2 Summary of the results of univariate tests.

3.1. Data analysis For the quantitative data analyses of both the immediate and delayed evaluations, we implemented a three-way repeated-measures multivariate analysis of variance (MANOVA) with the input modality and time as within-subjects factors and story knowledge as a between-subjects factor. To compute the four DVs, we separately averaged the eight pairs of EQ, seven pairs of HQ, eight pairs of APPEAL, and four items of PE into individual composite scales. Because we found a significant interaction effect in the APPEAL scores, we also performed Bonferroni-adjusted paired t-tests for the interaction to determine the significant pairwise differences using a manually calculated Bonferroni correction. For the qualitative content analyses of participants' impressions about the game and their remembered gaming experiences across different input methods, we conducted two subsequent analyses. To compare the critical characteristics of Monument Valley across the story-knowledge groups, we classified all responses collected from the 80 participants into subordinate and superordinate characteristics (percentage agreement between two coders: 94.2% for eWOM-only and 93.6% for eWOM-Visual storytelling). Furthermore, we addressed whether the groups who read the same eWOM about the mobile game had certain characteristics in common. To find differences in gameplay experiences with the different input modalities, we analyzed the open-ended responses from 67 participants. First, all the data were classified as the advantages or disadvantages of each input method. On the basis of those categorized data, we re-classified the preference for input modality into response tendencies in terms of their response rates (percentage agreement between two coders: 95.4%). In the analyses, we used proportion scores as the unit of analysis rather than frequency scores to prevent verbose responses from being more heavily weighted than terse responses.

3.2. Multivariate and univariate statistics We found no significant multivariate effect for between-subject DVs (the combined EQ and HQ, APPEAL, and PE scores) across the story-knowledge groups (irrespective of input modality and time point): F (4, 73) ¼ 1.708, p ¼ 0.157, h2p ¼ 0.086. However, a betweengroup univariate analysis found significant differences across the story-knowledge groups in HQ, F (1, 76) ¼ 5.725, p < 0.05, h2p ¼ 0.070 (Table 1): participants in the eWOMeVisual storytelling group (M ¼ 5.590, SE ¼ 0.091) rated the HQ of the mobile game significantly higher than those in the eWOM-only group (M ¼ 5.286, SE ¼ 0.088). Based on the analyses, we draw the following points (Table 1):

Between subjects Story knowledge Hedonic quality Error Within subjects Input modality Hedonic quality Error Judgment of appeal Error Perceived enjoyment Error Time Ergonomic quality Error Input modality £ Time Judgment of appeal Error Note: *p < 0.05,

**

SS

df

MS

F-value

Sig.

h2p

7.169 95.169

1 76

7.169 1.252

5.725

0.019*

0.070

2.743 38.423 2.235 21.764 5.439 39.700

1 76 1 76 1 76

2.743 0.506 2.235 0.286 5.439 0.522

5.425

0.023*

0.067

2.462 41.382

1 76

2.297 15.578

1 76

7.803

0.007

**

0.093

10.412

0.002**

0.120

2.462 0.545

4.521

0.037*

0.056

2.297 0.205

11.204

0.001**

0.128

p < 0.01.

C The input modality produced significant multivariate effects in the repeated-measures MANOVA analyses (F (4, 73) ¼ 2.817, p < 0.05, h2p ¼ 0.134), as did time (F (4, 73) ¼ 2.933, p < 0.05, h2p ¼ 0.138). C There was a significant interaction effect between the input modality and time (F (4, 73) ¼ 3.720, p < 0.01, h2p ¼ 0.169). C The univariate analyses found significant main effects for the input modality on HQ (F (1, 76) ¼ 5.425, p < 0.05, h2p ¼ 0.067), APPEAL (F (1, 76) ¼ 7.803, p < 0.01, h2p ¼ 0.093), and PE (F (1, 76) ¼ 10.412, p < 0.01, h2p ¼ 0.120). C In the pairwise comparisons, participants who played the mobile game with their fingertips rated the HQ, APPEAL, and PE of the mobile game significantly higher than those who played the game with a stylus: (1) HQ, fingertip (M ¼ 5.532, SE ¼ 0.075) vs. stylus (M ¼ 5.344, SE ¼ 0.075); (2) APPEAL, fingertip (M ¼ 5.535, SE ¼ 0.061) vs. stylus (M ¼ 5.366, SE ¼ 0.066); and (3) PE, fingertip (M ¼ 5.901, SE ¼ 0.069) vs. stylus (M ¼ 5.637, SE ¼ 0.085). C Time produced a significant main effect on EQ, F (1, 76) ¼ 4.521, p < 0.05, h2p ¼ 0.056 (Table 2), such that participants in the delayed evaluation (M ¼ 3.659, SE ¼ 0.072) rated the EQ of the game significantly higher than they did in the immediate evaluation (M ¼ 3.837, SE ¼ 0.075). This can be attributed to the fact that the positive and negative adjectives of the EQ scales were anchored from 1 to 7, respectively (e.g., controllable (1)euncontrollable (7)). From the ANOVA results, we draw the following points (Table 2):

Table 1 Summary of the results of multivariate tests.

Between subjects Story knowledge Within subjects Input modality Input modality  Story knowledge Time Time  Story knowledge Input modality  Time Input modality  Time  Story knowledge Note: *p < 0.05,

**

p < 0.01.

Pillai's Trace

F-value (df, error df)

Sig.

0.086

1.708 (4, 73)

0.157

0.134 0.023 0.138 0.050 0.169 0.011

2.817 0.429 2.933 0.965 3.720 0.207

(4, (4, (4, (4, (4, (4,

73) 73) 73) 73) 73) 73)

h2p 0.086 *

0.031 0.787 0.026* 0.432 0.008** 0.934

0.134 0.023 0.138 0.050 0.169 0.011

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

C There was a significant interaction between the input modality and time for APPEAL (F (1, 76) ¼ 11.204, p < 0.01, h2p ¼ 0.128) but not for EQ, HQ, or PE. C In the interaction with APPEAL scores, playing the mobile game with the fingertip (M ¼ 5.576, SE ¼ 0.067) was perceived to be significantly more appealing than playing it with the stylus (M ¼ 5.235, SE ¼ 0.079) in the delayed evaluation, t (77) ¼ 4.639, p < 0.001 (two-tailed). C In the paired t-tests results, a significant difference emerged over time for playing the game with the stylus, t (77) ¼ 3.040, p < 0.008 (two-tailed). In the delayed evaluation, the gaming experience with the stylus was considered to be significantly less appealing than it was in the immediate evaluation (M ¼ 5.496, SE ¼ 0.078).

185

Fig. 1. Interaction effect of input modality and time on judgment of appeal.

3.3. Correlations in dependent measures EQ was negatively correlated with HQ (Table 3). In the immediate and delayed evaluations, the experience of playing the game with both the fingertip and the stylus showed very weak negative correlations between the EQ and HQ scores (jrj < 0.19) (see Fig. 1). 3.4. Analyses of participants' impressions of the game and remembered gaming experiences across input methods Our content analysis classified 347 responses (eWOM-only: n ¼ 176; eWOM-Visual storytelling: n ¼ 171) as reflecting 44 subordinate and 14 superordinate characteristics of Monument Valley. The 14 superordinate characteristics were: (1) impossible architecture, (2) challenge, (3) character, (4) fantasy, (5) femininity, (6) novelty, (7) review, (8) simplicity, (9) enjoyment, (10) sound, (11) quality, (12) adventure, (13) input modality, and (14) art. The category distribution of the most frequently mentioned characteristics varied depending on participants' story knowledge (Fig. 2). For the eWOM-only group, “impossible architecture” (13.07%) and “challenge” (13.07%) were the leading superordinate characteristics, followed by “character” (12.50%) and “fantasy” (11.93%). For the eWOM-Visual storytelling group, “novelty” was the most noticeable characteristic, representing 20.47% of all superordinate game characteristics. “Impossible architecture” and “fantasy” were ranked second with a 19.88% share each. “Art” (8.19%) was a distinguished feature that rarely appeared in the responses of the eWOM-only group. The most interesting part of the responses from both groups is the fact that other players' reviews were transmitted mouth to mouth even though the participants never experienced

Fig. 2. Comparison of Monument Valley characteristics across the two story-knowledge groups.

the events during the experiment. For example, participants cleared four or five stages on average, and no one cleared all ten stages of the game. Nonetheless, they said that it was too short for the price or that more levels should be added. The advantages and disadvantages of each input modality were re-classified into five different types of preferences for each input modality (Fig. 3). The results show that the preference rates for finger touch and the stylus were nearly identical: 43.28% and 41.79%, respectively. For both input modalities, a feeling of comfort

Table 3 Pearson's correlation matrix of dependent measures (Note: e1 ¼ immediate evaluation; e2 ¼ delayed evaluation; eF ¼ finger; eS ¼ stylus).

1-1. 1-2. 2-1. 2-2. 3-1. 3-2. 4-1. 4-2. 5-1. 5-2. 6-1. 6-2. 7-1. 7-2. 8-1. 8-2.

EQeF EQeF HQeF HQeF APPeF APPeF PEeF PEeF EQeS EQeS HQeS HQeS APPeS APPeS PEeS PEeS

1e1

1e2

2e1

2e2

3e1

3e2

4e1

4e2

5e1

5e2

6e1

6e2

7e1

7e2

8e1

8e2

1 0.271 0.009 0.138 0.125 0.014 0.052 0.035 0.092 0.011 0.147 0.030 0.047 0.014 0.029 0.006

1 0.095 0.141 0.107 0.177 0.141 0.331 0.009 0.219 0.180 0.213 0.037 0.160 0.040 0.211

1 0.468 0.667 0.487 0.568 0.392 0.246 0.176 0.315 0.365 0.300 0.300 0.273 0.079

1 0.384 0.485 0.381 0.516 0.260 0.252 0.253 0.381 0.275 0.145 0.339 0.153

1 0.592 0.495 0.409 0.075 0.112 0.253 0.449 0.356 0.408 0.206 0.123

1 0.434 0.566 0.076 0.022 0.218 0.360 0.381 0.498 0.363 0.279

1 0.566 0.095 0.114 0.264 0.220 0.296 0.232 0.447 0.333

1 0.170 0.224 0.232 0.178 0.232 0.241 0.342 0.298

1 0.363 0.163 0.102 0.188 0.128 0.255 0.134

1 0.039 0.104 0.150 0.355 0.125 0.292

1 0.470 0.519 0.280 0.483 0.341

1 0.394 0.567 0.408 0.535

1 0.417 0.644 0.345

1 0.362 0.570

1 0.550

1

Bold text indicates a statistically significant difference with a p-value less than 0.05.

186

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

Fig. 3. Preference for input modality for playing Monument Valley.

and the combination of high speed, sensitivity, and accuracy were regarded as fundamental determinants for preference. Most of all, feeling in control of the touch was essential to satisfactory interactions between the game and participants. When those aspects were achieved, the participants found themselves completely immersed and involved in the game, leading to greater enjoyment and better performance. When interrupted in their gameplay by the low speed, sensitivity, or accuracy of the input modality, they felt uncomfortable, irritated, and frustrated, leading to greater difficulty or confusion and poorer performance. In addition to their shared advantages and disadvantages, each input modality had its own pros and cons. Whereas finger touch allowed participants to feel familiar and friendly touch sensations, the stylus led them to feel unfamiliar and artificial touch sensations. In particular, the participants explicitly perceived the stylus as a user interface between the game and themselves that required an additional process for accurate positioning on the game elements. On the other hand, the stylus helped them play the game without tightening their grip, whereas finger touch caused hand or finger fatigue as they pressed directly on the hard glass surface of the touchscreen with their fingertips. Furthermore, they reported hand occlusion and fat finger problems, stating that their views were obscured by their fingers, making it difficult to play the game: “I am having a difficulty as my fingers are visually blocking the screen”; “My fingers are too fat and stubby to suitably fit in the small screens.” When using the stylus, some participants raised design issues such as that it was rather blunt, stiff, and heavy. Irrespective of those issues, other participants viewed the sense of touch and the sound of the stylus on the surface of the touchscreen as one of the greatest pleasures of playing the game; i.e., the change in the input modality itself was an exciting new experience for them. 4. Discussion 4.1. The findings from the univariate test Even though we found no main effect for the group difference in story-knowledge exposure in the multivariate test, the univariate test suggested that exposure to narrative game-metaphor-based knowledge led to a significant difference in assessing the HQ of Monument Valley. The participants exposed to both eWOM and M.C. Escher's works rated the HQ aspects of Monument Valley as significantly higher than those who were exposed only to the eWOM. Given that HQ is closely linked to sub-factors such as novelty, originality, and innovativeness, the qualitative analysis (Fig. 2) revealed that the eWOM-Visual storytelling group was more likely than the eWOM-only group to identify the link between the artist's works and the game, which apparently led to a wide variety of thoughts about novelty, impossible architecture, fantasy, and art.

In contrast, the eWOM-only group brought a similar proportion of game element-related thoughts from a more objective and realistic perspective. Surprisingly, the eWOM-Visual storytelling group rarely mentioned that the game presented a challenge, whereas the eWOM-only group mentioned how challenging the game was more than how novel it was or how closely it was related to the works of M.C. Escher. Even though the fact that the design of Monument Valley was inspired by his artwork was included in the eWOM to help participants find a critical clue while reading the experienced players' reviews, the eWOM-only group did not readily realize the connection between the drawings and the game. Perhaps they found it difficult to notice the clue because they lacked narrative game-metaphor-based knowledge, and they therefore underestimated the HQ of Monument Valley. Both story-knowledge groups were exposed to the same eWOM content in the immediate evaluation, and several individuals from each group left comments similar to the eWOM as if they had really experienced those events during the experiment. Likewise, a few player reviews that participants read before the immediate evaluation were passed from mouth to mouth in the delayed evaluation. These findings show that the interpretation of product information or new product experiences is closely related to users' preexisting cognitive structures or knowledge about the product, which are made up of what consumers actually know about the product (objective knowledge) and what they think they know about the product (subjective knowledge). Because many apps reward users for leaving a review on the App Store, it is difficult to rule out the possibility of either scanning others' reviews and then selectively copying and pasting parts of their reviews or writing down a quick comment. Indeed, this behavioral pattern has been observed even in controlled experimental situations. For skeptical customers, acquiring additional narrative game-metaphor-based knowledge along with eWOM might be more persuasive than reading only eWOM. In particular, mental imagery-eliciting strategies using game-related pictures, words, stories, or imagery could encourage prospective users to perceive a high degree of HQ in a game. Despite concerns about copycat reviews, some participants tried to express the thoughts that came to their mind first in their own words as follows: “A game appealing to adult sensibilities”; “A game worth playing with earphones in a crowded subway”; “It's not just a short game. It has a story. All the game stories are connected as one”; “The Monument Valley collection is definitely worth it.” Idiosyncratic comments such as those about users' overall impressions of the gaming experience can show game designers and developers how players' game-related schemata were structured. Moreover, referring to their impressions can help designers and developers prevent disconfirmation, dissatisfaction, and discontinuance in updates. 4.1.1. Findings from qualitative analysis The findings in the qualitative analysis rather conflict with those from the quantitative analysis. Coping with the issue of “fat finger” (Baudisch & Chu, 2009) and “hand occlusion” (Zaman et al., 2010), the qualitative analysis showed that a stylus was one of the most convenient and affordable alternatives to intuitive finger-based input, enhancing both precision and enjoyment in the gameplay. However, the univariate tests revealed that playing the game via finger touch led to significantly higher ratings on the perception of HQ, APPEAL, and PE than playing it via stylus. When interacting with time, the two input modalities showed a significant difference in the judgment of appeal, not in the immediate evaluation but in the delayed evaluation, such that the stylus led to a significantly unfavorable attitude toward the gaming experience over time. Based on those statistical results, we can completely neglect the

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

fact that a stylus could provide rich multisensory gaming experiences while dealing with the two main weaknesses of finger touch. Although it is true that the participants preferred finger touch to a stylus while playing the game, there was very little initial difference between the preferences for the two input modalities, as shown in Fig. 3. It is difficult to draw any clear conclusions on which of the two input modalities is better or worse in this research. However, the stylus we used might have had limited capabilities within the current technology. A variety of stylus-related issues will be improved with the release of new styli, and a solid foundation for offering a better multisensory experience will be established in the near future. A close look at the findings of the content analysis shows that finger touch elicited negative feelings about finger fatigue, fat finger, hand occlusion, and the size of display/game elements. Players wanted to play the game on an iPad: “I felt fatigue in my hand.”; “Finger touch forced me to use uncomfortable postures to perform manipulations such as rotating structures and moving the character to the exact point”; “I found it more difficult to play the game when blocking the view of the scene”; “It's not easy to touch with fingers because of the small screen.” In fact, Monument Valley was designed for both the iPhone and iPad to address precisely those issues and offer complete enjoyment of the gameplay and aesthetic design; however, purchasing a new iPad is not an option for an iPhone user. From this viewpoint, a genuinely affordable alternative to finger touch would be a stylus, as shown by the following stylus-related responses: “I'm glad I don't have to tighten my hand”; “I can move more accurately”; “It is more helpful to be immersed in playing the game”; “I felt the feeling of rotating architecture was one of the greatest pleasures of this game.” In spite of a stylus's great potential, the most important point is that high levels of speed, accuracy, and sensitivity are fundamental prerequisites for both input modalities to ensure that the players do not lose control of the game. When those prerequisites are not met, players might have considerable confusion and misunderstanding about whether they played the game incorrectly or the stylus did not work correctly, thereby leading them to underestimate the gaming experience itself. In fact, the stylus they used never broke down during the experiment, but it did autonomously go into the sleep mode during long idle times while they figured out solutions to the gaming tasks. This unexpected event interrupted the players' feeling of engagement by ruining their impression of the gameplay experience: “The stylus was not sensitive as I expected, so I couldn't know if the path was impassable because it was blocked or if the pen was not touching it exactly”; “I just completely wasted my time because the stylus malfunctioned so often compared to the finger touch”; “I think the gameplay was rather disturbed by the stylus use.” Furthermore, for the stylus to strengthen its weaknesses against the finger touch, it is important to bridge the gap between expectation and disconfirmation of the perceived gaming experience with different input modalities. In general, people tend to observe the nib and shaft of the stylus and the surface of the touchscreen pre-interaction, and then they expect, observe, and assess the results of the interaction between them (Chung & Shin, 2015). From the two different stages of interactions, the visual and haptic domains, it can be inferred that the participants experienced both their pre-interaction expectations and their post-interaction memories in the immediate and delayed evaluations: “I expected that I could play the game via the stylus better than via finger touch because its tip looked very sharp and fine-tuned, but it was uncomfortable during actual use.” Ideally, players should be able to not only be in control of the game whatever they use in the gameplay but also decide how to control games and which input modality to use to fully enjoy gameplay experiences with new perspectives and manipulation

187

skills. The tactile perceptions of the two direct input methods led to completely different gameplay experiences in terms of the viewpoint and the distance between the input modality and the game character: “I felt I put myself in the main character's shoes and uncovered hidden paths (Finger); I felt as if I manipulated architectural structures and moved the main character in the omniscient third person (Stylus).” A possible explanation for those distinctive experiences is that the strengths of finger touch and the weaknesses of the stylus might influence players' assessments of their multisensory gaming experiences. On the basis of our participants' responses to the double-edged haptic sensations of the two input modalities, we suggest that finger touch gave the participants more natural and familiar touch sensations than the stylus. We can carefully infer that the feeling of familiarity they experienced with finger touch made it easier to look at the game elements from a different viewpoint and distance. The participants seemed to experience the finger touch-based gameplay from a proximal, first-person viewpoint and the stylus-based gameplay from distal, omniscient thirdperson viewpoints: “It felt good controlling it myself (Finger).” “It seemed the character followed my movements (Finger).” “I felt like the game could understand my intention well (Finger).” “I felt like I wasn't playing it on my own (Stylus).” “I felt there was an artifact (also called a stylus) between the game and me (Stylus).” “I felt less able to play the game at a close distance/I felt I was playing it at a distant distance (Stylus).” In this sense, the difference in input modalities would be sufficient to generate interactivity effects on perception and engagement. The concept of input modality can be extended to modality interactivity (Sundar, Xu, & Bellur, 2010) by adding an interactive motor dimension to the game-story experience while enjoying the gameplay. If the input modality is appropriately chosen by players according to the game genre or combinations of the character's viewpoint and camera distance, the richness of players' multisensory gaming experiences will be enhanced. 4.1.2. Comparison of finger and stylus interaction game Based on our findings, we compare the usability of the stylus and finger interaction (Table 4). It is a worthwhile comparison because only a few studies have compared finger and stylus interaction. 4.1.3. Time-based data analysis As far as time is concerned, our findings in this study are inconsistent with the argument of previous studies (Diefenbach & Hassenzahl, 2011) that HQ can be achieved only by sacrificing EQ and vice versa because of their partial incompatibility. In this research, the statistical result revealed that the EQ of the game significantly increased over time without sacrificing the HQ. Moreover, the results of the correlation analysis showed very weak negative correlations between EQ and HQ. Because the positive and negative adjectives of the EQ scales were anchored from 1 to 7 (e.g., predictable (1)eunpredictable (7)), the negatively correlated relationships indicated that increased EQ led to increased HQ, and vice versa. These findings might be closely related to customers' and other critics' reviews, which praised the paid app Monument Valley for its visually and aurally aesthetic game design and criticized it for its lack of difficulty and replayability. In fact, the game was successful in that it was designed to provide players with an aesthetic experience in which every scene could be regarded as an artwork in itself. Indeed, almost all the players were satisfied with the artistic gameplay experience. Nevertheless, the lack of replayability of the paid game app might have led users to underestimate the difficulty and challenge of the game. Because increased EQ does not simply imply ease-of-use in this context, we cannot directly

188

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

Table 4 Comparison of stylus and finger interaction. Finger

Stylus

Usability

Tapping and moderate swiping.

Advantage

A finger tracking device tracks fingers in the 3D space or close to the surface without contact with the screen.

Limitations

Because fingers are far wider than a stylus, handwriting and other precise actions are rather difficult. - Hands can easily cover screen space that a user might want to see. - Fat fingers inevitably block some part of the display when touching the screen. Sudoku, puzzles. Higher accuracy in walking postures/mobile devices. Limited hand occlusion/fat finger problems.

Drawing, writing, or another kind of constant attention to the touchscreen. Good for games that involve navigating an onscreen maze, tracing the alphabet at various sizes, sketching a variety of items, or tablet tasks such as tapping around in apps and surfing the web. - With a stylus, hands do not block the view. The stylus makes transitions easy and provides a much less obstructed view of the puzzle field. Stylus fails to provide the tactile connection to the game screen that a finger provides.

Major use Accuracy Handwriting performance

conclude that an increase in EQ did not lead to a decrease in HQ or vice versa. 4.1.4. Theoretical implications Theoretically, the effect of story knowledge and input modality on the evaluation of gaming experience can be accounted for by the construal level theory (Trope & Liberman, 2010), which describes the relation between psychological distance and the extent to which people's thinking is abstract or concrete. Although having different story knowledge influenced only HQ ratings, the different input modalities influenced perceptions of HQ, APPEAL, and PE. According to the research model of Diefenbach and Hassenzahl (2011), perceived HQ plays a key role in users' judgment of appeal, and whether a software system is viewed as appealing by the users affects their emotional outcomes, such as fun/enjoyment and anger/frustration. From this perspective, input modality can be considered as a variable in judging whether a main task or distant future event is attractive (i.e., goal-relevant schemata as high-level construals), whereas story knowledge can be regarded as a variable to evaluate cognitive sub-tasks or near future events (i.e., goalirrelevant schemata as low-level construals). Compared to the exposure to different story knowledge, deciding how to control and enjoy a game using different modalities is closer to being a goalrelevant activity. This can be attributed to the fact that a fundamental goal of playing a game is to experience fun/enjoyment. Accordingly, game designers and developers need to pay as much attention to the issue of input modality as they do to marketing strategy, without dissociating hardware and software in their mind, even when designing mobile games. 5. Conclusion Based on that discussion, we summarize the following key highlights:  Assessment of a paid mobile game experience changes over time.  Both story-knowledge groups repeated the reviews of experienced players received by word-of-mouth.  The stylus shows potential as an alternative to fingers for touchscreen-based game input.  After a four-week delay, participants reported increased ergonomic quality without sacrificing hedonic quality.

iPad Pro, DSi, or 3DS. Higher accuracy in static situations (sitting)/stationary devices. - Stylus is more accurate and is good for resistive touchscreens. Good writing.

 A commercialized hedonic app and realistic gaming tasks could offer a complete picture of actual user experience. The results show significant main and interaction effects only for input modality and time in the evaluation of the complete gaming experience. However, the results of the univariate analysis show significant differences across story-knowledge groups in the assessment of hedonic quality. Although it is difficult to draw a clear conclusion about which of the two input modalities is better, we did observe that the stylus showed good potential as an alternative game controller. The results of this study contribute to our understanding of the role of narrative game-metaphor-based knowledge in eWOM communications and of input modality in the perception of a gaming experience over time. That expanded perspective on usability could take stakeholders another step toward designing players' gaming experiences and not merely considering game usability as the emotional consequence of EQeHQ trade-offs. Unlike previous studies on the dynamics of user experience with utilitarian software and hardware (Diefenbach & Hassenzahl, 2011; Karapanos et al., 2012; Shin & Shin, 2011), we used a direct-touchbased mobile game app, Monument Valley, a paid hedonic app with a definite end and low replayability. More important, our investigation of stylus-based interactions with realistic and ecologically valid tasks and applications has offered a relatively complete picture of actual user experience (Chung & Shin, 2015). Reviewing experienced players' comments in the App Store is no longer a key driver in making a purchase decision, but exposing participants to the artworks of M.C. Escher, which they might have heard of or seen before, might attract their attention and trigger their curiosity to play the game. Last but not least, our new experimental attempts to elicit insights into a neglected aspect of the gaming experience has provided practical and theoretical findings that will be helpful for game designers and developers as they prepare guidelines and recommendations for app development and evaluation methods.

6. Limitations and future studies Unlike another line of research to broaden the traditional concept of usability, the effects of aesthetic design on the value users assign to a mobile game before actually using it was not taken into account in this research to prevent participants from being desensitized to the game and to implicitly measure the effects of its

D.-H. Shin, K.-m. Chung / Computers in Human Behavior 68 (2017) 180e189

visual aesthetic design in further research. In this study, questions about the feelings and thoughts generated in participants' minds while seeing a series of M.C. Escher's works, which served as a metaphor in designing Monument Valley, could not be answered. Compared to other types of game genres, a second question yet to be answered is how the game's visual aesthetic design influenced affective or cognitive processing; several participants mentioned that it was a calm and peaceful game and not a brutal and violent one. The second limitation is that we chose only one puzzle game for this research. Despite our insightful findings on the effects of the input modality, time, and their interaction, whether the findings can be generalized to all story-based puzzle games or even other game genres with a definite story and ending has not yet been determined. In addition, the look and feel of Monument Valley is pretty and feminine, so its stakeholders' greatest concern could be to make more male players interested in their soft womanly game. Given that there are gender differences in game usage, behavior, and preference, further research with respect to different types of game genres should be carried out. Third, if the control mechanism to play mobile games is a major concern in future studies, deciding which smartphone and stylus to use should be carefully considered. During our experiment, the stylus rarely caused any inconvenience other than a short delay from the auto-sleep mode, which interrupted the players' engagement. Consequently, a stylus could negatively influence their performance and satisfaction in the immediate evaluation, which could then affect their impressions of the complete gaming experience in the delayed evaluation. Playing a mobile game consists of a variety of tapping, holding, and dragging tasks, so choosing a stylus design suitable for all those tasks will be important in future studies. References Ahn, D., & Shin, D. (2015). Differential effect of excitement versus contentment, and excitement versus relaxation: Examining the influence of positive affects on adoption of new technology with a Korean sample. Computers in Human Behaviors, 50, 283e290. Baudisch, P., & Chu, G. (2009). Back-of-device interaction allows creating very small touch devices. In CHI'09 proceedings of the 27th international conference on human factors in computing systems, Boston, USA (pp. 1923e1932). http:// dx.doi.org/10.1145/1518701.1518995. Chen, C., Nguyen, B., Klaus, P., & Wu, M. (2015). Exploring electronic Word-of-Mouth in the consumer purchase decision-making process. Journal of Travel & Tourism

189

Marketing, 32(8), 953e970. Chen, Y., & Xie, J. (2008). Online consumer review. Management Science, 54(3), 477e491. Cheung, C., & Thadani, D. (2012). The impact of electronic word-of-mouth communication. Decision Support Systems, 54(1), 461e470. Chung, K. M., & Shin, D. H. (2015). Effect of elastic touchscreen and input devices with different softness on user task performance and subjective satisfaction. International Journal of Human-Computer Studies, 83, 12e26. http://dx.doi.org/ 10.1016/j.ijhcs.2015.06.003. Diefenbach, S., & Hassenzahl, M. (2011). The dilemma of the hedoniceappreciated, but hard to justify. Interacting with Computers, 23(5), 461e472. Erkan, I., & Evans, C. (2016). The influence of eWOM in social media on consumers' purchase intentions. Computers in Human Behavior, 61, 47e55. Karapanos, E., Martens, J. B., & Hassenzahl, M. (2012). Reconstructing experiences with iScale. International Journal of Human-Computer Studies, 70(11), 849e865. http://dx.doi.org/10.1016/j.ijhcs.2012.06.004. Liu, Y., & Li, H. (2011). Exploring the impact of use context on mobile hedonic services adoption. Computers in Human Behavior, 27(2), 890e898. http:// dx.doi.org/10.1016/j.chb.2010.11.014. Mudambi, S. M., & Schuff, D. (2010). What makes a helpful review? MIS quarterly, 34(1), 185e200. See-To, E., & Ho, K. (2014). Value co-creation and purchase intention in social network sites: The role of electronic Word-of-Mouth and trust. Computers in Human Behavior, 31, 182e189. Shin, D. (2015). Quality of experience: Beyond the user experience of smart services. Total Quality Management, 26(8), 919e932. Shin, D. (2016). Cross-platform users' experiences towards designing interusable systems. International Journal of Human-Computer Interaction, 32(7), 503e514. Shin, D., & Ahn, D. (2013). Associations between game use and cognitive empathy. Cyberpsychology, Behavior, and Social Networking, 16(8), 599e603. Shin, D., & Shin, Y. J. (2011). Why do people play social network games? Computers in Human Behavior, 27(2), 852e861. Shin, D., Choi, M., Kim, J., & Lee, J. (2016). Interaction, engagement, and perceived interactivity in single-handed interaction. Internet Research, 26(5), 1134e1157. Sundar, S. S., Xu, Q., & Bellur, S. (2010). Designing interactivity in media interfaces: A communications perspective. In CHI'10 proceedings of the SIGCHI conference on human factors in computing systems, Atlanta, USA (pp. 2247e2256). http:// dx.doi.org/10.1145/1753326.1753666. Sweeney, J., Soutar, G., & Mazzarol, T. (2014). Factors enhancing word-of-mouth influence. European Journal of Marketing, 48(1/2), 336e359. http://dx.doi.org/ 10.1108/EJM-06-2012-0336. Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological distance. Psychological Review, 117(2), 440e463. Tsao, W., & Hsieh, M. (2012). Exploring how relationship quality influences positive eWOM. Total Quality Management, 23(7), 821e835. Vasa, R., Hoon, L., Mouzakis, K., & Noguchi, A. (2012). A preliminary analysis of mobile app user reviews. In Proceedings of the 24th Australian computer-human interaction conference (pp. 241e244). http://dx.doi.org/10.1145/ 2414536.2414577.  zquez-Casielles, R., Sua rez-Alvarez, L., & Río-Lanza, A. (2013). The word of mouth Va Dynamic. Journal of Advertising Research, 53(1), 43e60. Zaman, L., Natapov, D., & Teather, R. J. (2010). Touchscreens vs. traditional controllers in handheld gaming. In Proceedings of the international academic conference on the future of game design and technology (pp. 183e190). http:// dx.doi.org/10.1145/1920778.1920804.