longitudinal comparison of finnish and us online …...the next stage in the buying process is...

devices almost throughout their entirelives, which makes them highly literatein online activities. This paper is focusedon the online shopping behaviour of akey segment of two of the world’s mostadvanced IT nations: Finnish and USuniversity students.2 Although there hasbeen ample discussion and criticism

INTRODUCTIONThe use of the internet has rapidlygrown from its early years, thanks to theso-called ‘Net Generation’. The NetGeneration is said to consist of youngpeople and adolescents born between1977 and 1997.1 These people have beenusing the latest information technology

336 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 4, 336–356 � Palgrave Macmillan Ltd 1479-1862/06 $30.00

Longitudinal comparison ofFinnish and US online shoppingbehaviour among universitystudents: The five-stage buyingdecision processReceived (in revised form): 10th August, 2006

Charles Comegysis the Ciejek Chair of Business and is Associate Professor of Marketing in the Girard School of Business & InternationalCommerce at Merrimack College, USA. Dr Comegys’ research interests include research methodologies and internet purchasebehaviour. His articles appear in numerous publications including the Journal of Marketing Theory and Practice, PracticalAssessment, Research & Evaluation, Journal of Internet Commerce, Journal of Business Education, Journal of Small BusinessManagement, Journal of the Academy of Business Education, Management Research, Atlanta Economic Review, OperationsManagement Review, and Marketing News. He has served on the Board of Directors of the American Marketing Association,the Marketing Research Association, and currently is on the Board of the Marketing Educators’ Association.

Mika Hannulaholds MSc (Eng) and DrTech in Industrial Management and Engineering. Professor Hannula is currently the Head of theInstitute of Business Information Management and the Director of the Degree Program of Information and KnowledgeManagement at Tampere University of Technology, Finland.

Jaani Vaisanenreceived his Master’s Degree in Statistics from the University of Tampere, Finland. He currently works as a researcher in theInstitute of Business Information Management at Tampere University of Technology, where he is also completing his doctoralstudies. His current research interests are related to electronic commerce and especially in the use of search engines in themarketing field.

Abstract This study investigates the online purchase behaviour of a key segment ofthe population, the ‘Net Generation’ university-aged student, from two of the world’smost advanced IT nations with the greatest potential in e-commerce: Finland and theUSA. Information about online shopping behaviour in 2002 is compared with 2004/2005for the two countries. The research also answers the question whether online shoppersfrom the two countries approach the consumer buying decision process differently overtime. The results provide useful guidance to consumer e-marketing companies.

Charles Comegys, PhDGirard School of Business& International CommerceMerrimack College315 Turnpike StreetNorth Andover, MA, USA

Tel: +1 (978) 837 5409Fax: +1 (978) 837 5086Email:[email protected]

that even with all these new innovationsand abilities, the internet has notchanged the basic buying patterns ofconsumers.6

These mixed opinions are to thebenefit of the present study, which aims tocombine the aspects of flexibility withfixed purchasing patterns. Thus, the studyassumes that buying behaviour follows thefive-stage model with some modificationsfor the online situation, and considers thechanges that have occurred in two yearswithin the framework of the model.Assuming that the model is a ‘closedspace,’ which completely covers everyaspect of purchase, it is possible to drawconclusions on which parts of the modelare more relevant to customers thanothers.

BACKGROUNDThe five-stage buying decision processmodel is a widely used tool formarketers to gain a better understandingabout their customers and theirbehaviour.7 The idea of the model is thatwhen a customer purchases an item, thepurchase event is a forward-movingprocess, which begins long before theactual purchase and continues even afterthe purchase is made. As the nameimplies, there are five different stages inthe process, which are need recognition,information search, evaluation ofalternatives, purchase decision, andpostpurchase behaviour (see Figure 1).Following is a review of each of thesestages in the consumer buying process.

Need recognition

The buying process begins with needrecognition (sometimes referred to asproblem recognition), where the buyersenses a difference between their actualstate and a state they desire. This needcan be triggered by either an internal

concerning the use of university studentsas test subjects, they fit well for thisstudy as university students in the targetcountries are typically extensive users ofthe internet. This study is not intendedto gather information pertaining to theproportion of the population that use theinternet, but to understand the buyerdecision process of those that do. Thus,university students, although not a trulyrepresentative sample, do reflect thebehaviour of heavy online users.Information about this segment has beengathered at two points in time: in 2002and 2004 in Finland, and in 2002 and2005 in the USA. The focal point of thisstudy is to compare the differences andrelations between the two samples, andthe two nations. Among different modelsof consumer behaviour, perhaps the mostwell known is the five-stage model,3

which describes the consumer purchaseevent as a process which starts longbefore the actual purchase isconsummated and continues after thepurchase. This model is the backbone ofthe present research, which depicts thechanges in the process and the factorsthat underlie them between the samples.The results give marketers an insight intohow the Finnish and US onlinemarketplace has changed in the past twoyears in the customer’s eyes, along withsome implications of where it might bein another two years.

Electronic commerce offers manyadvantages over traditional commercethat are useful to both marketers andcustomers. For example, customerrelationship management (CRM) helpsthe company to gather information andmaintain relations to the customer, whoin turn will get precisely the informationthey need.4 This flexibility is said to beone of the internet’s greatest values formarketers when compared withconventional methods of commerce.5 Onthe other hand, there are still opinions

� Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 4, 336–356 Journal of Targeting, Measurement and Analysis for Marketing 337

Finnish and US online shopping behaviour among university students

have suggested that higher purchasefrequencies stimulate repeat purchasingrather than variety-seeking behaviour,which gives e-marketers a good reasonto make the threshold for shopping aslow as possible.

Information search

The next stage in the buying process isinformation search, where the consumeruses different channels to gatherinformation about available products,which might fulfil the needs discussedabove. Kotler15 defines two levels ofarousal during the information searchphase. In the milder state, heightenedattention, the consumer just becomesmore familiar with the different productsthat might be suitable for them. In thisstate, the consumer only pays attentionto advertisements and conversationsabout the subject. In the next state,active information search, the consumeractively engages these conversations andsearches for information about differentbrands, models, etc.

Kotler16 also defines four methods bywhich consumers receive product/serviceinformation. Personal sources includefamily and friends. Commercial sourcesinclude advertising and salespersons.Public sources include mass media andconsumer-rating organisations.Experimental sources include examiningand using the product itself. Most of theinformation comes from the commercialsources, but according to Dubois,17 the

(hunger, thirst) or external (passingthrough McDonald’s stimulates hunger)stimuli.8

Other factors besides differences in thebuyer’s actual and desired state influenceneed recognition as well. One of thedirectly observable influences isdemographic factors, including age, sex,income, race, education, household size,and marital status.9 There are alsoinferred influences which affect needrecognition. Psychological factors play amajor role in these inferred influences.

Of these psychological factors,motivation is the basis of all consumerbehaviour.10 Although there is no generalagreement among psychologists on thebest way to classify consumer needs asfar as motivation goes, Kinnear andBernhardt11 divide the motivational needsinto physiological and psychologicalneeds. The former include, for example,the need for food and shelter, while thelatter are generated by one’s socialenvironment.

Perception reflects how the customerssee themselves and their surroundings,which in turn affects not only the needrecognition phase, but the other phasesas well.12 Depending on the consumer’sneeds and perceptions, need recognitionmay take different forms. In addition topurchasing something totally new orreplacing a broken/outdated item,consumers may replace a product thatfills their needs completely with another.This phenomenon is called varietyseeking.13 Van Trijp, Hoyer and Inman14


Comegys, Hannula and Vaisanen

Figure 1: Five stage buying decision process model

Need

Recognition

Information

Search

Evaluation Purchase

Decision

Postpurchase

Behavior

increase the amount of searching, as thecomprehension of the results would beno problem. Knowledge uncertainty onthe other hand seemed to lower theamount of search. Insights from Urbanyet al. disagree somewhat with Alba andHutchinson. The conclusions derivedfrom both of these studies is that acertain amount of knowledge about thedesired product is necessary to increaseactive information searching, eventhough potentially improved knowledgemay reduce it.

Of all the possible products and brandsthat might satisfy a customer’s needs,only a handful will be brought to theconsumer’s attention. Kotler22 definesfour sets of alternatives from which thefinal purchase decision will be made.The total set includes all the possiblebrands available to the customer. Derivedfrom the total set, is an awareness set,which includes the brands the customerhas come to know. Brands that meet thebuyer’s purchasing criteria are taken fromthe awareness set to form theconsideration set. As more informationabout the products of the considerationset is gathered and evaluated, the finalpurchase decision will be made from thechoice set.

Peter and Olson23 define a similarprocess, which divides all the brands inthe product class into familiar andunknown brands. From unknown brands,those that are found accidentally andthose found through intentional searchmake it to the choice set. From familiarbrands, only those that are activated frommemory (evoked set) make it to thechoice set. Among the familiar brands,there may be some that once have beenfamiliar to the customer, but are lost inmemory. Of the evoked set, brands thatare liked more and thought more typicalto the product class tend to emerge intothe choice set.24

Narayana and Markin25 have extended

most effective information comes frompersonal sources.

There are also differences ininformation search with different levels ofexpertise. Alba and Hutchinson18 reportthat experts are able to search moreefficiently and therefore have moreknowledge of available products. Theypoint out a common case, however,where experts do less actual searchingbecause they have already gatheredprevious information about the productsin their memory (internal search). WhenBeatty and Smith19 propose that productclass knowledge is negatively associatedwith the search effort, it can beconcluded that often those withintermediate knowledge of the desiredproduct do most of the searching,although its efficiency might not rival theexperts’. Considering the onlineenvironment, expertise about the productand expertise about the search processmust be separated. As previously stated,those with excellent knowledge aboutthe product might reduce the amount ofinformation search, and better knowledgeabout the online search methods canactually do the same, as the searchprocedure becomes more effective. Inaddition, better knowledge about thedesired product can improve the qualityof the consideration set (discussed later inthis paper), thus also reducing the searchtime and improving the searchefficiency.20

Another factor affecting of the amountand type of information search is theperceived risk (uncertainty) involvedwith the purchase. In their study,Urbany, Dickson and Wilkie21 divideuncertainty into two categories.Knowledge uncertainty includesuncertainty regarding information aboutalternatives, and choice uncertaintymeans uncertainty about whichalternative to choose. Urbany et al. notedthat choice uncertainty appeared to



Evaluation of alternatives

Consumers tend to set rules, or attributecut-offs for the products in their choiceset. These are the minimum acceptablelevels that an alternative must possess inorder to be considered as the finalpurchase. Huber and Klein28 haveshowed two characteristics to thesecut-offs. The first one states that whenthe reliability of the attribute thecustomer is considering is high (theinformation comes from a reliable source,at least in the customer’s eyes), thecut-offs on that attribute are more severethan when reliability is low. The secondcharacteristic is that when there is apositive correlation between twoattributes (for example, low rent in ahigh-quality apartment), the cut-offs onthose attributes are more severe thanthey would be if the correlation wasnegative or of equal magnitude. Withrespect to price, these cut-offs mayhowever vary between online and offlinetransactions. Bhatnagar and Ghose29

found that price was not one of themajor factors for online shoppers whenthey evaluate different alternatives. Thismay be because consumers might thinkthe web prices are broadly similar and sothey do not need to pay much attentionto the price tag.

As consumers reduce their alternativesto the choice set, they have first acquiredinformation about the products and thencompared and evaluated them. As noconsumer has unlimited resources (time,in this case) at their disposal, a line mustbe drawn as to when to stop theevaluation process and make the actualpurchase decision. Hauser, Urban andWeinberg30 propose that consumersallocate their time in such a fashion thatwhen the trouble of getting extrainformation about the product outweighsthe value of the additional informationitself, information search and evaluationstop. More precisely, consumers try to

the awareness set to consist of threedifferent subsets. They define the evokedset as a set of products of which theconsumer has a positive opinion andfrom which they are likely to make theirpurchase. Inert set includes products forwhich the consumer has neither negativenor positive opinion. The consumer maybe aware of such products but not haveformed an opinion, possibly due to lackof knowledge. The final set is called theinept set, which includes the products forwhich the consumer has formed anegative opinion, and therefore theseproducts will not to be purchased.

No matter how one conceptualisesthese sets, the online environment givesthe consumers a variety of different toolsfor information search to form these sets.Many companies have utilised so-calledrecommendation agents, which areinteractive tools that assist the consumersin their screening of alternatives based onthe information they have provided.Haubl and Trifts26 have gatheredevidence that these recommendationagents ‘reduce the number of alternativesfor which detailed product information isviewed’, thus reducing the search time ofthe consumers. One of the mostcommon online search methods isbrowsing. Rowley27 asserts that browsingcan be either general or purposive.Purposive browsing occurs when theconsumer has fairly specific requirementsof the needed product, as opposed togeneral browsing which is used to keepup to date with the latest knowledgeconcerning the product range. Rowleydefines three distinctive purposes wherebrowsing is preferred to search engines:(1) the search objective cannot be clearlydefined; (2) the cognitive burden (ie theknowledge the consumer needs to havein order to do the searching) makesbrowsing easier; (3) the consumer’ssystem interface encourages browsingover other methods of searching.



technology has its advantages. Bykeeping record of individual customers’purchases and personal information,companies can now target theiradvertising accurately to each individualcustomer,33 thus having a chance toexploit the concept of conceptual fluencydiscussed above.

Purchase decision

After the evaluation stage, the consumerhas ranked the items in the choice set insome sort of order but not always willthe number one (if there is a numberone assigned) item be chosen. There aretwo factors that come between theevaluation and purchase decision stages.34

First there are the attitudes of others,where best friends or communitypressure may change a consumer’spreference ranking for a certain brandeven if they intended to buy a differentone. Secondly there might be someunexpected situational factors that affectthe purchase decision. The price of theproduct may have suddenly gone up, orsome other purchase becomes moreurgent. As online shopping usually occursin a more private environment, otherpeople’s influences are reduced at thepoint of purchase. Even when aconsumer has decided the exact productthey are going to buy, there are still afew purchase sub-decisions to be made.35

These sub-decisions include price range,point of sale, time of purchase, volumeof purchase, and method of payment.

In spite of all the theories on needrecognition, information search, andevaluation as a fundamental basis for thepurchase decision itself, there aresituations where none of these processesapply. This phenomenon is calledimpulse purchasing. Impulse buyingoccurs when the consumer purchases anitem based solely on their impulses andemotions. Baumeister36 submits that there

maximise the equation �svs(ts) � v0(t0),where vs(ts) is the value of time ts spentgaining information about the product,and v0(t0) is the value of time t0 spent inother activities. As online evaluations canbe done from a personal computer,saving the trouble of travelling fromplace to place, the equation changes sothat we have an increase in t0 and adecrease in ts.

Still, consumers do not make theirdecisions based solely on the informationabout different brands, but also includehow easy it is to process the informationabout the brand. A brand that is easilyrecognised by its physical characteristics isperpetually fluent, while a brand thatcomes to mind easily is conceptuallyfluent. Lee and Labroo31 demonstratedthat in their experiments that a brandbecomes more conceptually fluent if theconsumer has been exposed to relatedproducts just before they come intocontact with the target product.Following their example, a billboardadvertising ketchup produced moreconceptual fluency, when it was precededwith an advertisement for french friescompared with a washing detergent, thusthe message from the billboard was easierto handle. In addition, with highconceptual fluency conditions, the brandevaluations tend to be more favourable.It is important to note, however, that ifconceptual fluency creates negativeassociations, its effect will be negative.For example, the aforementionedadvertisement of french fries may yieldnegative associations to the people thatdislike them, which leads to lessfavourable evaluations of the ketchupad.32

This produces a great challenge formarketers in an online environment, ascustomers may be faced with a plethoraof different advertisements, prior to theone that is needed to get their attention.This is where the new customisation



and (postpurchase) satisfaction. Typically,loyal customers tend also to be satisfiedbut, according to Oliver, satisfaction doesnot necessarily produce loyalty. However,studies on several areas of consumerresearch such as Auh and Johnson,42 andBall, Coelho and Machas43 emphasisehow satisfaction indeed does produceloyalty.

The importance of satisfaction is asrelevant in an online environment as it isin an offline world. Evanschitzky et al.44

have replicated a study by Szymanski andHise45 and have concluded that the singlemost important factor forming satisfactionin electronic commerce is the shoppingconvenience, which is a typicaladvantage of online shopping comparedwith conventional shopping. Typically, ifa customer is not satisfied with thepurchase, there is a chance that they willcomplain about the product/service. Asthe channels involving traditional andelectronic commerce differ, it is notsurprising that there are noted differencesin the complaining threshold and thedegree of dissatisfaction of the purchasein an online vs an offline environment asreported by Cho et al.46

In addition to satisfaction and loyalty,important issues in postpurchase actionsinclude brand preference and repurchaseintentions and how they affect eachother. Hellier et al.47 have proposed amodel where all these aspects arediscussed and a number of hypotheses arederived. The following conclusions weremade: loyalty has a positive effect onbrand preference; satisfaction has apositive effect on loyalty; and brandpreference and the strength of brandpreference have a positive effect onrepurchase intentions. If theaforementioned assumption byEvanschitzky et al. concerning the basicimportance of satisfaction and itssubsequent effects hold, it may beassumed that the conclusions offered by

are some genuinely irresistible impulses,mainly physiological, from which there isno escape, although they do notnecessarily lead to purchase. This authorcontinues to define three characteristics,which form consumer’s self-control. Tofail one of them might result inimpulsive shopping given the rightcircumstances.

The first of these characteristics are thestandards the individual has set forthemselves. If consumers have set certaingoals and norms and they know exactlywhat they want, they are less likely toact on impulse. This also reduces thevulnerability to sales personnel andadvertisers. The second characteristic ismonitoring. People who keep bettertrack on their relevant behaviour are lessprone to lose their self-control. Thirdand most important, is the consumer’scapacity to change. Even if the twoprevious ingredients fail, there must besomething inside the consumer that iswilling to make the change and purchasethe certain item their impulses suggest.37

Similar results were found by Sojka andGiese.38

Postpurchase behaviour

The purchase process continues evenafter the actual purchase is made. Ifmarketers and retailers want thecustomers to come back, they mustunderstand their behaviour after thepurchase as well. Postpurchase behaviourcan be divided into two subgroups:postpurchase satisfaction and postpurchaseactions.39 When it comes to postpurchasesatisfaction, there is evidence offered byMittal and Kamakura40 that consumerswith different characteristics havedifferent thresholds when it comes toloyalty towards the store even if theywere dissatisfied with their purchase.

Oliver41 has further discussed theconcepts of loyalty (postpurchase action)



this conclusion, the following generalhypothesis is derived:

H1: Gender influences the onlinebuying process.

The first hypothesis is adjusted to fit thefive-stage model. It is proposed that ifthe whole online buying process isconsidered to consist of five stages, itwould be reasonable to test theassumption that gender influences all ofthe stages and not just some of them.Hence, the following hypotheses:

H1a: Gender influences the needrecognition phase of the onlinebuying process.

H1b: Gender influences theinformation search phase of theonline buying process.

H1c: Gender influences the evaluationphase of the online buyingprocess.

H1d: Gender influences the purchasedecision phase of the onlinebuying process.

H1e: Gender influences thepostpurchase phase of the onlinebuying process.

Differences in consumer online buyingbehaviour between 2002 and 2004/2005

Once the connection between gender andthe purchase process stages is established,an understanding of how the samplingperiod affects these stages is undertaken.There is also evidence that electroniccommerce is growing rapidly in Finland53

and in the USA.54 It would therefore bereasonable to assume that as electroniccommerce is growing, online buyingbehaviour might also be changing. Thisleads to the next general hypothesis:

H2: The time of sampling affects theonline buying process.

Hellier et al. are also valid in an onlineenvironment.

AREAS OF STUDY ANDPROBLEM CONCEPTUALISATION

Gender

The first hypotheses discuss thedifferences between online shoppingpatterns of men and women. The basisof this research area is the work reportedby Palanisamy,48 and Rodgers andHarris.49 According to Palasinamy, genderinfluences the relationship betweenonline advertisement interpretation andthe attitude toward the advertisementwhich in turn reflects the attitudetowards the company. Thus, if theinterpretation of an online advertisementis the same for both genders, but theattitude towards the ad is more positiveby one gender, this would impact theperception of the entire buying process.Rodgers and Harris support this theorywith their results. They claim thatwomen are emotionally less gratified andhave less satisfaction from onlineshopping than men. Men would also bemore trusting than women and thinkthat online shopping is more convenient.

Dholakia and Chiang50 suggest thatmen are often considered to be moretechnologically literate than women, afact that is reflected especially instereotypes that portray the onlineshopping event. There has been evidencefound by Garbarino and Strahilevitz51 tosupport these stereotypes. Even thoughwomen have been closing the gendergap considering internet usage, it is stillthought that women perceive higherrisks than men in online purchasing inboth probabilities and consequences.52

All the evidence above seems toindicate that there is indeed a differencein the way men and women perceivethe entire online buying process. From



following general hypothesis isproposed:

H3: There is a positive relationshipbetween the consumer’s affiliationto the different stages of thebuying process and the onlinepurchase volume.

Next, it is assumed that the cognitiveabsorption proposed by Shang et al.influences all of the components in thebuying behaviour process. Therefore, thegeneral H3 hypothesis is broken downinto five parts for each individualpurchase process stage, where it is statedthat those who use the internet more forneed recognition etc. also purchase moreonline:

H3a: There is a positive relationshipbetween the amount of consumeronline need recognition and theonline purchase volume.

H3b: There is a positive relationshipbetween the amount of consumeronline information search and theonline purchase volume.

H3c: There is a positive relationshipbetween the amount of consumeronline evaluation and the onlinepurchase volume.

H3d: There is a positive relationshipbetween the amount of consumeronline purchase decisions and theonline purchase volume.

H3e: There is a positive relationshipbetween the amount of consumeronline postpurchase satisfactionand the online purchase volume.

If a customer is satisfied with earlierpurchases, it should produce store loyalty,which means more repurchases.57,58 If acustomer has made several purchasesfrom a particular website, they may havedeveloped a routine pattern forpurchasing online, or at least the

To examine this general hypothesis morethoroughly, it is separated into fivesmaller sub-hypotheses:

H2a: The time of sampling affects theamount of need recognition.

H2b: The time of sampling affects theamount of information search.

H2c: The time of sampling affects theamount of evaluation.

H2d: The time of sampling affects theamount of purchase decisions.

In this study, postpurchase behaviour ismeasured in terms of loyalty (thelikelihood of repurchase). As Oliver55 hasstated, loyal customers tend also to besatisfied, so the term postpurchasesatisfaction is used here.

H2e: The time of sampling affects theamount of postpurchasesatisfaction.

The effects of consumer onlinebehaviour patterns on purchase volume

The next point of interest is therelationship between purchase volumeand different aspects of consumerbehaviour. The following is proposed:the stronger (more positive) theconsumer’s affiliation to any givenphase of the purchase process in anonline environment, the higher thepurchase volume will be. Evidence byShang, Chen and Shen56 suggests thatonline shopping is not so much agoal-oriented activity rather than aresult from cognitive absorptionexperiences from the web. So,according to theory by Shang et al., ifa consumer discovers a product on theinternet and uses it to obtaininformation about the said product, thecognitive absorption obtained duringthe process from the web may wellresult in online shopping. Therefore the



online information search and thepossible changes in purchasevolume.

H4c: There is a positive relationshipbetween the amount of consumeronline evaluation and the possiblechanges in purchase volume.

H4d: There is a positive relationshipbetween the amount of consumeronline purchase decisions and thepossible changes in purchasevolume.

H4e: There is a positive relationshipbetween the amount of consumeronline postpurchase satisfactionand the possible changes inpurchase volume.

METHODOLOGYThe sampling plan required that amulti-page questionnaire beself-administered to randomly selectedsamples of students. The four separateand independent samples used in thisstudy were gathered in Finland and theUSA during 2002 and again in2004/2005. The questionnaires used inthe USA and Finland were identicalexcept the Finnish students responded tothe questionnaire in Finnish. The contentof each question was carefully translatedin order to match the English version. Inaddition, the Finnish students respondedto the expenditure questions in terms oftheir local currency (Finnish markka).The Finnish currency responses wereconverted to US dollars using theexchange rate for the Finnish markka atthe time of the data gathering.

The first two samples consisted of datagathered in Finland and the USA during2002. The first sample consisted of 152students attending a college located innortheast USA. The second sample of194 was drawn from students attending auniversity located in Tampere, Finland.The third sample gathered data in 2004

threshold of purchasing online should belower from that particular site. It istherefore concluded that customers whoare likely to make a repurchase may havelarger online purchase volume than thosewho have lower likelihood forrepurchases, which leads to a followinghypothesis:

H3e: There is a positive relationshipbetween the amount of consumeronline postpurchase satisfactionand the online purchase volume.

The effects of consumer onlinebehaviour patterns to purchasevolume change

The former hypothesis combines certainparts of the H2 and H3 hypotheses. TheH2 and H3 hypotheses questioned howthe different behavioural characteristicsaffected the possible change in purchasevolume between the sampling periods.That is to say that the possible increase inpurchase volume might be greater amongthose with higher amount of online needrecognition, etc. So if H2 and H3 aresupported by the evidence, the followinghypotheses may be proposed:

H4: There is a positive relationshipbetween the consumer’s affiliationto the different stages in buyingprocess and the possible changesin online purchase volume.

As with the preceding hypotheses, thisgeneral hypothesis may be separated intofive sub-hypotheses:

H4a: There is a positive relationshipbetween the amount of consumeronline need recognition and thepossible changes in onlinepurchase volume.

H4b: There is a positive relationshipbetween the amount of consumer



Postpurchase actions have been studiedthrough store loyalty. The respondentswere asked whether they were likely tomake a repurchase from an internet site,assuming they were happy with theirearlier purchase.

All of these questions were measuredon a three-point scale. The mainchallenge of this study was to determinethe differences between the portions ofthe samples in relation to the explanatoryvariable (usually one of the buyingprocess stages). When dealing withcategorical variables, such as the 1-2-3scale, chi-square testing was used. Alongwith the chi-square statistic and acorresponding p-value, the contingencycoefficient was also computed to find outthe strengths of the possible relationships.

When confronted with scale variables,such as purchase volume, the differenceswere sorted with the analysis of variance(ANOVA). If the homogeneity ofvariances held true, the regular F-statisticto calculate the p-value was used, but ifthis assumption did not hold, the p-valuewas calculated by using theWelch/Brown-Forsythe statistics.60,61 AsANOVA provides only results for samplemean values to be equal or not,regression analysis with dummy variableswas used to find out which of thesample means differed from each other,along with their direction, magnitudeand 95 per cent confidence limits. Formore information on the statisticalanalyses carried out in this study, see egGreenwood and Nikulin,62 Christensen,63

and Draper and Smith.64

RESULTS

H1: Gender influences the online buying

process

The first hypothesis considered thatgender affected the purchase process and,in particular, it was postulated that

from 194 students attending two differentuniversities in Tampere, Finland. Thefinal sampling in 2005 consisted of 162students at two institutions, one smallprivate college and one large publicuniversity located in northeast USA.

In this study, the first of the fivestages, need recognition, means eitherneeds that are recognised directly onlineas external stimulus, or possible solutions(found online) to already existing needs.In the latter case the phase actuallymoves from need recognition toinformation search as the product shiftsfrom the total set to awareness set, asdefined by Kotler.59 In either case, thiswas regarded as need recognition andwas measured by the question ‘Whileonline, I have discovered productsand/or services that I am interested in.’

Information search means informationacquired only from the internet throughactive searching. Therefore personalsources like instant messaging services oremail do not apply, leaving commercial(advertising) or public (third-partywebsites) sources. The correspondingquestion is ‘I have searched websites forinformation about products and/orservices I am interested in.’

Evaluation has been measured as activeevaluation. Active evaluation means thatthe evaluation process is happeningthrough the internet with a possibleintent to make a purchase at some point.Measurement of evaluation was carriedout with the question ‘I use websites toevaluate different services and/orproducts, brands, prices, features, andtheir availability.’

Perhaps the most crucial factor in thewhole online buying process is thepurchase decision itself. By definition, theonline purchase decision means that theactual intent to purchase online has indeedoccurred online. The question ‘I make myfinal product and/or service purchasedecisions while online’ measures this.



relations between gender and thedifferent purchase process stages areabout equally strong. Although therelation between gender and informationsearch was statistically significant, it isclearly on the weaker side (comparedwith need recognition, evaluation, andpurchase decision). In all of these cases,men were found to be more active thanwomen.

The US sample showed that none ofthe purchase process stages differed whenmen and women were compared,providing strong evidence against thehypotheses. In the USA, the 2002sample yielded similar results, showingthat the online purchase process hasremained approximately the same withboth men and women over the last fewyears. In Finland, however, the earliersample showed that the purchase processwas much more similar between menand women two years ago than it isnow. Should this trend continue inFinland, men would continue to bemore active in the purchase process inthe following years, a fact that companiesand marketers need to acknowledge inorder to get their messages to the desiredaudience. Given the trend in the USA,this is something that should not be ofconcern when targeting the US-basedaudience.

In summary, the Finnish results give

gender influenced all of the five stages ofpurchase behaviour when these stageswere observed separately. Table 1 displaysthe chi-square and p-values along withcontingency coefficients where necessaryfor all of the five stages of purchaseprocess. For most up-to-date results, onlythe 2004/2005 samples were used.

Table 1 clearly show how the onlineshopping process differs in terms ofgender in Finland and the USA. Thetable shows that with a risk level of� � 0.05, the only variable that does notreject the null hypothesis ofindependence between genders inFinland is the postpurchase behaviour.The need recognition phase is the mostproblematic, as one of the expectedfrequencies was less than unity. However,the difference is so small that the resultswould be significant even if theobservations were arranged so that theexpected frequencies were greater thanone. (If one female respondent hadchanged her mind and answered ‘never’instead of ‘sometimes’ to thecorresponding question, the p-valuewould have been 0.003). Therefore, theconclusion is that the difference issignificant even with an expected valueless than one.

Need recognition, evaluation, andpurchase decision have fairly similarcontingency coefficients meaning that the



Table 1: Hypotheses H1: Gender

Variable Chi-square value p-value Contingency coefficient

FinlandNeed recognition 13.152 0.001 0.254Information search 7.534 0.023 0.194Evaluation 28.653 0.000 0.360Purchase decision 17.179 0.000 0.287Postpurchase behaviour 4.856 0.088 –USANeed recognition 1.844 0.398 –Information search 1.229 0.268 –Evaluation 1.248 0.536 –Purchase decision 0.674 0.714 –Postpurchase behaviour 1.227 0.541 –

in a positive direction, it was concludedthat there was a positive relation to thepurchase volume, thus accepting thehypotheses H2a–H2e. It is to be notedthat even though a positive relationshipbetween the possible increases in thebuying process stages and purchasevolumes were confirmed, no conclusionsconcerning the causal directions of theserelationships may be established. It isequally likely that a large amount ofonline evaluation triggers the purchasedecision because avid online buyers tendto evaluate the products online beforepurchase. Table 3 summarises the resultsof the chi-square analysis for differentstages.

With a 5 per cent risk level, the onlyvariables that supported the hypothesis ofindependence (contrary to thehypotheses) was need recognition in theUSA. More importantly, ANOVAshowed that all the increases measuredwere significant. Therefore, it can safelybe concluded that there is a positiverelationships between the increases in theaforementioned purchase process stages

support to four of the first fivehypotheses, while the US sample did notsupport any of them. Table 2 gives anoverview of these results.

H2: Differences in consumer online

buying behaviour between 2002 and2004/2005

The second hypothesis suggested thatthere have been changes in the wayconsumers approach the general purchaseprocess. As expected, the amount ofonline shopping has increased in bothFinland and in the USA. It was alsonoticed that the average monetaryamount of purchase has decreased inFinland, while the purchase quantitieshave increased, giving clues about more‘everyday’ type of shopping. This givesreason to assume that should there beany changes in consumer behaviour, theyare likely to be positive (more needrecognition, information search etc). Achi-square test was first performed on allof the purchase process stages and if astatistically significant change was found



Table 2: Summary of results for H1 hypotheses

H1a: Gender influences the need recognition phase of online buying process. A(FIN)/R(US)H1b: Gender influences the information search phase of online buying process. A(FIN)/R(US)H1c: Gender influences the evaluation phase of online buying process. A(FIN)/R(US)H1d: Gender influences the purchase decision phase of online buying process. A(FIN)/R(US)H1e: Gender influences the postpurchase phase of online buying process. R(FIN)/R(US)

A = accepted, R = rejected

Table 3: Hypotheses H2: Time of sampling

Variable Chi-square value p-value Contingency coefficient

FinlandNeed recognition 12.649 0.002 0.178Information search 12.249 0.002 0.175Evaluation 30.022 0.000 0.268Purchase decision 36.868 0.000 0.296Postpurchase behaviour 8.644 0.013 0.187USANeed recognition 4.557 0.102 –Information search 14.564 0.001 0.212Evaluation 19.027 0.000 0.240Purchase decision 38.199 0.000 0.334Postpurchase behaviour 10.569 0.005 0.199

surfers indicated that pop-ups interferedwith their use of webpages. Additionally,there may be something fundamentallydifferent in the Finnish marketplace.Possibly the strategies used to triggerneed recognition are more effective forthe Finnish customers at this stage.

Thus, it has been established that thepurchase process is dependent on genderand sampling time. It is no surprise thatit has also been determined that theamount of online shopping has increasedin both countries between the twosamples. Hypotheses H3 and H4 exploremore deeply into the relationshipsbetween purchase volume increase andpurchase process behaviour. Table 4summarises the results derived from H2.

H3: The effects of consumer online

behaviour patterns to the purchasevolume

The third hypothesis suggested that thereis a positive relationship between theaffiliation to the different factors thatinfluence the buying process and thepurchase volume. This hypothesis wastested with ANOVA and regressionanalysis using dummy variables. Most ofthe respondents answered that theiraffiliation to any given factor was eitherhigh or moderate, and therefore theresponses where the number was equalto or less than 5 per cent of the totalresponses were omitted. The results fromthe Finnish 2004 sample supported thistheory for all of the five stages. Table 5shows the means of purchased items inthe last 12 months for all the variables

and the increase in purchase quantity.When the samples were divided betweenmen and women, it was found that allthe stages that were significantlydependent were also significantlydependent when men and women wereobserved separately, so neither of thegenders was ‘responsible’ for this positiverelationship.

By definition, need recognition istriggered by either internal or externalstimuli. Therefore, need recognition ispassive in nature. Need recognition doesnot require any active measures from theend user. Marketers could benefit fromidentifying the circumstances andconditions where these stimuli aretriggered, so that they couldcommunicate the appropriate informationto the customers as suggested by Kotler.65

In Finland, this information has beenreceived rather successfully in the pastcouple of years, while in the USA thereappears to be no visible change. Due tothis finding, one may question whetherthe US online markets are already toosaturated with information fromadvertisements. US customers havechanged their attitudes and behaviourtowards such promotional information bynot paying as much attention to it asthey did in the past. These customersfeel that many forms of delivery ofonline promotional information such asbanner and skyscraper ads, pop-up andpop-under ads are a nuisance, intrusiveand annoying, and they should beblocked from their online browsing. Forexample, a 2001 study by nua InternetSurveys66 found that 62 per cent of web




H2a: The time of sampling affects the amount of need recognition. A(FIN)/R(US)H2b: The time of sampling affects the amount of information search. A(FIN)/A(US)H2c: The time of sampling affects the amount of evaluation. A(FIN)/A(US)H2d: The time of sampling affects the amount of purchase decisions. A(FIN)/A(US)H2e: The time of sampling affects the amount of postpurchase satisfaction. A(FIN)/A(US)


the hypotheses, with the exception ofpostpurchase behaviour on all occasions.The results from the female sample onlysupport the cases of need recognition,purchase decision, and postpurchasebehaviour. Table 6 summarises the resultsfrom H3a–H3e.

H4: The effects of consumer online

behaviour patterns to purchase volumechange

It has been established that in bothcountries there is an increase in thepurchase volume between the earlier andthe later samples, and that there is apositive relationship between some of thepurchase process stages and purchasevolume. The fourth hypothesis suggeststhat there is a positive relationshipbetween the purchase process stages andthe increase in purchase volume. Whilethe mean results support this hypothesisfairly well, in reality things are not assimple as they seem. In most of theFinnish samples’ cases, the majority ofthe respondents belong to the middlegroup, leaving rather few observations forthe extreme ends of the spectrum. Thisleads to large standard deviations and tothe fact that the test statistics forANOVA and regression analysis wouldneed to be much larger for the extreme

followed by p-values from regressionanalysis classified by the affiliation (high,moderate and low) for each variable. Thep-values indicate the difference from‘high’ internet usage with null hypothesisbeing that this difference is 0. Given the5 per cent risk level, there is a clearindication of a positive relationship withneed recognition, information search,evaluation, purchase decision andpostpurchase satisfaction in Finland.

Table 5 shows how the purchasedecision mean values drop in Finland asthe level of internet usage in differentstages decreases with the first fivevariables. When men and women wereobserved separately, few interestingfindings were made. From the Finnishsample, the results showed how onlineevaluation is positively related topurchase volume. However neither men’snor women’s separate samples supportedthese findings. The most likely reason isthat as the p-value was quite high(0.042) to begin with, the decreases inusable observations reduced the F-valuesenough to raise the p-value above 5 percent. The data also suggest that there areno dependencies between the purchasevolume and the levels of informationsearch or purchase decision as far aswomen are concerned. In the USA, theresults from the male sample supported



Table 5: Hypotheses H3: Purchase volume

AmountHigh Moderate LowMean p-value Mean p-value Mean p-value

FinlandNeed recognition 6.54 0 1.80 0.000 – –Information search 4.99 0 2.03 0.003 – –Evaluation 4.98 0 2.52 0.017 – –Purchase decision 6.24 0 2.85 0.003 0.84 0.004Postpurchase behaviour 8.02 0 2.83 0.000 – –USANeed recognition 17.76 0 8.21 0.001 – –Information search 14.62 0 7.93 0.009 – –Evaluation 14.97 0 8.70 0.019 – –Purchase decision 24.96 0 8.77 0.000 2.20 0.000Postpurchase behaviour 14.74 0 8.08 0.020 – –

differences between the mean values ofthe 2002 and 2004/2005 samples.

The conclusion for this hypothesis isunclear. The mean results give support insome form to all of hypotheses H4a–H4e,but statistical evidence only supportshypotheses H4a and H4b in Finland, andH4a, H4b and H4d in the USA. As thereis always the possibility that the meanvalues were due to chance (that is,indeed, what the results from ANOVAindicate), all hypotheses that did nothave statistical support from the analysesare rejected. However, it must berecognised that there might be sometruth in the hypotheses that wererejected. Given more observations, thetest statistics might show significantdifferences similar to the hypotheses.Table 8 sums up the hypotheses of H4 asthey were computed above.

CONCLUSIONS ANDLIMITATIONSFrom the results it seems clear thatonline shopping has increased inpopularity among both male and femaleportions of the target groups in Finland,and more so in the USA. The internetalso has increased in popularity as a toolused to contribute to and assist in thebuying process. In spite of the increasein both genders, it seems that in Finland,men tend to recognise more needsonline and use the internet forinformation search and evaluation more

groups than for the middle group tomake any statistical significance. In theUSA, most of the respondents fell intothe ‘high’ and ‘moderate’ groupsreducing the usefulness of the statisticalanalyses as far as the ‘low’ group isconcerned.

In Finland, the statistical differencesthat suggest any support for the generalH4 hypothesis, were significant only withH4a and H4b. The increases in purchasevolume were significantly larger(� � 0.05) for people with a highfrequency in online need recognition andinformation search. In the USA,supportive evidence was found for H4a,H4b, and H4d. As was the case withFinland, the increases in purchase volumein the USA was larger for people with ahigh frequency in online needrecognition and information search, andno differences were found in the ‘low’groups. With postpurchase satisfaction,the differences were less between thehigh and moderate groups, but as in thelow group, there was no increase at all,there is enough support for H4d in theUSA. Overall the USA increases weresignificantly larger than Finnish. Table 7lists the p-values from ANOVA andregression analysis. If the assumption ofhomogenous variances failed,67,68 thestatistics were computed instead of theusual F-statistic. Table 7 also lists the�-coefficients (when statistically differentfrom zero) obtained from regressionanalysis. These coefficients show the




H3a: There is a positive relationship between the amount of consumer online need A(FIN)/A(US)recognition and the online purchase volume.

H3b: There is a positive relationship between the amount of consumer online A(FIN)/A(US)information search and the online purchase volume.

H3c: There is a positive relationship between the amount of consumer online evaluation A(FIN)/A(US)and the online purchase volume.

H3d: There is a positive relationship between the amount of consumer online purchase A(FIN)/A(US)decisions and the online purchase volume.

H3e: There is a positive relationship between the amount of consumer online postpurchase A(FIN)/A(US)satisfaction and the online purchase volume.


�

shoppers is bound to be changing overtime, as new services and productsemerge in the markets, along with betterlogistic solutions for delivery andhandling. The two-year span of thisstudy, however, was not enough tocapture any large changes. Affordableairline tickets made available online are

than women. Finnish men also showed ahigher frequency in online purchasedecisions and postpurchase behaviour. Inthe USA, there was no such gender gap,showing that the online shoppingorientation between men and womendid not differ significantly.

The distribution for most frequent



Table 7: Hypotheses H4: Change in purchase volume

p-value �

FinlandAmount of online need recognition High 0.033 3.522

Moderate 0.000 1.072Low –a –

Amount of online information search High 0.019 2.68Moderate 0.000 1.312Low –a –

Amount of online evaluation High 0.069 –Moderate 0.002 1.496Low 0.810 –

Amount of online purchase decision High 0.176 –Moderate 0.023 1.300Low 0.008 0.685

Amount of postpurchase behaviour High 0.070 –(likelihood of repurchase) Moderate 0.635 –

Low 0.585 –USAAmount of online need recognition High 0.005 12.018

Moderate 0.000 4.945Low 0.743 –

Amount of online information search High 0.002 9.049Moderate 0.000 4.656Low –a –

Amount of online evaluation High 0.024 8.286Moderate 0.000 5.336Low 0.015 8.449

Amount of online purchase decision High 0.082 –Moderate 0.006 3.526Low 0.062 –

Amount of postpurchase behaviour High 0.015 8.825(likelihood of repurchase) Moderate 0.007 4.538

Low –a –

aStatistics could not be computed due to a zero-variance variable


H4a: There is a positive relationship between the amount of consumer online need recognition A(FIN)/A(US)and the possible increase in online purchase volume.

H4b: There is a positive relationship between the amount of consumer online information A(FIN)/A(US)search and the possible increase in purchase volume.

H4c: There is a positive relationship between the amount of consumer online evaluation R(FIN)/R(US)and the possible increase in purchase volume.

H4d: There is a positive relationship between the amount of consumer online purchase R(FIN)/R(US)decisions and the possible increase in purchase volume.

H4e: There is a positive relationship between the amount of consumer online postpurchase R(FIN)/A(US)satisfaction and the possible changes in purchase volume.


different stages of the purchase process,tends to buy more from the internet.This phenomenon is occurring generallywith men and women in both of thetarget countries. The need recognitionphase in particular is of interest becausethe results show that if a potentialcustomer frequently discovers productsthey need while online, they are likelyto make more purchases than those whouse the internet frequently for onlyinformation search or evaluation. InFinland, the results suggest that thefrequency of need recognition has agreater influence on the number ofgoods bought from the internet than thefrequency of online purchase decisionsthemselves. The results also clearly showthat it is worthwhile for e-marketers tokeep their customers satisfied. If thee-marketer satisfies the customer, thatcustomer is a prime candidate for arepurchase. As the total number ofonline purchases continues to increase,the e-marketer who effectively serves,satisfies, and delights their online buyerswill enjoy repeat patronage.

This study has shown that the widelydiscussed five-stage model for consumerbuying behaviour has implications in theonline environment. Although onlineshopping in general does not depend onthe geographical location (neither buyer’snor vendor’s), the cultural differences,along with the ICT infrastructure, doaffect the process as was illustrated in thisstudy. Even though Finland and the USAboth have a very high level ofsophistication in the ICT field, the USonline shopping field could be describedas somewhat more convergent, becausemen and women are frequently marketedto as separate segments. The increase inUS online shopping also was significantlylarger than in Finland. As there is cleargrowth in online purchase volume inboth countries and in fact globally,e-marketers may be guided by the

still a good example of a product thathas gained popularity, and although notlisted in the questionnaire, onlineordering of train tickets was mentionedby many respondents in Finland. Onlinetrain ticket services are still a fairly newphenomenon in Finland, and if theincrease in airline tickets is an indication,the demand for online train tickets mayincrease in the future, especially amongthe Net Generation in Finland (forexample, students get a 50 per centdiscount from train tickets, plus many ofthem live away from their hometown).The effects on the general populationremain to be seen.

When the buying process is definedby the five-stage purchase model, it isevident that the internet is playing acontinuously increasing role. When theamount of internet usage related todifferent phases of the model wasmeasured, a statistical increase in bothcountries from 2002 to 2004/2005 wasfound with the largest increases being inthe purchase decision itself. There is noreason to assume that this trend willdecrease anytime soon. The NetGeneration in both Finland and the USAhave been enjoying affordable broadbandconnections for some time now, andthese consumers are familiar with theadvantages the internet has to offer. Thistrend is also likely to spread to the restof the population, as the number ofinternet service providers offeringinexpensive broadband connectionsincreases. This will certainly push theinternet penetration rate even higher. Nochanges in need recognition weremeasured, but a slight increase in thetrust factor was noted. In light of currentinternet connectivity technology (ICT)developments, this trend is likely tocontinue in both countries.

This study shows clearly that the NetGeneration segment of the populationwith high internet usage relative to the



would have been beneficial in obtainingmore accurate information about thebehavioural trends in online shopping,instead of just drawing conclusions fromprevious trends of development.According to this and previous reports,however, there is no need to assume thatthe amount of online shopping willdiminish anytime soon.

References1 Alch, M. (2000) ‘Get ready for a new type of

worker in the workplace: The Net Generation’,Supervision, Vol. 61, No. 4, pp. 3–7.

2 World Economic Forum (2004) ‘NetworkedReadiness Index’; available at:http://www.weforum.org/pdf/Global_Competitiveness_Reports/Reports/GITR_2004_2005/Networked_Readiness_Index_Rankings.pdf,accessed 26th May, 2005.

3 Kotler, P. and Kelle, K. L. (2006) ‘MarketingManagement’ (12th edn), Prentice Hall, UpperSaddle River, NJ, pp. 191–199.

4 Jobber, D. and Lancaster, G. (2003) ‘Selling andSales Management’ (6th edn), Prentice Hall, UpperSaddle River, NJ, p. 216.

5 Roth, R. (2001) ‘Yes, customer control isfundamental change’, InternetWeek, February, p. 23.

6 Fader, P. (2001) ‘No, it doesn’t change basic buyingpatterns’, InternetWeek, February, p. 23.

7 Kotler and Kelle (2006), see ref. 3 above.8 Kotler, P. and Armstrong, G. (2005) ‘Marketing: An

Introduction’, Prentice Hall, Upper Saddle River,NJ, p. 165.

9 Kinnear, T. and Bernhardt, K. (1986) ‘Principles ofMarketing’ (2nd edn), Scott, Foresman and Co.,Glenview, IL, p. 146.

10 Ibid, p. 149.11 Ibid.12 Ibid, p. 150.13 Dubois, B. (2000) ‘Understanding the Consumer’,

Prentice Hall, Upper Saddle River, NJ, p. 229.14 Van Trijp, H., Hoyer, W. and Inman, J. (1996)

‘Why switch? Product category-level explanationsfor true variety-seeking behavior’, Journal ofMarketing Research, Vol. 33, No. 3, pp. 281–292.

15 Kotler and Kelle (2006), see ref. 3 above,pp. 191–192.

16 Ibid.17 Dubois (2000), see ref. 13 above, p. 231.18 Alba, J. and Hutchinson, J. (1987) ‘Dimensions of

consumer expertise’, Journal of Consumer Research,Vol. 13, No. 4, pp. 411–454.

19 Beatty, S. and Smith, S. (1987) ‘External searcheffort: an investigation across several productcategories’, Journal of Consumer Research, Vol. 14,No. 1, pp. 83–95.

20 Biswas, D. (2004) ‘Economics of information in theWeb economy: Towards a new theory?’, Journal of

findings of hypothesis H4. E-marketersshould continue to design and delivereffective online support for the needrecognition, information search, andpurchase decision stages of the consumerbuying decision process. Additionally, theonline evaluation and postpurchasesatisfaction stages should be prioritised forcontinued investigation, betterunderstanding, and improvement.Particularly in emerging online markets,during the introduction and growthphases of online purchasing, gender mayinfluence all but the postpurchase stageof the consumer buying decision process,with males being more active thanfemales. These results provide importantguidance as to how the five-stagepurchase process model could be utilisedby e-vendors to improve their marketingonline.

It is essential to acknowledge thelimitations of this study that reduce thegeneralisability of the findings. The mainlimitations of this study are the samples.University students, as part of the NetGeneration, are extensive users of thelatest information technology. This makesthem ideal for this study because they allutilise computers frequently. The biasmay come from the fact that thesesamples represent only a portion of thepopulation that have above averagecomputer skills, as well as free access tothe internet. Thus, any generalisations tothe general population should be madewith caution. Internet usage is, however,increasing in all segments of thepopulation. Given that this study was notintended to establish the proportion ofthe population that use the internet topurchase online, but was intended toinvestigate the buyer decision process ofthose that do, the Net Generation’shighly literate and heavy online users arecertainly more qualified and better ableto provide the requisite information.

Additionally, a third reference point



42 Auh, S. and Johnson, M. (2005) ‘Compatibilityeffects in evaluations of satisfaction and loyalty’,Journal of Economic Psychology, Vol. 26, No. 1,pp. 35–37.

43 Ball, D., Coelho, P. and Machas, A. (2004) ‘Therole of communication and trust in explainingcustomer loyalty’, European Journal of Marketing, Vol.38, No. 9/10, pp. 1272–1293.

44 Evanschitzky, H., Iyer, G., Hesse, J. and Ahlert, D.(2004) ‘E-satisfaction: a re-examination’. Journal ofRetailing, Vol. 80, No. 3, pp. 239–247.

45 Szymanski, D. M. and Hise, R. T. (2000)‘E-satisfaction: an initial examination’, Journal ofRetailing, Vol. 76, No. 3, pp. 309–322.

46 Cho, Y., Im, I., Hiltz, R. and Fjermestad, J.(2002) ‘The effects of post-purchase evaluationfactors on online vs offline customer complainingbehavior: implications for customer loyalty’,Advances in Consumer Research, 29, No. 1, pp.318–326.

47 Hellier, P., Geurse, G., Carr, R. and Rickard, J.(2003) ‘Customer repurchase intention: a generalstructural equation model’, European Journal ofMarketing, Vol. 37, No. 11/12, pp. 1762–1800.

48 Palanisamy, R. (2004) ‘Impact of gender differenceson online consumer characteristics on web-basedbanner advertising effectiveness’, Journal of ServicesResearch, Vol. 4, No. 2, pp. 45–74.

49 Rodgers, S. and Harris, M. (2003) ‘Gender ande-commerce: an exploratory study’, Journal ofAdvertising Research, Vol. 43, No. 3, pp. 322–329.

50 Dholakia, R. and Chiang, K. (2003) ‘Shoppers incyberspace: are they from Venus or Mars and does itmatter?’, Journal of Consumer Psychology, Vol. 13,No. 1/2, pp. 171–176.

51 Gabriano, E. and Strahilevitz, M. (2004) ‘Genderdifferences in the perceived risk of buying onlineand the effects of receiving a site recommendation’,Journal of Business Research, Vol. 57, No. 7,pp. 768–775.

52 Ibid.53 Sahkoisen Kaupan Palvelukeskus (2004) ‘Suomi:

internet-kuluttajakaupan arvo 2,1 miljardia euroa’,available at: http://www.e-finland.org/cgi-bin/center/xhtml?/pageID=53//list=3/from=4/size=1/sort=down+25/find=+/show,accessed 24th May, 2005.

54 The Census Bureau of the Department ofCommerce (2005) US Census Bureau News,http://www.census.gov/mrts/www/data/pdf/05Q1.pdf, accessed 24th May, 2005.

55 Oliver (1999), see ref. 41 above.56 Shang, R., Chen, Y. and Shen, L. (2005) ‘Extrinsic

versus intrinsic motivations for consumers to shopon-line’, Information and Management, Vol. 42, No. 3,pp. 402–413.

57 Auh and Johnson (2005), see ref. 42 above.58 Ball, Coelho and Machas (2004), see ref. 43 above.59 Kotler and Kelle (2006), see ref. 3 above, pp.

192–193.60 Welch, B. L. (1951) ‘On the comparison of several

means values: an alternative approach’, Biometrika,

Business Research, Vol. 57, No. 7, pp. 724–733.21 Urbany, J., Dickson, P. and Wilkie, W. (1989)

‘Buyer uncertainty and information search’, Journal ofConsumer Research, Vol. 16, No. 2, pp. 208–215.

22 Kotler and Kelle (2006), see ref. 3 above, pp.192–193.

23 Peter, J. and Olson, J. (1996) ‘Consumer Behaviorand Marketing Strategy’ (4th edn),Irwin/McGraw-Hill, Boston, Massachusetts, p. 200.

24 Klenosky, D. and Rethans, A. (1988) ‘The formationof consumer choice sets: a longitudinal investigationat the product class level’, Advances in ConsumerResearch, Vol. 15, No. 1, pp. 13–18.

25 Narayana, C. and Markin, R. (1975) ‘Consumerbehavior and product performance: an alternativeconceptualization’, Journal of Marketing, Vol. 39,No. 4, pp. 1–6.

26 Haubl, G. and Trifts, V. (2000) ‘Consumer decisionmaking in online shopping environments: the effectsof interactive decision aids’, Marketing Science, Vol.19, No. 1, pp. 4–21.

27 Rowley, J. (2000) ‘Product search in e-shopping: areview and research propositions’, Journal ofConsumer Marketing, Vol. 17, No. 1, pp. 20–35.

28 Huber, J. and Klein, N. (1991) ‘Adapting cutoffs tothe choice environment: the effects of attributecorrelation and reliability’, Journal of ConsumerResearch, Vol. 18, No. 3, pp. 346–357.

29 Bhatnagar, A. and Ghose, S. (2004) ‘A latent classsegmentation analysis of e-shoppers’, Journal ofBusiness Research, Vol. 57, No. 7, pp. 758–767.

30 Hauser, J., Urban, G. and Weinberg, B. (1993)‘How consumers allocate their time when searchingfor information’, Journal of Marketing Research, Vol.30, No. 4, pp. 452–466.

31 Lee, A. and Labroo, A. (2004) ‘The effect ofconceptual and perceptual fluency on brandevaluation’, Journal of Marketing Research, Vol. 41,No. 2, pp. 151–165.

32 Ibid.33 Kotler and Armstrong (2005), see ref. 8 above,

pp. 36–37.34 Ibid, p. 169.35 Dubois (2000), see ref. 13 above, p. 239.36 Baumeister, R. (2002) ‘Yielding to temptation:

self-control failure, impulsive purchasing andconsumer behavior’, Journal of Consumer Research,Vol. 28, No. 4, pp. 670–676.

37 Ibid.38 Sojka, J and Giese, J. (2003) ‘Using individual

differences to detect customer shopping behavior’,The International Review of Retail, Distribution andConsumer Research, Vol. 13, No. 4, pp. 337–353.

39 Kotler and Kelle (2006), see ref. 8 above,pp. 198–199.

40 Mittal, V. and Kamakura, W. (2001) ‘Satisfaction,repurchase intent and repurchase behavior:investigating the moderating effect of customercharacteristics’, Journal of Marketing Research, Vol. 38,No. 1, pp. 131–142.

41 Oliver, R. (1999) ‘Whence consumer loyalty?’,Journal of Marketing, Vol. 63, No. 4, pp. 33–44.



108–140.64 Draper, N. and Smith H. (1981) ‘Applied

Regression Analysis’ (2nd edn), John Wiley andSons, Inc., Hoboken, NJ, pp. 241–250.

65 Kotler and Armstrong (2005), see ref. 8 above, p.204.

66 nua Internet Surveys, 3 May, 2001,http://www.nua.ie, accessed 28th May, 2001.

67 Welch (1951), see ref. 60 above.68 Brown and Forsythe (1974), see ref. 61 above.

Vol. 38, No. 3/4, pp. 330–336.61 Brown, M. B. and Forsythe, A. B. (1974) ‘The

ANOVA and multiple comparisons for data withheterogeneous variances’, Biometrics, Vol. 30, No. 4,pp. 719–724.

62 Greenwood, P. and Nikulin, M. (1996) ‘A Guide toChi-Squared Testing’, John Wiley and Sons, Inc.,Hoboken, NJ, pp. 86–90.

63 Christensen, R. (1996) ‘Analysis of Variance, Designand Regression’, Chapman & Hall, London, pp.



longitudinal comparison of finnish and us online …...the next stage in the buying process is...

Documents