sustainability

Article Pattern Preference Analysis of Black-and-White Plaid

Qianling Jiang 1 , Li-Chieh Chen 1, Chun Yang 2 and Jie Zhang 3,*

1 The Graduate Institute of Design Science, Tatung University, Taipei 104, Taiwan; [email protected] (Q.J.); [email protected] (L.-C.C.) 2 Graduate School of Design, National Yunlin University of Science & Technology, Yunlin 640, Taiwan; [email protected] 3 Institute of and Clothing, The Hong Kong Polytechnic University, Hong Kong, China * Correspondence: [email protected]; Tel.: +86-15217014324

 Received: 1 September 2018; Accepted: 15 October 2018; Published: 17 October 2018 

Abstract: The market economy has shifted the decision-making power of the garment industry from the enterprise to the consumer. Research on consumer clothing preferences is an essential part of sustainable development of the garment industry. Based on data statistics from eight fast fashion brands, black and white are most commonly used in two-color plaid shirts. This paper carried out a psychophysical experiment to investigate factors affecting pattern preferences for black-and-white shirts and the differences and similarities between male and female pattern preferences. Twenty-eight different representative patterns of plaid shirts were selected by five fashion designers together from 190 different black-and-white plaid shirts from eight fast fashion brands, which were then classified into three categories: , , and windowpane. Based on these patterns, 28 male and female shirts were simulated in three dimensions and presented on a calibrated computer display. The simulations were assessed by 42 observers (consisting of 21 males and 21 females) in terms of four semantic scales, including light–dark, delicate–rough, simple–complex, and like–dislike. The experimental results revealed that there was no significant difference of pattern preference between females and males for 89.29% of the black-and-white plaid shirts, and also described features of the patterns that the females and males liked or disliked. Furthermore, the study also demonstrated the formulation between the four semantic scales and three pattern features (including the percentage of black region, the size of the minimum repeat unit, and the descriptor of the pattern complexity). The findings could be used to develop a more robust and comprehensive theory of pattern preferences and provide a reference for pattern design for black-and-white plaid shirts. More comprehensive pattern preference theory is not only an effective tool to solve the problem of plaid inventory in the garment industry but also an important theoretical basis for the “sustainable design” of clothing, which is of great significance to the sustainable development of the garment industry.

Keywords: pattern preference; plaid shirts; pattern categories

1. Introduction

1.1. Research Background and Motivation For the garment industry, inventory has always been one of its main pain points. The product backlog not only occupies the company’s operating capital but also increases the company’s management and profit costs, reducing the company’s overall profits, and is also an obstacle to the sustainable development of the garment industry. Linkshop [1] collated the annual inventory

Sustainability 2018, 10, 3739; doi:10.3390/su10103739 www.mdpi.com/journal/sustainability Sustainability 2018, 10, 3739 2 of 18

Sustainability 2018, 10, x FOR PEER REVIEW 2 of 18 data of eight Chinese listed garment enterprises from 2012 to third quarter 2016, including HLA, YOUNGOR, SEMIR,SEMIR, HODO, HODO, METERS METERS BONWE, BONWE, SHANSHAN, SHANSHAN, JOEONE, JOEONE, and and BUSEN; BUSEN; these these data weredata calculatedwere calculated from thefrom annual the annual reports reports of enterprises, of enterprises, as shown as shown in Figure in Figure1. 1.

Figure 1. Stock status of eight Chines Chinesee clothing companies in 2012–2016.

The figurefigure above above shows shows that that the the inventory inventory value value of YOUNGOR of YOUNGOR exceeded exceeded 10 billion 10 RMB billion annually RMB everyannually year every from year 2012 from to third 2012 quarter to third 2016, quarter and 2 even016, and reached even 23.473 reached billion 23.473 RMB billion in 2012. RMB Even in 2012. the lowestEven the inventory lowest inventory value of thevalue eight of companies,the eight companies, BUSEN, stillBUSEN, exceeded still exceeded 200 million 200 RMB million annually RMB betweenannually 2012between and 2012 2016. and It can2016. be It seen can be that seen the that inventory the inventory problem problem is becoming is becoming an obstacle an obstacle to the sustainableto the sustainable development development of the of garment the garment industry. indust Thery. studyThe study of pattern of pattern preference preference is not is not only only an effectivean effective tool tool with with which which to solve to solve the inventory the inventory problem problem of the garment of the garment industry butindustry also an but important also an theoreticalimportant theoretical basis for the basis sustainable for the sustainable design of apparel. design of It isapparel. of great It significanceis of great significance to the sustainable to the developmentsustainable development of the garment of the industry. garment industry. To solvesolve the the inventory inventory problem problem of the of garment the garm industryent industry and promote and the promote sustainable the development sustainable ofdevelopment the garment of the industry, garment we industry, need not we only need a not strong only supply a strong chain supply system chain but system also but rapid also market rapid responsemarket response ability to ability support to thesupport extremely the extremel high inventoryy high turnover.inventory Consumerturnover. Consumer clothing preference clothing researchpreference is anresearch effective is toolan effective with which tool to with support which the marketto support reaction the market abilityof reaction the garment ability industry. of the Thegarment market industry. economy The causes market the economy decision-making causes the power decision-making of the garment power industry of the to be garment transferred industry from enterprisesto be transferred to consumers. from enterprises By understanding to consumers. the preferences By understanding of consumers, the preferences clothing enterprises of consumers, can avoidclothing unsold enterprises clothing can as much avoid as unsold possible clothing and reduce as much inventory. as possible By understanding and reduce the inventory. preferences By of consumers,understanding the the design preferences and production of consumers, of clothing the des enterprisesign and production can reduce of undesirable clothing enterprises clothing andcan thusreduce reduce undesirable inventory. clothing and thus reduce inventory. “Sustainable design” is formed in the deep thinkingthinking of the relationship between human development andand environmental environmental problems problems and and the practicethe practice of constantly of constantly seeking seeking change. change. The strategy The ofstrategy the “sustainable of the “sustainab design”le design” of clothing of clothing can besummed can be summed up in three up in ways: three sustainableways: sustainable design design based onbased a material on a material foundation foundation aims at ai reducingms at reducing the use ofthe materials use of mate andrials optimizing and optimizing the use of the materials, use of andmaterials, intuitively and reflectsintuitively the reflects ecological the concept ecological of environmental concept of environmental protection; sustainable protection; design sustainable based ondesign conveying based on specific conveying ideas specific uses emotional ideas uses design emotional methods design to conveymethods sustainable to convey sustainable ideas sensibly ideas to consumers;sensibly to consumers; and sustainable and sustainable design based design on lifebased cycle on considerationslife cycle considerations aims at prolonging aims at prolonging clothing usageclothing and usage service and life. service Overall, life. the Overall, third method the thir isd themethod easiest is tothe implement easiest to and implement can be achieved and can in be a shortachieved period in a ofshort time; period the study of time; of clothingthe study preference of clothing can preference improve can the durabilityimprove the of durability the relationship of the betweenrelationship clothing between and clothing people, therebyand people, reducing thereb consumptiony reducing consumption and waste of an resources.d waste of resources. From the above point of view, this paper carried out a psychophysical expe experimentriment to investigate factors affecting pattern preferences for black-and-whiteblack-and-white shirts and the differences and similarities between male and female pattern preferences. Psyc Psychophysicshophysics has has been defined defined as “the analysis of perceptual processes by studying the effect on a subject’s experience or behavior of systematically varying the properties of a stimulus along one or more physical dimensions” [2]. Sustainability 2018, 10, 3739 3 of 18 perceptual processes by studying the effect on a subject’s experience or behavior of systematically varying the properties of a stimulus along one or more physical dimensions” [2]. Research on fabric preferences has mostly focused on color. For instance, Jiang investigated the color of clothing and emphasized the rule that color brightness influences positive–negative emotional categorization [3]. Hsu carried out a psychophysical experiment to investigate factors affecting color preferences for Taiwanese floral pattern fabrics as a case study of object color preference [4]. These studies mainly stressed color. Compared with color, pattern is also an important part of textiles. This researcher summarized some studies on patterns.

• Eun Yi Kim discussed challenging issues in how to index images based on human emotions and presented a heuristic approach to emotion-based textile indexing using colors and texture [5]. • Soo-jeong Kim proposed a textile indexing system that can classify textile images based on human emotions [6]. • Na Yeon Kim proposed a neural network-based approach for emotion-based textile indexing [7]. • So-Ra Lee found out the effects of textile motifs and application methods on the wearer’s image perception [8]. • Na Yeon Kim proposed a novel method using color and pattern information for recognizing emotions included in a textile [9]. • Soo-Kyoung Choi investigated the image of casual shirts according to the color combination, tone, and interval of the checked pattern in tone-in-tone coloration. The images, according to color combination, tone, and interval of checked pattern, consisted of five dimensions of attractiveness, youth-activity, appeal, modesty, and freshness. This study discussed how the color combination, tone, and interval of checked pattern affects the five dimensions [10]. • Felecia Davis presented a study that related the communication of emotion to people via the aesthetic textural expression of computational textiles or textiles that transform through electronic commands over time [11]. • Felecia Davis investigated what still and shape-changing textural expressions of computational textiles can communicate emotionally to people [12]. • Siri Homlong studied how simple and complex patterns in printed textile fabrics are perceived and expressed verbally, as well as how judgments, concepts, and values in relation to designed textile patterns are expressed by schoolchildren, consumers, teachers of textile handicrafts, and designers [13].

The above research did not explore the pattern preference of -dyed fabrics. Therefore, this study probed into textile patterns and how patterns influence consumer preferences. This study hypothesized that men and women will have different or similar pattern preferences for textiles. It was worth exploring the characteristics of these patterns. The object of this paper was to investigate the patterns of plaid shirts made of two-color yarn-dyed fabrics. There were several reasons why these objects were selected for investigation. First, from the design of technologies to the wide application of plaids in modern clothing, plaids have evolved along with the history and civilization of humans. It was therefore significant to regard plaids as the basic research objects of patterns of textiles. Second, this study focused on the aesthetic differences and similarities between men and women in terms of plaids. Clothing has no gender; it is classified into men’s and women’s clothing because of the gender of the wearers. It can be seen that the gender of clothing is a consensus established by society. No other commodity in the world has such distinct gender differences as clothing. Hence, it was reasonable to consider clothing rather than other textiles as the research object. Lastly, the type of clothing that is commonly worn by males and females is plaid shirts. The application of plaids has a long history. Today, is thought to have originated in ; however, tartan was first discovered far from the British Isles. According to the textile historian E. J. W. Barber, the of Central Europe, which is linked with ancient Celtic Sustainability 2018, 10, 3739 4 of 18 populations and flourished between the 8th and 6th centuries BC, produced tartan-like textiles [14]. Some of them were discovered in 2004, remarkably preserved, in the Hallstatt salt mines near , Austria [15]. Textile analysis of fabric from the Tarim in Xinjiang, northwestern China has also shown it to be similar to that of the Iron Age Hallstatt culture [16]. At around 1700 AD, plaid patterns in Scotland started to reflect distinct geographical features. Martin Martin, in A Description of the Western Islands of Scotland, published in 1703, wrote that Scottish tartans could be used to distinguish the inhabitants of different regions. He expressly wrote that the inhabitants of various islands and the mainland of the Highlands were not all dressed alike, but that theSustainability setts and 2018, 10 colours, x FOR PEER of REVIEW thevarious tartans varied from isle to isle [15].4 of People 18 in the same Sustainability 2018, 10, x FOR PEER REVIEW 4 of 18 Sustainability 2018, 10, x FOR PEER REVIEW 4 of 18 area tendedSustainability to adopt 2018, the10, x FOR same PEER plaids.REVIEW This also meant that plaid patterns in Scotland4 of 18 were formed. SustainabilityAt around 2018, 10 1700, x FOR AD, PEER plaid REVIEW patterns in Scotland started to reflect distinct geographical features.4 of 18 SustainabilityAt around 2018, 10 1700, x FOR AD, PEER plaid REVIEW patterns in Scotland started to reflect distinct geographical features.4 of 18 Furthermore,Martin everyAt Martin,around family in1700 A AD,Description or plaid tribe patterns of had the itsWesternin Scotland own Islands plaid started of patterns,Scotland,to reflect distinctpublished known geographical in as1703, family wrote features. that plaids, which were MartinAt Martin,around 1700in A AD,Description plaid patterns of the Westernin Scotland Islands started of toScotland, reflect distinctpublished geographical in 1703, wrote features. that MartinScottishAt Martin, aroundtartans incould1700 A AD,Description be used plaid to patterns distinguishof the Westernin Scotland the i nhabitantsIslands started of Scotland,toof reflectdifferent distinctpublished regions. geographical Hein 1703, expressly wrote features. wrote that also a symbolMartinScottish of Martin, statustartans incould and A Description be wealth. used to distinguishof In the addition, Western the inhabitantsIslands plaids of Scotland,of were different laterpublished regions. developed Hein 1703, expressly wrote for wrote that different occasions. ScottishMartinthat the Martin, inhabitantstartans incould A of Description bevarious used toislands distinguishof the and Western the the main i nhabitantsIslandsland ofof the Scotland,of differentHighlands published regi wereons. not Hein all1703, expressly dressed wrote wrotealike, that For instance,Scottishthat there the inhabitantstartans were could plaids of be various used specific toislands distinguish for and formal the the main inhabitants andland of hunting the of differentHighlands clothes. regi wereons. not WithHe all expressly dressed the advent wrote alike, of the industrial thatScottishbut thatthe inhabitantstartansthe setts could and of becoloursvarious used oftoislands thedistinguish various and the thetartans main inhabitantsland varied of thefrom of Highlandsdifferent isle to isleregi were [15].ons. not HePeople all expressly dressed in the wrotealike,same thatbut thatthe inhabitantsthe setts and of coloursvarious ofislands the various and the tartans mainland varied of thefrom Highlands isle to isle were [15]. not People all dressed in the alike,same age, variousbutthatarea countries that thetended inhabitantsthe settsto adopt and and of regionsthe coloursvarious same ofislands plaids. beganthe various andThis to the also tartans register main meantland va riedthat theirof the fromplaid Highlands own islepatterns to plaids. isle were in [15]. Scotland not InPeople all the dressedwere in combination the formed. alike,same of aesthetics butarea that tended the settsto adopt and thecolours same of plaids. the various This also tartans meant va riedthat fromplaid isle patterns to isle in [15]. Scotland People were in the formed. same areabutFurthermore, that tended the settsto every adopt and family thecolours same or oftrib plaids. thee had various This its own also tartans plaidmeant va patterns, riedthat fromplaid know islepatterns nto as isle family in [15]. Scotland Peopleplaids, were whichin the formed. samewere and the consumerareaFurthermore, tended psychologyto every adopt family the same orof trib plaids. moderne had This its ownalso society, plaidmeant patterns, that plaids plaid know patterns becamen as family in Scotland a plaids, design were which elementformed. were and source of Furthermore,areaalso atended symbol to everyof adopt status family the and same or wealth. trib plaids.e had In Thisaddition, its own also plaidmeantplaids patterns, thatwere plaid later know patternsdevelopedn as family in Scotlandfor differentplaids, were which occasions. formed. were Furthermore,also a symbol everyof status family and or wealth. tribe had In addition, its own plaidplaids patterns, were later know developedn as family for differentplaids, which occasions. were inspirationalsoFurthermore,For in laterinstance,a symbol eras everythereof status and familywere and were plaids orwealth. trib fully specifice hadIn appliedaddition, its for own formal plaidplaids to theandpatterns, were hunting design later know developed clothes. ofn as textiles family With for differentplaids, the such advent which asoccasions. clothing.of were the alsoFor instance,a symbol thereof status were and plaids wealth. specific In addition, for formal plaids and were hunting later developed clothes. With for different the advent occasions. of the Foralsoindustrial instance,a symbol age, thereof various status were countriesand plaids wealth. andspecific In regions addition, for beganformal plaids to and registerwere hunting later their developed clothes. own plaids. With for different Inthe the advent combination occasions. of the ManyForindustrial variations instance, age, there various of thewere countries majorplaids andspecific patterns regions for formalbegan exist, toand register as hunting designers their clothes. own plaids. constantly With Inthe the advent combination come of the up with new and industrialForof aesthetics instance, age, and there various the were consumer countries plaids psychology andspecific regions for of beganformal mode to rnand registersociety, hunting theirplaids clothes. own became plaids. With a design Inthe the advent combinationelement of andthe different styles.industrialof aesthetics Shirt age, and patternsvarious the consumer countries are oftenpsychology and regions enhanced of began mode torn or registersociety, can betheirplaids a own combination became plaids. a designIn the ofcombinationelement different and patterns. ofindustrialsource aesthetics of inspiration age, and various the inconsumer countrieslater eras psychology andand regionswere fullyof began mode applied torn registersociety, to the designtheirplaids own becameof textiles plaids. a design suchIn the as combinationelement clothing. and ofsource aesthetics of inspiration and the consumerin later eras psychology and were fullyof mode appliedrn society, to the designplaids becameof textiles a designsuch as element clothing. and The classicsourceof plaids aestheticsMany of inspiration shownvariations and the inconsumer inof later the Table erasmajor psychology 1and consistpatterns were fully of exist, ofmode applied horizontalasrn designers society, to the designplaids constantly and ofbecame verticaltextiles come a designsuch up lines withas element clothing. new and and planes processed sourceMany of inspiration variations in of later the erasmajor and patterns were fully exist, applied as designers to the design constantly of textiles come such up aswith clothing. new and sourcedifferentMany of styles.inspiration variations Shirt inpatterns of later the erasmajor are andoften patterns were enhanced fully exist, applied or as can designers be to athe combination design constantly of textiles of comedifferent such up checkwithas clothing. newpatterns. and by yarn-dyeddifferentMany technology. styles. variations Shirt patterns of The the picturesmajor are often patterns enhanced in Tableexist, or as can1 designerswere be a combination drawn constantly usingof comedifferent Matlabup checkwith newpatterns. in and this study, and the differentThe classicMany styles. variationsplaids Shirt shown patterns of thein Table major are often 1 patternsconsist enhanced of exist, horizontal or as can designers be and a combination vertical constantly lines of and comedifferent planes up checkwith processed newpatterns. and by differentThe classic styles. plaids Shirt shown patterns in Table are often 1 consist enhanced of horizontal or can be and a combination vertical lines of anddifferent planes check processed patterns. by descriptionThedifferentyarn-dyed of classic check styles. plaidstechnology. patterns Shirt shown patterns The refersin Tablepictures are often to1 consist thein enhanced Table research of horizontal1 orwere can ofdrawn be and Christinaa combination vertical using Matlablines Lee of and different in [17 planesthis]. study,check processed patterns.and theby Theyarn-dyed classic plaidstechnology. shown The in Tablepictures 1 consist in Table of horizontal1 were drawn and vertical using Matlablines and in planesthis study, processed and theby yarn-dyedThedescription classic technology.plaidsof check shown patterns The in Tablepictures refers 1 toconsist inthe Table research of horizontal1 were of Christina drawn and vertical usingLee [17]. Matlablines and in planesthis study, processed and theby yarn-dyeddescription technology. of check patterns The pictures refers to inthe Table research 1 were of Christina drawn using Lee [17]. Matlab in this study, and the descriptionyarn-dyed technology.of check patterns The pictures refers to inthe Table research 1 were of Christina drawn usingLee [17]. Matlab in this study, and the descriptionyarn-dyedTable technology.of 1. checkClassic patterns The plaids picturesrefers consisting to inthe Table research of 1 horizontalwere of Christina drawn andusingLee [17]. vertical Matlab in lines this andstudy, planes and the [17 ]. description ofTable check 1. patterns Classic plaids refers consisting to the research of horizontal of Christina and vertical Lee [17].lines and planes [17]. description ofTable check 1. patterns Classic plaids refers consisting to the research of horizontal of Christina and vertical Lee lines[17]. and planes [17]. Table 1. Classic plaids consisting of horizontal and vertical lines and planes [17]. PlaidPlaid Table 1.Example Classic plaids consisting of horizontal and vertical Difinition lines Difinition and planes [17]. Plaid Table 1.Example Classic plaids consisting of horizontal and vertical Difinition lines and planes [17]. Plaid Example Gingham usually comes Difinition in a checkered pattern and is Plaid Example Gingham usually comes Difinition in a checkered pattern and is Plaid Example GinghamGingham usually usually comes comes Difinition in in a checkereda checkered pattern pattern and and is distinguishedis by white Plaid Example distinguishedGingham usually by white comes and Difinition incolored, a checkered even-sized pattern checks. and is This anddistinguished colored, even-sized by white checks.and colored, This patterneven-sized is formed checks. by This horizontal and vertical GinghamGingham patterndistinguishedGingham is formed usually by bywhite horizontal comes and incolored, aand checkered vertical even-sized patternstripes checks. (usuallyand isThis of Gingham patternstripesdistinguished (usuallyis formed by of bywhite the horizontal same and color) colored, and that vertical even-sized cross eachstripes checks. other (usually on This a whiteof background to Gingham thepatterndistinguished same is color) formed bythat bywhite cross horizontal and each colored, other and verticalon even-sized a white stripes background checks. (usually This ofto Gingham thepattern same is color) formed that by cross horizontal each other formand verticalon even a white checks. stripes background (usually ofto Gingham thepattern same is color) formed that by cross horizontalform each even other and checks. verticalon a white stripes background (usually ofto the same color) that crossform each even other checks. on a white background to the same color) that crossform each even other checks. on a white background to the same color) that crossform each even other checks. on a white background to Tartan plaid is the patternform that even is checks.most often found on Scottish TartanTartan plaid plaid is is the the pattern patternform that even that is is checks.most most oftenoften found found on on Scottish Scottish . This plaid Tartan Tartankilts. This plaid plaid is the consists pattern of that vertical is most and often horizontal found or on diagonal Scottish TartanTartan Tartankilts.consists This plaid plaid of is vertical the consists pattern and of horizontal that vertical is most and or often diagonalhorizontal found stripes oron diagonal Scottish that cross each other to Tartan Tartankilts.stripes This plaid that plaid crossis the consists eachpattern other of thatvertical to is form most and differently often horizontal found sized or on diagonal checks.Scottish Tartan kilts.stripes This that plaid cross consists each other ofform vertical to differentlyform and differently horizontal sized checks.sized or diagonal checks. Tartan kilts.stripes This that plaid cross consists each other of vertical to form and differently horizontal sized or diagonal checks. Tartan kilts.stripes This that plaid cross consists each other of vertical to form and differently horizontal sized or diagonal checks. Thisstripes pattern that crossis a -weave each other ofto small,form differently even-sized, sized colored, checks. and Thisstripes pattern that iscross a twill-weave each other ofto small,form differently even-sized, sized colored, checks. and Thiswhite pattern checks. is aWhile twill-weave this check of small, often resembleseven-sized, the colored, gingham and Shepherd’s ThiswhiteThis pattern patternchecks. is a isWhile twill-weave a twill-weave this check of small, ofoften small, even-sized,resembles even-sized, the colored, gingham colored, and and white checks. Shepherd’s Thiswhitecheck, pattern checks. the is visible aWhile twill-weave twillthis check weave of small, often is what resembleseven-sized, distinguishes the colored, gingham the and Shepherd’sCheck whiteWhilecheck, checks. this the check visible While often twillthis resembles check weave often is the what resembles gingham distinguishes check,the ginghamthe thevisible twill weave is Shepherd’sShepherd’sCheck Check whiteshepherd’scheck, checks. the check visible While from twillthis gingha check weave m.often is Thewhat resembles distinguishes the gingham pattern the Shepherd’sCheck shepherd’swhatcheck, distinguishes the check visible from twill the gingha weave shepherd’sm. is Thewhat check houndstooth distinguishes from gingham. pattern the The houndstooth Check shepherd’scheck, theoriginated check visible from twill from gingha weave the m.Shepherd’s is Thewhat houndstooth distinguishes check. pattern the Check shepherd’soriginated checkpattern from from gingha originated the m.Shepherd’s The from houndstooth the check. Shepherd’s pattern check. Check shepherd’soriginated check from from gingha the m.Shepherd’s The houndstooth check. pattern Theshepherd’s windowpaneoriginated check check from from is gingha a thepatte Shepherd’sm.rn The that houndstooth resembles check. the pattern pattern The windowpaneoriginated check from is a pattethe Shepherd’srn that resembles check. the pattern Windowpane The windowpaneof panesoriginated on a checkwindow. from is a Thepattethe Shepherd’sstripesrn that that resembles check.cross to the form pattern Windowpane The windowpaneof panes on a check window. is a patteThe stripesrn that that resembles cross to the form pattern WindowpaneCheck windowpaneTheThe windowpaneof windowpanepanes checks on a checkwindow. are check oftenis a is Thepatte athic pattern stripesrnker that and that that resembles farther resembles cross apart to the form the thanpattern pattern the of panes on a WindowpaneCheck windowpaneof panes checks on a window. are often The thic stripesker and that farther cross apart to form than the WindowpaneWindowpaneCheck Check windowpanewindow.of panes The checks on stripespattern a window. are that foundoften cross The thic in to stripesgraphker form and windowpane checks.that farther cross apart to form checks than arethe often thicker and Check windowpane checkspattern are foundoften thic in graphker and checks. farther apart than the Check windowpane checksfartherpattern are apart foundoften than thic in thegraphker pattern and checks. farther found apart in graph than checks.the Check Thiswindowpane is a check checks patternpattern are that foundoften rese thicinmbles graphker theand checks. crossing farther apartlines ofthan graph the This is a check patternpattern that found rese inmbles graph the checks. crossing lines of graph Thispaper. is a Thecheck graph patternpattern check that found pattern rese inmbles is graph characterized the checks. crossing by lines solid, of graphthin, Thispaper. is a Thecheck graph pattern check that pattern resembles is characterized the crossing by lines solid, of graphthin, Graph Check Thispaper.and is single-colored a The check graph pattern check stripes that pattern rese thatmbles crossis characterized the each crossing other byto lines formsolid, of even graphthin, Graph Check paper.and Thissingle-colored The is graph a check check patternstripes pattern that that crossis resembles characterized each other the crossing byto formsolid, lines even thin, of graph paper. Graph Check andpaper.and small-sized single-colored The graph checks. check stripes The pattern thatstripes crossis characterizedthat each create other a graph byto formsolid, check even thin, are Graph Check andTheand small-sized graphsingle-colored check checks. pattern stripes The is characterizedthatstripes cross that each create by other solid, a graph to thin, form check and even single-colored are stripes GraphGraph Check Check andand small-sized single-coloredthinner thanchecks. thestripes The stripes thatstripes in cross a thatwindowpane each create other a graph to check. form check even are andthat small-sized crossthinner each thanchecks. other the to The stripes form stripes even in a andthat windowpane small-sizedcreate a graph check. checks. check The are stripes that create and small-sizedthinner thanchecks. the The stripes stripes in a thatwindowpane create a graph check. check are andTattersall small-sizedthinnera graph is athan checks. checkcheck the arepattern Thestripes thinner stripes that in thana consiststhatwindowpane the create stripes of thin,a graph check. in regularly a windowpanecheck are check. is a check pattern that consists of thin, regularly Tattersallspacedthinner stripes is athan check in alternatingthe pattern stripes that incolors a consistswindowpane that are of thin,repeated check. regularly both Tattersallspaced stripes is a check in alternating pattern that colors consists that are of thin,repeated regularly both Tattersall horizontallyTattersallspaced stripes isand a check vertically.in alternating pattern The that colors stripes consists that that are ofcreate thin,repeated the regularly tattersall both Tattersall horizontallyspaced stripes and vertically.in alternating The colors stripes that that are create repeated the tattersall both Tattersall horizontallyspacedTattersall stripes and is a checkvertically.in alternating pattern The thatcolors stripes consists that that are create of repeated thin, the regularly tattersall both spaced stripes in Tattersall horizontallypatternspaced stripesoften and come vertically.in alternating in two The different colorsstripes thatcolors that are create and repeated are the usually tattersall both alternatingpattern often colors come that in are two repeated different both colors horizontally and are usually and vertically. The stripes Tattersall horizontallypattern often anddarker come vertically. thanin two the The different background stripes colors that color. create and are the usually tattersall Tattersall pattern oftendarker come thanin two the different background colors color. and are usually thatpattern create often thedarker tattersall come thanin patterntwo the different background often comecolors color. in and two are different usually colors and are usually pattern oftendarker come than in two the different background colors color. and are usually Sustainability 2018, 10, x FOR PEER REVIEW darker than the background color. 5 of 18 darker than the background color.

ThisThis is a pattern is a pattern consisting consisting of very of very small small and andeven-sized even-sized checks. checks. It usually Mini-checkMini-check Itconsists usually ofconsists one color of one with color white with and white often and resembles often resembles the gingham check-except the gingham check-exceptthat that it is it much is much smaller. smaller.

1.2. Research Purpose As mentioned above, with the changing times, more and more types of plaid fabrics have been produced. Each has its own unique characteristics. Pattern and color are two important components of plaids. As a basic study on plaid preferences, this study explored two-color plaids in the first part to lay a foundation for follow-up research on multi-color plaids. This study probed into the effects of these characteristics on the preferences of males and females in the hope of identifying the factors influencing the preference of two-color plaid shirts. The research questions included the following: 1. What is the relationship between the plaid features and aesthetic emotions? 2. What types of two-color plaid shirts show significant differences in preferences between men and women? 3. What are the characteristics of two-color plaid shirts that are liked or disliked by both men and women? In order to answer these questions, this study conducted a psychophysical experiment that covered four semantic scales, including light–dark, delicate–rough, simple–complex, and like– dislike. A 10-point scale was adopted to assess the fabric and patterns of plaid shirts. Regression analysis between the four semantic scales and three pattern features was performed, including an independent sample t-test of differences in the preferences between men and women, a one-sample t-test of differences in the preferences of different patterns, and Two-Way Mixed-Design ANOVA.

2. Methods

2.1. Data Collection Through an online survey, this study asked two questions: Do you often buy plaid shirts? What brands do you choose? A total of 67 replies confirmed that they purchased plaid shirts. Eight brands were mentioned the most, including H&M, Zara, Uniqlo, C&A, GAP, Muji, Forever 21, and New Look. After receiving these responses, this researcher used the Internet to collect pictures of 190 pieces of two- color plaid shirts from the eight brands. The selection criteria included (a) two-color plaids and (b) shirts, regardless of material. Of the total, 117 pieces were men’s shirts, while the other 73 were women’s. The colors in the pictures of these shirts were extracted by Adobe Photoshop software, which is the industry standard in digital imaging and is used worldwide for design, photography, video editing, and more. In line with the L values (Li,1, Li,2) under the CIE LAB color space, the 117 pairs of colors were classified into 117 light colors and 117 dark colors, in which i indicates the i-th shirt, and 1 and 2 stand for Color 1 and Color 2 of the i-th shirt respectively, as shown in Figure 2. It could be seen from the above figure that black and white were used the most in the two-color plaid shirts, and that black and white was the most common color combination. Therefore, this study analyzed the preference of black-and-white plaid shirts. Through the analysis of the 190 pictures of two-color plaid shirts, five fashion designers discussed and drew 28 representative plaids. Matlab 2016 was used to generate simulated drawings of these yarn-dyed fabrics. See Figure 3 for the simulated drawings of the 28 plaids (the figure displays the simulated drawing of fabrics with the size 15 cm × 15 cm). Sustainability 2018, 10, 3739 5 of 18

1.2. Research Purpose As mentioned above, with the changing times, more and more types of plaid fabrics have been produced. Each has its own unique characteristics. Pattern and color are two important components of plaids. As a basic study on plaid preferences, this study explored two-color plaids in the first part to lay a foundation for follow-up research on multi-color plaids. This study probed into the effects of these characteristics on the preferences of males and females in the hope of identifying the factors influencing the preference of two-color plaid shirts. The research questions included the following:

1. What is the relationship between the plaid features and aesthetic emotions? 2. What types of two-color plaid shirts show significant differences in preferences between men and women? 3. What are the characteristics of two-color plaid shirts that are liked or disliked by both men and women?

In order to answer these questions, this study conducted a psychophysical experiment that covered four semantic scales, including light–dark, delicate–rough, simple–complex, and like–dislike. A 10-point scale was adopted to assess the fabric and patterns of plaid shirts. Regression analysis between the four semantic scales and three pattern features was performed, including an independent sample t-test of differences in the preferences between men and women, a one-sample t-test of differences in the preferences of different patterns, and Two-Way Mixed-Design ANOVA.

2. Methods

2.1. Data Collection Through an online survey, this study asked two questions: Do you often buy plaid shirts? What brands do you choose? A total of 67 replies confirmed that they purchased plaid shirts. Eight brands were mentioned the most, including H&M, Zara, Uniqlo, C&A, GAP, Muji, Forever 21, and New Look. After receiving these responses, this researcher used the Internet to collect pictures of 190 pieces of two-color plaid shirts from the eight brands. The selection criteria included (a) two-color plaids and (b) shirts, regardless of material. Of the total, 117 pieces were men’s shirts, while the other 73 were women’s. The colors in the pictures of these shirts were extracted by Adobe Photoshop software, which is the industry standard in digital imaging and is used worldwide for design, photography, video editing, and more. In line with the L values (Li,1, Li,2) under the CIE LAB color space, the 117 pairs of colors were classified into 117 light colors and 117 dark colors, in which i indicates the i-th shirt, and 1 and 2 stand for Color 1 and Color 2 of the i-th shirt respectively, as shown in Figure2. It could be seen from the above figure that black and white were used the most in the two-color plaid shirts, and that black and white was the most common color combination. Therefore, this study analyzed the preference of black-and-white plaid shirts. Through the analysis of the 190 pictures of two-color plaid shirts, five fashion designers discussed and drew 28 representative plaids. Matlab 2016 was used to generate simulated drawings of these yarn-dyed fabrics. See Figure3 for the simulated drawings of the 28 plaids (the figure displays the simulated drawing of fabrics with the size 15 cm × 15 cm). Sustainability 2018, 10, 3739 6 of 18 Sustainability 2018, 10, x FOR PEER REVIEW 6 of 18

Sustainability 2018, 10, x FOR PEER REVIEW 6 of 18

(a) (b) (a) (b)

FigureFigure 2. Color 2. Color clustering clustering scatter scatter diagramdiag diagramram basedbased based onon 190190 plaid plaid shirts shirtsshirts in in inCIE CIECIE LAB LABLAB color colorcolor space. space.space. (a) Light (a) LightLight colorcolor distribution. distribution. ( b) ( Darkb) Dark color color distribution. distribution.

FigureFigure 3. 3.Simulated Simulated drawings of of plaids. plaids.

AfterAfter obtaining obtaining the the simulated simulated drawings drawings ofof thethe 28 fabrics, fabrics, CLO CLO 3D 3D software software was was used used to simulate to simulate the men’s and women’s shirts usingFigure these 3. Simulated patterns. drawings CLO 3D is of a plaids.3D fashion design software program the men’s and women’s shirts using these patterns. CLO 3D is a 3D fashion design software program creating virtual, true-to life garment visualization with cutting-edge simulation technologies for creating virtual, true-to life garment visualization with cutting-edge simulation technologies for fashion Afterfashion obtaining designers. the CLO simulated 3D can drawingsbe used as ofa thedesign 28 fabrics,tool, specifically CLO 3D forsoftware design wasdevelopment, used to simulate to designers.the men’svisualize and CLO garments women’s 3D can based shirts be usedon using their as a2Dthese design CAD patterns. patterns. tool, specifically CLO It is 3Dused is toa for 3Dvisualize design fashion true-to-life development, design software 3D garments. to visualizeprogram garmentscreatingExamples virtual, based of on thetrue-to their men’s 2Dlife shirts CAD garment are patterns. shown visualizat in It Figure is usedion 4. to with visualize cutting-edge true-to-life simulation 3D garments. technologies Examples for of thefashion men’s designers. shirts are CLO shown 3D in can Figure be 4used. as a design tool, specifically for design development, to visualize garments based on their 2D CAD patterns. It is used to visualize true-to-life 3D garments. Examples of the men’s shirts are shown in Figure 4. Sustainability 2018, 10, 3739 7 of 18 Sustainability 2018, 10, x FOR PEER REVIEW 7 of 18

FigureFigure 4. 4.Simulated Simulated drawings of of men’s men’s plaid plaid shirts. shirts.

2.2.2.2. Classification Classification of of Plaids Plaids TheThe 28 28 plaids plaids were were drawn drawn byby fivefive fashionfashion designersdesigners by by referring referring to to 190 190 two-color two-color plaid plaid shirts shirts fromfrom the the eight eight brands. brands. TheseThese plaidsplaids containedcontained five five types types of of plaids plaids (gingham, (gingham, tartan, tartan, mini-check, mini-check, windowpane,windowpane, and and graph graph check) check) as as classified classified byby Christina Lee, Lee, which which were were the the most most common common patterns patterns amongamong the the 190 190 two-color two-color plaid plaid shirts. shirts. ThroughThrough the observations observations of of this this study, study, the the mini-check mini-check could could bebe regarded regarded as as a a smaller smaller type type of of gingham,gingham, andand the graph check check could could be be seen seen as as a asmaller smaller type type of of windowpane.windowpane. Thus, Thus, these these plaids plaids could could be be classified classified into into gingham, gingham, tartan, tartan, and and windowpane. windowpane. Among Among the 28the plaids, 28 plaids, 16 were 16 firstwere selected. first select Checksed. Checks 1–4 under 1–4 under gingham gingham had anhad evenly-distributed an evenly-distributed black black and and white pattern.white pattern. In terms In of terms the other of the plaids, other plaids, after the after exchange the exchange of black of black and white, and white, there there were were obvious obvious visual differences.visual differences. Therefore, Therefore, the black the and black white and colors white of co thelors other of the 12 other plaids 12 wereplaids exchanged. were exchanged. The research The objectsresearch were objects added were to the added 28, asto shownthe 28, as in shown Table2 inbelow. Table 2 below.

TableTable 2.2. ClassificationClassification of the 28 28 plaids. plaids.

TypeType of Plaid of Plaid Plaid Plaid No. No. Ginham Checks 1/2/3/4 Ginham Checks 1/2/3/4 TartanTartan ChecksChecks 5/6/7/8/9/10/11/12/13/14/15/16/17/18 5/6/7/8/9/10/11/12/13/14/15/16/17/18 WindowpaneWindowpane ChecksChecks 19/20/21/22/23/24/25/26/27/28 19/20/21/22/23/24/25/26/27/28

AfterAfter the the 28 28 plaids plaids were were selected selected byby thethe fivefive fashion designers, designers, this this researcher researcher asked asked the the fashion fashion designersdesigners to to write write classification classification criteriacriteria forfor thethe selectionselection of of these these plaids. plaids. They They did did not not need need to towrite write thethe criterion criterion for for the the typetype of of plaid, plaid, as as each each type type of plaid of plaid could could have havemore morethan one than type one of typefabric. of After fabric. discussion, they identified three criteria: light–dark, delicate–rough, and simple–complex. This study After discussion, they identified three criteria: light–dark, delicate–rough, and simple–complex. used the three pairs of adjectives to analyze and describe the 28 plaids and hypothesized that the This study used the three pairs of adjectives to analyze and describe the 28 plaids and hypothesized three pairs of adjectives could be used to describe the plaids associated with the preferences. that the three pairs of adjectives could be used to describe the plaids associated with the preferences. Sustainability 2018, 10, 3739 8 of 18 Sustainability 2018, 10, x FOR PEER REVIEW 8 of 18

2.3.2.3. Test Test Stimuli Stimuli InIn order order to to avoid avoid the the three three adjectives adjectives of of ligh light–dark/delicate–rough/simple–complext–dark/delicate–rough/simple–complex interfering interfering withwith the the choice choice of of preferences, preferences, the the experiment experiment wa wass divided divided into into two two parts. parts. Meanwhile, Meanwhile, this this study study consideredconsidered that that those those who who had had received received training training could could conduct conduct a amore more reliable reliable assessment assessment of of the the threethree pairs pairs of of adjectives. adjectives. The The images images for for the the test test in in the the first first part part were were the the above above 28 28 simulated simulated images images ofof fabrics fabrics and and the the 3D 3D simulated simulated images images of of clothes. clothes. In In the the experiment, experiment, only only one one preference preference indicator, indicator, like–dislike,like–dislike, was was used used to tojudge judge the the extent extent of the of thepref preferenceerence for fordifferent different patterns patterns of the of observers. the observers. In theIn first the firstpart partof the of experiment, the experiment, a 10-step a 10-step mandat mandatoryory scale scale with with choice choice questions questions was wasemployed employed to judgeto judge the theextent extent of preference. of preference. The The 10 10steps steps includ includeded “extremely “extremely dislike”, dislike”, “exceptionally “exceptionally dislike”, dislike”, “fairly“fairly dislike”, dislike”, “dislike”, “dislike”, “somewhat “somewhat dislike”, dislike”, “somewhat “somewhat like”, like”, “like”, “like”, “fairly “fairly like”, like”, “exceptionally “exceptionally like”,like”, and and “extremely “extremely like”. like”. The The respondents respondents were were forced forced to to make make a achoice choice between between like like and and dislike dislike ratherrather than than selecting selecting a aneutral neutral response. response. The The images images used used in in the the test test are are shown shown in in Figure Figure 55 (in (in which which thethe image image on on the the left left (a) (a) is is a abasic basic type type of of men’s men’s shirt shirt for for a aheight height of of 175 175 cm, cm, the the middle middle (b) (b) is is a abasic basic typetype of of women’s women’s shirt shirt for for a height a height of of165 165 cm, cm, and and the the right right (c) is (c) a issimulated a simulated fabric fabric with with a size a sizeof 15 of cm15 × cm 15 ×cm).15 cm).The respondents The respondents were wereinformed informed of these of these images images before before the test. the test.

(a) (b) (c)

FigureFigure 5. 5. ImagesImages used used in in the the questionnaire. questionnaire. (a ()a Men’s) Men’s shirt, shirt, (b ()b women’s) women’s shirt, shirt, and and (c) ( csimulated) simulated fabric. fabric.

InIn the the second second part part of of the the experiment, experiment, 30 30 observer observerss from from the the first first part part were were selected selected purposefully. purposefully. TheyThey received received training training on on graphic graphic design design to to can can make make more more precise precise judgments judgments of of the the three three pairs pairs of of adjectives.adjectives. They They were were asked asked to to fill fill in in the the same same 10-st 10-stepep mandatory mandatory scale scale with with choice choice questions questions to to assess assess thethe 28 28 plaids plaids using using the the three three pairs pairs of adjectives: of adjectives: light–dark, light–dark, delicate–rough, delicate–rough, and simple–complex. and simple–complex. The fourThe semantic four semantic scales scales of aesthetic of aesthetic emotions emotions are arepresented presented in inChinese Chinese during during the the experiment. experiment. An An EnglishEnglish translation translation of of the the four four scales scales is is shown shown in in Table Table 3.3.

TableTable 3. 3. FourFour semantic semantic scales scales of of aesthetic aesthetic emotions. emotions.

ChineseChinese English English 明亮的明亮的-黑暗的-黑暗的 Light–darkLight–dark 精緻的精緻的-粗獷的-粗獷的 Delicate–roughDelicate–rough 簡單的簡單的-複雜的-複雜的 Simple–complexSimple–complex 不喜歡的 喜歡的 不喜歡的--喜歡的 Dislike-likeDislike-like

2.4.2.4. Display Display InIn the the firstfirst part part of theof experiment,the experiment, this study this usedstudy two used Apple two 27-inch Apple iMacs 27-inch (model: iMacs MD095CH/A) (model: MD095CH/A)as the display. as The the two display. monitors The weretwo monitors under the we samere under indoor the environment same indoor and environment light source. Duringand light the source.experiment, During other the experiment, interference other factors interference were avoided, factors and were the sameavoided, brightness and the and same display brightness calibration and displaywas set calibration for the two was monitors. set for the During two monitors. the test, an During image the was test, displayed an image for was 10 sdisplayed to avoid interferencefor 10 s to avoidwith interference the next image. with the next image. In the second part of the experiment, this study used the A0 international standard-sized paper to print the 28 plaid shirts and fabrics shown above. All 28 plaids were displayed simultaneously, Sustainability 2018, 10, 3739 9 of 18

In the second part of the experiment, this study used the A0 international standard-sized paper to print the 28 plaid shirts and fabrics shown above. All 28 plaids were displayed simultaneously, which helped the observers to make a judgment. Unlike the main part of the experiment, the observers could change their scores on fabric that they had already assessed.

2.5. Observers The first part of the experiment was aimed to collect data. A team of 42 observers participated in the first part of the experiment, including 21 males and 21 females aged 18 to 33 years old. During the experiment, each observer was asked to sit in front of a monitor that was placed approximately 55 cm away from him/her. Each observer then rated his/her preference of the 28 images. In the second part of the experiment, 30 observers were selected among the 42 observers from the first part. These 30 observers, including 15 males and 15 females, received training on graphic design. There were 11 doctoral students, 13 postgraduates, and six undergraduates specializing in design-related majors, and seven had backgrounds related to fashion design. During the experiment, each observer was asked to sit approximately 1.3 m in front of the A0 paper. Each observer scored the 28 images according to the following three pairs of adjectives: light–dark, delicate–rough, and simple–complex.

2.6. Research Limitation All 42 observers are young Chinese people, and 30 of them are majoring in design (at various education levels). This is a very specific consumer group that cannot be used directly to translate the results into universal consumer preference of patterns. However, the proposed investigation method can be used to analyze the patterns of preference consumers of different ages and nationalities. Furthermore, this paper just focuses on the black-and-white plaid shirts and does not include the solid and multi-color plaid shirts, which will be investigated further in the future.

3. Results

3.1. Analysis of Reliability The reliability of the assessment data for the preference and the adjective pairs was analyzed. The reliability coefficient values were 0.938, 0.990, 0.993, and 0.991, all of which were greater than 0.9 and therefore indicated that the reliability of the research data was high. With respect to Cronbach’s Alpha after deletion of an item, the reliability coefficient value after the analysis item was deleted did not obviously improve; therefore, all items were retained, and the reliability of the research data was high. The corrected item-total correlation (CITC) values of the analysis items were higher than 0.9, indicating good correlation between items and good reliability. In summary, the reliability coefficient of the research data was higher than 0.9 and did not increase after an item was deleted. Therefore, the data were reliable and suitable for further analysis.

3.2. Regression Analysis In accordance with the simulated drawings of the 28 plaid fabrics, this study obtained three important characteristic fabric parameters: (1) the black regional percentage, x1; (2) the physic size of the minimum repeat unit (centimetre), x2; and (3) the descriptor of the pattern complexity, x3:

N x = black (1) 1 w × h

(w + h) x = (2) 2 2 w h 2 w/2 h/2 2 w/3 h/3 2 w/4 h/4 2 (pi,j−p1) 4×(pi,j−p2) 9×(pi,j−p3) 16×(pi,j−p4) x3 = ∑ ∑ w×h + ∑ ∑ w×h + ∑ ∑ w×h + ∑ ∑ w×h (3) i = 1 j = 1 i = 1 j = 1 i = 1 j = 1 i = 1 j = 1 Sustainability 2018, 10, x FOR PEER REVIEW 10 of 18

in which pij, is the gray level of the (i, j)-th pixel in the simulated images of yarn-dyed fabrics and p1 , p2 , p3 , and p4 stand for the average gray levels of the minimum repeat unit of the 1, 1/2, 1/3, and 1/4 parts. The correlation scatter matrix diagram of light–dark (y1), delicate–rough (y2), simple–complex (y3), and x1/x2/x3 is shown in Figure 6. The correlations between y1 and x1 and between y2 and x2 were strong and presented a good linear correlation. Sustainability 2018, 10, 3739 10 of 18 This study adopted regression analysis to obtain the linear relationship between light–dark (y1), delicate–rough (y2), simple–complex (y3), and x1/x2/x3, as shown in Table 4. The values of correlation coefficientin which p ir,j betweenis the gray the level fitted of value the (i ,andj)-th the pixel actual in the value simulated were 0.96, images 0.91, of and yarn-dyed 0.96, respectively, fabrics and indicatingp1, p2, p3, strong and p4 correlationstand for the and average good fitting gray effects. levels of Analysis the minimum of variance repeat was unit performed of the 1, next. 1/2, 1/3,The Fand test 1/4 was parts. employed to judge the regression effect of the regression model. According to the results, p = 2EThe − 12, correlation which was scatter significantly matrixdiagram less than of 0.05. light–dark The overall (y1), regression delicate–rough equation (y2), was simple–complex significantly effective.(y3), and xTherefore,1/x2/x3 is it shown was possible in Figure to6. describe The correlations subjective between assessmentsy1 and yx1,1 yand2, and between y3 throughy2 and thex2 characteristicswere strong and of presentedobjective fabric a good images linear x correlation.1, x2, and x3.

Figure 6. CorrelationCorrelation analysis analysis of of the the scatter scatter matrix matrix diagra diagramm (Corr (Corr indicates indicates the the correlation correlation coefficient). coefficient).

This study adoptedTable regression 4. Summary analysis of regression to obtain analysis the linear of the relationship fitting results. between light–dark (y1), delicate–rough (y ), simple–complex (y ), and x /x /x , as shown in Table4. The values of correlation Factors 2 Regression Equation3 1 Multiple2 3 R Adjusted R Square Significance F coefficient r between the fitted value and the actual value were 0.96, 0.91, and 0.96, respectively, Light–dark y1 = 11.09 x1 − 0.39 0.96 0.93 2.28E − 16 2 F indicatingDelicate–rough strong correlation y2 = 1.32 and + 0.50 good x2 − fitting 0.01 x2 effects. 0.91 Analysis of variance 0.81 was performed 4.63E next.− 10 The test wasSimple–complex employed toy judge3 = 2.84 the x3 − regression 0.11 x2 + 2.03 effect x1 of the0.96 regression model. 0.88 According 5.88E to the− 14 results, p = 2E − 12, which was significantly less than 0.05. The overall regression equation was significantly effective.The locally Therefore, weighted it was scatterplot possible tosmoothing describe (LOWESS subjective or assessments LOESS) is usedy1, y 2to, andregressiony3 through analyze the thecharacteristics relationship of between objective different fabric images genderx1 ,preferencesx2, and x3. and three adjective pairs (y1: light–dark, y2: delicate–rough, and y3: simple–complex). Locally weighted regression, including LOWESS and LOESS, originally proposedTable 4.bySummary Cleveland of regression[18] and further analysis developed of the fitting by results. Cleveland and Devlin [19], specificallyFactors denotes a method Regression that is Equation also known as Multiple locallyR weightedAdjusted polynomialR Square regression Significance [18,19].F Locally weighted regression is based on a smoothing procedure that pays greater attention to the Light–dark y1 = 11.09 x1 − 0.39 0.96 0.93 2.28E − 16 2 local points.Delicate–rough Parametersy2 =of 1.32 polynomial + 0.50 x2 − 0.01 regressionx2 model0.91 can change 0.81 with different 4.63E independent− 10 variables;Simple–complex in other words,y3 = differ 2.84 x3ent− 0.11 observationsx2 + 2.03 x1 can correspond0.96 to different 0.88 group of parameters 5.88E − 14 and given corresponding estimated values of regression function. The biggest advantage LOESS has over The locally weighted scatterplot smoothing (LOWESS or LOESS) is used to regression analyze the relationship between different gender preferences and three adjective pairs (y1: light–dark, y2: delicate–rough, and y3: simple–complex). Locally weighted regression, including LOWESS and LOESS, originally proposed by Cleveland [18] and further developed by Cleveland and Devlin [19], specifically denotes a method that is also known as locally weighted polynomial regression [18,19]. Locally weighted regression is based on a smoothing procedure that pays greater attention to the local points. Parameters of polynomial regression model can change with different independent variables; Sustainability 2018, 10, 3739 11 of 18

Sustainabilityin other words, 2018, 10 different, x FOR PEER observations REVIEW can correspond to different group of parameters and 11 given of 18 corresponding estimated values of regression function. The biggest advantage LOESS has over many manyother other methods methods is the is fact the that fact itthat does it does not requirenot require the specificationthe specification of a of function a function to fit to afit model a model to allto allof theof the data data in thein the sample. sample. Instead, Instead, the the analyst analys onlyt only has has to provide to provide a smoothing a smoothing parameter parameter value value and andthe degreethe degree of the of localthe local polynomial, polynomial, which which can becan used be used to smooth to smooth the datathe data and and remove remove the noisethe noise [20]. [20].In addition, In addition, LOESS LOESS is very is very flexible, flexible, making making it ideal it ideal for modelingfor modeling complex complex processes processes for for which which no notheoretical theoretical models models exist. exist. The The analysis analysis results results are shown are shown in Figure in Figure7, in which 7, in which the black the line black is the line LOESS is the LOESScurves approximatecurves approximate the relationship the relationship between t differentbetween gendert different preferences gender and preferences three adjective and three pairs. adjective pairs. 1. Female preference is more correlated with y2. There is a negative linear relationship between 1. Femalefemale preference andis morey1, y 3correlated. Female preferencewith y2. There has a is positive a negative linear linear relationship relationship with ybetween2. 1 3 2 2. femaleMale preference preferenceis and more y , y correlated. Female preference with y3. Male has preferencea positive linear has a relationship positive linear with relationship y . 2. Male preference is more correlated with y3. Male preference has a positive linear relationship with y1, y2 and a negative linear relationship with y3. 3. withMen y prefer1, y2 and dark a negative plaids, linear and women relationship prefer with light y3 plaids.. Both men and women prefer rough 3. Menplaids, prefer while dark men plaids, are more and receptive women to prefer delicate light plaids plaids. than Both women. men Both and men women and womenprefer rough prefer plaids,simple while plaids, men and are the mostmore complexreceptive plaid to delic (checkate 9)plaids is the than least women. liked by Both men andmen women. and women prefer simple plaids, and the most complex plaid (check 9) is the least liked by men and women.

FigureFigure 7.7. CorrelationCorrelation analysisanalysis between between different different gender gend preferenceser preferences and and three three adjective adjective pairs (Corrpairs (Corrindicates indicates the correlation the correlation coefficient). coefficient).

3.3.3.3. Independent Independent Sample Sample t t Test Test AnAn independent independent sample sample tt-test-test was was used used to to analyze analyze the the differences differences in in the the preference preference for for gingham, gingham, tartan,tartan, and and windowpane windowpane from from the the viewpoint viewpoint of of gender gender.. As As shown shown in in Figure Figure 88 ofof thethe heatheat mapmap ofof thethe independentindependent sample tt-test-test andand thethe test test data, data, the the results results showed showed that that observers observers of differentof different genders genders did didnot not show show a significant a significant difference difference for gingham for gingham (p > 0.05),(p > 0.05), which which indicated indicated that differentthat different genders genders had a hadconsistent a consistent preference preference for gingham. for gingham. In other words,In othe therer words, was nothere difference was no in difference preference. in Additionally, preference. Additionally,observers of differentobservers genders of different showed significantgenders showed differences significant for tartan differences and windowpane for tartan (p < 0.05),and windowpaneimplying that ( differentp < 0.05), genders implying had that varied different preferences gender fors tartan had andvaried windowpane. preferences According for tartan to and the windowpane.specific analysis, According the significance to the specific level of analysis, gender for the tartan significance was 0.05 level (t = 2.04;of genderp = 0.04). for tartan In line was with 0.05 the (specifict = 2.04; differences,p = 0.04). In line the averagewith the forspecific the males differences, (5.83) was the average significantly for the higher males than (5.83) that was for significantly the females higher(5.19). Thethan significancethat for the females level of gender(5.19). The for significance windowpane level was of 0.01 gender (t = 2.90;for windowpanep < 0.01). In linewas with0.01 ( thet = 2.90;specific p < differences,0.01). In line the with average the specific for the differences, males (5.99) the was average significantly for the higher males than (5.99) that was for significantly the females higher than that for the females (4.89). Based on the above, each gender did not show significant differences in their preferences for gingham but showed significant differences in their preferences for tartan and windowpane, and the preferences of the males for the two types of plaids were significantly higher than those of the females. Sustainability 2018, 10, x FOR PEER REVIEW 12 of 18

An independent sample t-test was used to analyze the differences in the preferences for the 28 plaids from the viewpoint of gender. The results demonstrated that observers of different genders did not show a significantly different preferences for 25 plaids (checks 1–19, 21, 23, and 25–28) (p > 0.05), indicating that different genders had consistent preferences for them. In other words, there was no difference in preferences. In addition, observers of different genders showed significant differences in preferences for three plaids (checks 20, 22 and 24) (p < 0.05), implying that different genders had varied preferences for them. In accordance with the specific analysis, the significance level of gender for tartan was 0.05 (t = 2.03; p < 0.05). In line with the specific differences, the average for the males (6.10) was significantly higher than for the females (4.48). The significance level of gender for check 22 was 0.05 (t = 2.54; p < 0.05). In line with the specific differences, the average for the males (6.48) was significantly higher than that for the females (4.33). The significance level of gender for check 24 was 0.05 (t = 2.63; p < 0.05). In line with the specific differences, the average for Sustainabilitythe males2018 (6.24), 10, 3739 was significantly higher than that for the females (4.24). Based on the above, gender12 of 18 did not show significant differences in preferences for 25 plaids (checks 1–19, 21, 23, and 25–28) but showed significant differences in preferences for three plaids (checks 20, 22, and 24). Based on the (4.89).data, Based the three on the plaids above, could each be identified gender did as liked not show by males significant and disliked differences by females. in their preferences for ginghamAn but independent showed significant sample t differences-test was performed in their preferencesto probe into for the tartan differences and windowpane, in light–dark andy1, the preferencesdelicate–rough of the males y2, and for simple–complex the two types y of3 between plaids were genders. significantly The results higher showed than that those there of were the females.no significant differences (p > 0.05) between genders in any of the three adjective pairs of the 28 plaids.

FigureFigure 8. Independent8. Independent sample sample t-test and and thermogram thermogram of oftest test data. data.

An independentAs shown in the sample above Figuret-test was8, the used values to of analyze simple–complex the differences y3 for checks in the 20, preferences22, and 24 were for the 5.2, 5.9, and 3, respectively. If 5.5 was regarded as the median value for y3, y3 had no influence on the 28 plaids from the viewpoint of gender. The results demonstrated that observers of different genders preferences for checks 20, 22, and 24. All the values of light–dark y1 for checks 20, 22, and 24 were did not show a significantly different preferences for 25 plaids (checks 1–19, 21, 23, and 25–28) (p > 0.05), more than 7, and all the values for delicate–rough y2 were less than 4. The other plaids did not have indicatingthese characteristics that different at genders the same had time. consistent Hence, this preferences study considered for them. that, In for other plaids words, liked by there males was no differenceand disliked in preferences. by females, In y addition,1 will be greater observers than 7, of and different y2 will be genders less than showed 4. When significant y1 and y2 meet differences these in preferencescharacteristics, for three y3 plaidswill not (checks affect preference. 20, 22 and 24) (p < 0.05), implying that different genders had varied preferences for them. In accordance with the specific analysis, the significance level of gender for tartan was 0.053.4. One-Sample (t = 2.03; p t-Test< 0.05). In line with the specific differences, the average for the males (6.10) was significantlyBecause higher a 10-point than for scale the was females used (4.48).in the experime The significancent, 5.5 points level represented of gender neither for check dislike 22 wasnor 0.05 (t = 2.54;like. pA< one-sample 0.05). In line t-test with was the used specific to study differences, the types the of averageplaids that for were the malesliked or (6.48) disliked was by significantly both highermales than and that females, for the as females shown in (4.33). Figure The 8. The significance one-sample level t-test of was gender used for to check if 24 the was preferences 0.05 (t = 2.63; p < 0.05).for gingham, In line withtartan, the and specific windowpane differences, were obviously the average not forequal the to males 5.5. The (6.24) results was showed significantly that there higher thanwere that forno significant the females differences (4.24). Basedin the preferen on theces above, for gingham, gender didtartan, not and show windowpane significant between differences males and females (p > 0.05), implying that the averages for gingham, tartan, and windowpane were in preferences for 25 plaids (checks 1–19, 21, 23, and 25–28) but showed significant differences in preferences for three plaids (checks 20, 22, and 24). Based on the data, the three plaids could be identified as liked by males and disliked by females. An independent sample t-test was performed to probe into the differences in light–dark y1, delicate–rough y2, and simple–complex y3 between genders. The results showed that there were no significant differences (p > 0.05) between genders in any of the three adjective pairs of the 28 plaids. As shown in the above Figure8, the values of simple–complex y3 for checks 20, 22, and 24 were 5.2, 5.9, and 3, respectively. If 5.5 was regarded as the median value for y3, y3 had no influence on the preferences for checks 20, 22, and 24. All the values of light–dark y1 for checks 20, 22, and 24 were more than 7, and all the values for delicate–rough y2 were less than 4. The other plaids did not have these characteristics at the same time. Hence, this study considered that, for plaids liked by males and disliked by females, y1 will be greater than 7, and y2 will be less than 4. When y1 and y2 meet these characteristics, y3 will not affect preference. Sustainability 2018, 10, 3739 13 of 18

3.4. One-Sample t-Test Because a 10-point scale was used in the experiment, 5.5 points represented neither dislike nor like. A one-sample t-test was used to study the types of plaids that were liked or disliked by both males and females, as shown in Figure8. The one-sample t-test was used to check if the preferences for gingham, tartan, and windowpane were obviously not equal to 5.5. The results showed that there were no significant differences in the preferences for gingham, tartan, and windowpane between males andSustainability females 2018 (p >, 10 0.05),, x FOR implying PEER REVIEW that the averages for gingham, tartan, and windowpane were 13 all of close18 to 5.5 without statistical difference. Both men and women neither liked nor disliked the three patterns veryall close much. to 5.5 without statistical difference. Both men and women neither liked nor disliked the three patternsA one-sample very much.t-test was employed to identify the types of plaids that were liked or disliked by A one-sample t-test was employed to identify the types of plaids that were liked or disliked by both males and females. The one-sample t-test was used to check if the preferences of the 28 plaids both males and females. The one-sample t-test was used to check if the preferences of the 28 plaids were obviously not equal to 5.5. The following Figure9 indicated that the averages for 25 plaids were obviously not equal to 5.5. The following Figure 9 indicated that the averages for 25 plaids (checks 1–8, 11–25, 27 and 28) were all close to 5.5 (p > 0.05), the average of the preference for check (checks 1–8, 11–25, 27 and 28) were all close to 5.5 (p > 0.05), the average of the preference for check 26 was significantly higher than 5.5 (p < 0.05), and the averages of the preference for checks 9 and 10 26 was significantly higher than 5.5 (p < 0.05), and the averages of the preference for checks 9 and 10 werewere obviously obviously lower lower than than 5.55.5 ((pp << 0.05).0.05). The The data data reflected reflected that that check check 26 26 was was liked liked very very much much by by bothboth males males and and females, females, while while checkschecks 99 andand 10 were disliked disliked very very much much by by both both males males and and females. females.

FigureFigure 9.9. ThermogramThermogram of of one-sample one-sample t-testt-test test test data. data.

ChecksChecks 20, 20, 22, 22, and and 24 24 werewere likedliked by men and and disliked disliked by by women, women, while while check check 26 26was was liked liked very very muchmuch by by both both men men and and women. women. BasedBased on the comparison comparison of of the the characteristics characteristics of ofthese these four four plaids, plaids, it wasit was possible possible to to find find out out why why thethe womenwomen disliked checks checks 20, 20, 22, 22, and and 24. 24. As As can can be beseen seen from from the the above figure, for checks 26, 20, 22, and 24, all the values of light–dark y1 were more than 7, and all the above figure, for checks 26, 20, 22, and 24, all the values of light–dark y1 were more than 7, and all the values of delicate–rough y2 were more than 4. Thus, when the value of light–dark y1 was more than values of delicate–rough y2 were more than 4. Thus, when the value of light–dark y1 was more than 7, 7, women’s preference for the plaid would be prone to the influence of delicate–rough y2. For checks women’s preference for the plaid would be prone to the influence of delicate–rough y2. For checks 12 12 and 28, their values of y1 were more than 7, and their values of y2 were more than 4. Their values and 28, their values of y1 were more than 7, and their values of y2 were more than 4. Their values of of delicate–rough y2 were 5.5 < 6 and 8.2 > 8, respectively. These two plaids were not liked very much delicate–rough y were 5.5 < 6 and 8.2 > 8, respectively. These two plaids were not liked very much by both men and2 women. Therefore, this study found that for plaids liked by both men and women, by both men and women. Therefore, this study found that for plaids liked by both men and women, y1 will be more than 7 and y2 will be between 6 and 8. The values of simple–complex y3 for checks 9 y will be more than 7 and y will be between 6 and 8. The values of simple–complex y for checks 9 1and 10 were 7.9 and 4.4, respectively.2 If 5.5 was regarded as the mean value, y3 would not3 influence and 10 were 7.9 and 4.4, respectively. If 5.5 was regarded as the mean value, y would not influence the preference for checks 9 and 10. The values of light–dark y1 for checks 9 and 103 were more than 5 theand preference less than for6. The checks values 9 and of delicate–rough 10. The values y of2 were light–dark more thany1 for 4 and checks less 9than and 5. 10 The were other more plaids than 5 anddid less not than have 6. these The characteristics values of delicate–rough at the same ytime.2 were Hence, more for than plaids 4 and disliked less than very 5. much The other by both plaids men did notand have women, these y characteristics1 will be more atthan the 5 sameand less time. than Hence, 6, and for y2 plaidswill be dislikedmore than very 4 and much less by than both 5. menWhen and women,these characteristicsy1 will be more are than met, 5 y and3 will less not than influence 6, and preference.y2 will be more than 4 and less than 5. When these characteristicsAt the same are met,time,y Figures3 will not 8 and influence 9 show preference. that male preferences, female preferences, and overall preferencesAt the same are time,not significantly Figures8 and close9 show to thatred maleor green preferences, in the thermogram. female preferences, From the and specific overall preferencesexperimental are notdata, significantly none of the close patterns to red were or green either in strongly the thermogram. disliked nor From liked, the specificbecauseexperimental the mean data,values none of male of the preferences, patterns were fema eitherle preferences, strongly and disliked overall nor preferences liked, because varied the somewhere mean values from of 4.14 male to 7.05, 4.19 to 6.29, and 4.16 to 6.67 for all 28 patterns, while the total scale varied from 1 to 10.

3.5. Two-Way Mixed-Design ANOVA We employed two-way mixed-design analysis of variance model (ANOVA) to analyze the differences in preference for gingham, tartan, and windowpane between genders with SPSS. In statistics, a mixed design variance model is used to test the differences between two or more independent groups, and the participants are measured repeatedly. Therefore, in the ANOVA model of mixed design, one factor (fixed effect factor) is the test variable, and the other factor (random effect factor) is the within-subjects variable. A repeated measures design is used when multiple Sustainability 2018, 10, 3739 14 of 18 preferences, female preferences, and overall preferences varied somewhere from 4.14 to 7.05, 4.19 to 6.29, and 4.16 to 6.67 for all 28 patterns, while the total scale varied from 1 to 10.

3.5. Two-Way Mixed-Design ANOVA We employed two-way mixed-design analysis of variance model (ANOVA) to analyze the differences in preference for gingham, tartan, and windowpane between genders with SPSS. In statistics, a mixed design variance model is used to test the differences between two or more independent groups, and the participants are measured repeatedly. Therefore, in the ANOVA model of mixed design, one factor (fixed effect factor) is the test variable, and the other factor (random effect factor) is the within-subjects variable. A repeated measures design is used when multiple independent variables or measures exist in a data set, but all participants have been measured for each variable [21]. Taking gingham as an example, different ginham lattice patterns will have different preferences. This study wants to observe whether gender is another factor affecting preferences, which can be regarded as a two-way mixed-design ANOVA.

3.5.1. Gingham Two-way mixed-design ANOVA was employed to analyze the differences in preference for gingham between genders; the results of the analysis are shown in Table5. The test of internal effect of the respondents showed that the F value of the interaction effect was 1.079; p = 0.361, which did not reach a significant level. Hence, there was no need to conduct the simple main effect analysis. The main effect analysis of the two independent variables implied that the differences in design independent variables (gender) among the respondents did not reach a significant level. F = 0.402; p > 0.05, implying that gender was not related to the preferences for different plaids in gingham. The average of four samples of the within-subject design reached a significant level. The F value of the inter-group effect was 2.823; p < 0.05, which demonstrated that the preferences for different plaids in gingham varied. The ex-post analysis showed that, with respect to the pairwise comparison of four averages, the difference between check 2 and check 4 was significant (the difference, on average, was 0.929; p < 0.05). The pairwise comparison of the other plaids did not reach a significant level. The average for check 2 was 5.83, while that for check 4 was 4.90. It could be judged that, in the category of gingham, the preference for check 2 was significantly higher than that for check 4.

Table 5. Summary of one-way ANOVA of dependent samples for gingham plaids.

Source of Variation SS Df MS F p Gender (independent factor) 3.429 1 3.429 0.402 0.010 Different plaids (independent factor) 24.786 3 8.262 2.823 0.055 Gender × different plaids 9.476 3 3.159 1.079 0.361 Intra-group 541.553 80 Among respondents (Block) 341.048 40 8.526 Residual 200.505 40 5.013 Total 579.224 87 Note: “Gender × different plaids” indicates the interaction between gender and plaids.

In terms of gingham, the preference for check 2 was significantly higher than that for check 4. The values of light–dark y1 and delicate–rough y2 for check 2 were between 5 and 6, while the other plaids did not have these characteristics at the same time. The value of light–dark y1 for check 4 was between 5 and 6, while the value of delicate–rough y2 was between 2 and 3. Check 3 had the same characteristics. The values of simple–complex y3 for checks 3 and 4 were 3.5 and 5.3, respectively. Hence, both men and women preferred simple plaids. Among gingham plaids, for the plaids liked the most by the respondents, the value of light–dark y1 and the value of delicate–rough y2 were between 5 and 6. Sustainability 2018, 10, 3739 15 of 18

3.5.2. Tartan Two-way mixed-design ANOVA was employed to analyze the differences in preference for tartan between genders; the results of the analysis are shown in Table6. The test of internal effect of the respondents showed that the F value of the interaction effect was 0.720; p = 0.745, which did not reach a significant level. Hence, there was no need to conduct simple main effect analysis. The main effect analysis of the two independent variables implied that the differences in design independent variables (gender) among respondents did not reach a significant level. F = 0.809; p > 0.05, implying that gender was not related to the preferences of different tartan plaids. The average of four samples of the within-subject design reached a significant level. The F value of the inter-group effect was 4.639; p < 0.05, which demonstrated that the preferences for different tartan plaids varied. The ex-post analysis showed that, with respect to the pairwise comparison of 14 averages, the differences between check 9 and the other plaids were significant. The differences between check 10 and checks 5–9 and 13–15 via pairwise comparison reached a significant level. The differences between check 17 and checks 5–9 and 12–15 via pairwise comparison reached a significant level. The averages of checks 9, 10, and 17 were 4.167, 4.762, and 4.905. Compared with the averages of the other plaids, it could be judged that for tartan, the preferences for checks 9, 10, and 17 were significantly lower than those of the other plaids.

Table 6. Summary of one-way ANOVA of dependent samples for tartan plaids.

Source of Variation SS Df MS F p Gender (independent factor) 32.859 1 32.859 0.809 0.34 Different plaids (independent factor) 142.580 13 10.968 4.639 0.000 Gender × different plaids 22.117 13 1.701 0.720 0.745 Intra-group 2854.667 560 Among respondents (block) 1625.150 40 40.629 Residual 1229.517 520 2.364 Total 3052.223 587 Note: “Gender × different plaids” indicates the interaction between gender and plaids.

As shown in the above table, the values of simple–complex y3 for checks 9, 10, and 17 were 7.9, 4.4, and 4.2, respectively. If 5.5 was regarded as the median value of y3, y3 would have no influence on the preferences for checks 9, 10, and 17. Both values of light–dark y1 for checks 9 and 10 were between 5 and 6, and both values of delicate–rough y2 were between 4 and 5. Other plaids did not have these characteristics at the same time. The value of light–dark y1 for check 17 was between 6 and 7 and the value of delicate–rough y2 was between 5 and 6. Other plaids did not have these characteristics at the same time. Among the tartan plaids, there were two scenarios for the plaids disliked the most: (1) a value of light–dark y1 between 5 and 6 and a value of delicate–rough y2 between 4 and 5; and (2) a value of light–dark y1 between 6 and 7 and a value of delicate–rough y2 between 5 and 6. When these characteristics were met, y3 did not affect preference.

3.5.3. Windowpane Two-way mixed-design ANOVA was employed to analyze the differences in preference for windowpane between genders; the results of the analysis are shown in Table7. The test of internal effect of the respondents showed that the F value of the interaction effect was 2.095; p = 0.029, reaching a significant level. Hence, simple main effect analysis was carried out. According to the summary, in terms of the males, F(9, 360) = 0.900; p > 0.05. With respect to females, F(9, 360) = 3.756; p < 0.01. The two values were different from those of the main effect regardless of gender (F(9, 360) = 2.328; p < 0.05), implying that the respondents had different preferences for windowpane, especially among females. The differences in preferences for windowpane among females were significant, while those among males were the opposite. Sustainability 2018, 10, 3739 16 of 18

Table 7. Summary of mixed-design simple main effect variance analysis of windowpane plaids.

Simple Main Effect SS Df MS F p Different plaids (independent factor) Males 36.233 9 4.026 0.900 0.527 Females 128.424 9 14.269 3.756 0.000 Error (residual) 1489.143 360 4.137 Gender (independent factor) Check 19 0.214 1 0.214 0.028 0.868 Check 20 27.524 1 27.524 4.123 0.049 Check 21 4.667 1 4.667 0.591 0.446 Check 22 48.214 1 48.214 6.474 0.015 Check 23 0.095 1 0.095 0.017 0.897 Check 24 42.000 1 42.000 6.896 0.012 Check 25 0.095 1 0.095 0.016 0.901 Check 26 6.095 1 6.095 1.154 0.289 Check 27 0.857 1 0.857 0.137 0.714 Check 28 18.667 1 18.667 3.753 0.060 Error 2561.619 400 6.404

On the other hand, in terms of the independent variable of gender and 10 dependent factors, the F values (1, 400) were 0.028, 4.123, 0.591, 6.474, 0.017, 6.896, 0.016, 1.154, 0.137, and 3.753, respectively. Checks 20, 22, and 24 reached a significant level, while the other plaids did not. These values differed from the main effect F(1, 400) = 2.627; p > 0.05 with complete gender variables. The different genders showed varied preferences for checks 20, 22, and 24. Their averages by gender were as follows: check 20 (males: 6.095; females: 4.476), check 22 (males: 6.476; females: 4.333), and check 24 (males: 6.238; females: 4.238). The data showed that the men’s preferences for these three patterns were significantly higher than those of the women, while the differences in the other plaids were not significant. The results were consistent with the results of the differences in the preferences of the 28 plaids between gender verified by the independent sample t-test. As shown in the above table, the values of dimple–complex y3 for checks 20, 22, and 24 were 5.2, 5.9, and 3, respectively. If 5.5 was regarded as the mean value of y3, y3 could not be used to describe the preferences of checks 20, 22, and 24. However, all the values of light–dark y1 for checks 20, 22, and 24 were more than 7, and all the values of delicate–rough y2 were less than 4. The other plaids did not have these characteristics at the same time. Hence, for a plaid liked by males and disliked by females, y1 will be greater than 7, and y2 will be less than 4. When y1 and y2 meet these characteristics, y3 will not affect preference.

4. Conclusions In this paper, a psychophysical experiment was implemented to investigate the factors affecting pattern preference for black-and-white shirts. Firstly, 28 different representative patterns of plaid shirts were selected by five fashion designers together from 190 different black-and-white plaid shirts from eight fast fashion brands, which were then classified into three categories: gingham, tartans, and windowpane. Secondly, based on these patterns, 28 male and female shirts were simulated in three dimensions and presented on a calibrated computer display. The simulations were assessed by 42 observers (consisting of 21 males and 21 females) in terms of four semantic scales, including light–dark, delicate–rough, simple–complex, and like–dislike. The experimental results revealed:

• There was no significant difference in the pattern preferences between females and males for 89.29% of the black-and-white plaid shirts. Sustainability 2018, 10, 3739 17 of 18

• There was a significant difference for tartan and windowpane preferences between females and males, and the preference degree of the males was more than that of the females. • There was no significant difference in the preference for gingham.

Furthermore, this study also demonstrated the formulation between the four semantic scales and three pattern features (including the percentage of black region, the size of the minimum repeat unit, and the descriptor of the pattern complexity). The preference degree of the females was easily affected by the semantic scale of delicate–rough. These findings could be used to develop a more robust and comprehensive theory of object pattern preference for black-and-white plaid shirts and provide a reference for pattern design. From the above results, we can provide suggestions for clothing enterprises to avoid producing black-and-white plaid shirts that men and women obviously do not like. At the same time, the same or different preferences of men and women for black-and-white plaid shirts can provide some theoretical basis for fashion designers to design black-and-white plaid shirts that are liked for longer time. A more comprehensive theory of pattern preference is not only an effective tool to solve the inventory problem of plaid shirts in the garment industry but also an important theoretical basis for the “sustainable design” of garments, which is of great significance to the sustainable development of the garment industry. As managers of textile enterprises and fashion designers, they should make use of our research results so that they can promote the sustainable development of the garment industry. The limit of this study is that all the 42 observers are young Chinese people, 30 of whom majored in design (at different educational levels). This is a very specific consumer group and cannot be directly used to translate the results into a universal consumer preference of patterns. However, the proposed investigation method can be used to analyze the patterns of preference of consumers of different ages and nationalities. A limitation of this paper is that it focused on black-and-white plaid shirts and did not include solid and multi-color plaid shirts. Such color combinations in the fabric pattern may have a great influence on pattern preferences, and this effect should be studied in the future.

Author Contributions: All authors contributed to the paper. Q.J., J.Z., and C.Y. collected and organized data; Q.J. wrote the manuscript with the supervision from L.-C.C. and J.Z.; and J.Z. acted as a corresponding author. Funding: This research received no external funding. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Linkshop. Exclusive: Youngor and Other Eight Clothing Five-Year Inventory Structure Analysis. Available online: http://www.linkshop.com.cn/web/archives/2017/368169.shtml (accessed on 13 October 2018). 2. Bruce, V. Visual Perception; Psychology Press: , UK, 1996. 3. Jiang, X.F. Positive-Negative Emotional Categorization of Clothing Color Based on Brightness. Engineering 2015, 5, 189–194. [CrossRef] 4. Hsu, M.Y. Colour preference for Taiwanese floral pattern fabrics. Color Res. Appl. 2016, 41, 43–55. [CrossRef] 5. Kim, E.Y. Emotion-Based Textile Indexing Using Colors and Texture. In Proceedings of the International Conference on Fuzzy Systems and Knowledge Discovery, Changsha, China, 27–29 August 2005; pp. 1077–1080. 6. Kim, S.J. Emotion-Based Textile Indexing Using Colors. In Proceedings of the International Symposium on Visual Computing, Lake Tahoe, NV, USA, 6–8 November 2006; pp. 9–18. 7. Na, Y.K. Emotion-Based Textile Indexing Using Neural Networks. In Proceedings of the International Conference on Human-Computer Interaction, Beijing, China, 22–27 July 2007; pp. 349–357. 8. Lee, S.R. Effect of Motif Designs on Preferences and Image Perception. Res. J. Costume Cult. 2007, 15, 193–202. 9. Na, Y.K. Emotion recognition using color and pattern in textile images. In Proceedings of the IEEE Conference on Cybernetics and Intelligent Systems, Chengdu, China, 21–24 September 2008; pp. 1100–1105. Sustainability 2018, 10, 3739 18 of 18

10. Chio, S.K. The Image Evaluation according to Checked Pattern Variable of Casual Shirts-Focus on Tone-in-Tone Coloration. J. Korean Soc. Cloth. Text. 2011, 35, 867–876. [CrossRef] 11. Davis, F. A Study Relating Computational Textile Textural Expression to Emotion. In Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA’15), Seoul, Korea, 18–23 April 2015; pp. 1977–1982. [CrossRef] 12. Davis, F. The Textility of Emotion: A Study Relating Computational Textile Textural Expression to Emotion. In Proceedings of the 2015 ACM SIGCHI Conference on Creativity and Cognition (C&C’15), , UK, 22–25 June 2015; pp. 23–32. 13. Homlong, S. The Language of Textiles Description and Judgement on Textile Pattern Composition. Ph.D. Thesis, Uppsala University, Uppsala, Sweden, 2006; p. 90. 14. Barber, E.J.W. Prehistoric Textiles: The development of Cloth in the Neolithic and Bronze Ages, with Special Reference to the Aegean; Princeton University Press: Princeton, NJ, USA, 1992. 15. Banks, J. Tartan: Romancing the Plaid; Rizzoli: New York, NY, USA, 2007; ISBN 978-0-8478-2982-8. 16. Fortson, B.W. Indo-European Language and Culture: An Introduction; Blackwell Publishing: , UK, 2004; ISBN 1-4051-0316-7. 17. Lee, C. Style Guide: Know Your Shirt Fabric Patterns [EB/OL]. Available online: https://www.alexander- west.com/styleguide/?p=288.2009-8-3 (accessed on 13 October 2018). 18. Cleveland, W.S. Robust locally weighted regression and smoothing scatterplots. J. Am. Stat. Assoc. 1979, 74, 829–836. [CrossRef] 19. Cleveland, W.S. Locally-weighted regression: An approach to regression analysis by local fitting. J. Am. Stat. Assoc. 1988, 83, 596–610. [CrossRef] 20. Zhang, J. Automatic inspection of yarn-dyed fabric density by mathematical statistics of sub-images. J. Text. Inst. 2014, 106, 823–834. [CrossRef] 21. Field, A. Discovering Statistics Using SPSS, 3rd ed.; Sage: Thousand Oaks, CA, USA, 2009.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).