In this section, we discuss the conceptual and methodological issues found in previous work. While we try to separate them, it is clear that conceptual and
methodological issues are related. For example, ambiguous definitions normally lead to inconsistent methods of operationalization and measurement.
2.4.1 Conceptual Issues
Issues related to the conceptualization of flow are manifested as (1) inconsistent and incomplete flow models and (2) inconsistent and inadequate definitions of core constructs. We will discuss each of these issues in detail below. Table 2.5 summarizes selected empirical research in terms of the flow models used. The majority of the studies included in this table involve Internet usage and online shopping. For a similar summary of studies on general information technology, please refer to Agarwal and Karahanna (2000).
Table 2.5 Summary of Selected Research on Flow
Authors Construct Dimensions Antecedents Consequences
1. Csikszentmihalyi (1988) Flow Concentration; Mergence; Control; Time distortion; Loss of self- consciousness Balance of challenges and skills; Clear goal; Clear feedback Autotelic experience 2. Csikszentmihalyi and LeFevre (1989)
Flow In flow vs. not in flow Balance of challenges and skills; Work vs. leisure Affect; Potency; Concentration; Motivation; Satisfaction 3. Hoffman and Novak (1996) Flow Mergence; Time distortion; Lost of self- consciousness; Telepresence; Gratifying state
Balance of skills and challenges; Interactivity; Focused attention Positive subjective experience; Increased learning; Exploratory and participatory behavior; Perceived control 4. Chen et al. (1998) Flow Enjoyment; Attention; Time distortion Challenges vs. skills; Clear goal
5. Nel et al. (1999) Flow Control; Attention focus; Curiosity; Intrinsic interest
Web site type (content, audience focus)
Return intention; Overall site rating
Table 2.5 Continued
Authors Construct Dimensions Antecedents Consequences
6. Agarwal and
Karahanna (2000) Cognitive Absorption Temporal dissociation; Focused immersion; Heightened enjoyment; Control; Curiosity Personal innovativeness; Playfulness Perceived usefulness; Perceived ease-of-use 7. Novak et al.
(2000) Flow One-dimensional Telepresence; Time distortion; Challenge/arousual; Skill/control; Interactivity
No conclusive results for consequences of flow
8. Wan and Nan (2001) Optimal online experience Web excitement (flow); Other dimensions: Positive emotions; Negative emotions; Evaluation of Web site structure and efficiency; Negative Web experience
Surfing motive; Web site design
Web actions;
Change of brand attitude
Shopping enjoyment Product involvement; Web skill; Challenge; Value-added search return intention 9. Koufaris (2002) Flow
Concentration Product involvement;
Web skill; Challenge 10. Novak et al. (2003) Flow Flow; Flow verbatim Goal directed vs. experiential activities; Skill; Challenge Novelty; Importance 11. Skadberg and Kimmel (2004) Flow experience Enjoyment; Lost track of time; Telepresence
Interactivity; Attractiveness; Proposed but dropped: Domain
knowledge/skill; Information in the Web site/challenge
Increased learning, in turn it leads to changes of attitude and behavior
12. Huang (2003a) Flow Attention; Control; Curiosity; Interest Complexity; Novelty; Interactivity Utilitarian; Hedonic 13. Finneran and
Zhang (2003) Flow People: trait and states Task;
2.4.1.1 Flow Models
First of all, from the table, it clear that there are discrepancies among flow models used in these studies. This inconsistency can be found in two aspects. One inconsistency is the total number of flow core constructs examined in a given study. It varies from thirteen (Novak et al. 2000) to two (e.g., Koufaris 2002). For example, the requirements for a clear goal and interactivity are examined in some studies, but are assumed to be embedded in Web activities or technology in other studies. This assumption might be worthy of reexamination. The second inconsistency resides in the theorization of those constructs: the antecedents and consequences of flow are modeled differently. The same constructs are placed in different stages in different studies. For instance, perceived control has been treated as a part of flow experience in some studies (e.g., Trevino and Webster 1992) but as an antecedent in others (Ghani 1995; Novak et al. 2000). In addition, with the help of structural equation modeling, some studies further differentiate antecedents of flow into direct and indirect factors, while others treat them the same. Although inconsistency is not necessarily deficient, it is a sign of lack of cohesiveness in theorization and maybe an indicator of the immaturity of the field.
More seriously, not all dimensions in the Csikszentmihalyi’s flow theory have been investigated. Few have studied time distortion and balance of challenges and skills; fewer have looked into either the merging of action and awareness or the loss of self-consciousness. Some studies omitted important constructs. For example, perceived challenges and skills are sometimes excluded (e.g., Nel et al. 1999; Wan and Nan 2001),
and we see this as a major flaw. In summary, none of the models is a complete model. More research is needed to answer questions such as what flow experiences are associated with Web activities, including Internet shopping, and what antecedents and consequences of flow are relevant to online environments.
2.4.1.2 Flow Constructs.
The second conceptual issue is related to the definition of core constructs. The operationalization of the most important constructs – flow and the ratio of perceived challenges and skills – is troublesome. First, although researchers agree on the underlying multidimensionality of flow, it seems they cannot agree on the number of dimensions. This issue is the direct result of the inconsistent models used in the studies and can be seen in Table 2.5.
Secondly, measures of perceived challenges and skills are also a concern. Ellis et al. (1994) point out that unidimensional scales of challenges and skills may not serve as valid measures. Although at first this appears to be a measurement issue, it is also rooted in conceptualization of the situated challenge and skill perception. It is especially true when collecting everyday-life data using the Experience Sampling Method, because challenges and skills can be context-based. Merely asking whether it is hard to carry out an activity will not yield enough information on that particular situation. In the case of Internet activities, challenges and skills are multi-faceted and activity-specific, because Internet activities consist of a wide range. For example, shopping online requires the skills of using a computer and the Internet, knowledge of the site and product in question, and payment ability (such as possessing a credit card). Failing to incorporate
challenges and skills, along with not recognizing this contextual aspect of challenges and skills, leads to unjustified research models and results that go astray.
Third, the ratio of perceived challenges and skills is thought of as the determinant of flow in many studies in other fields. However, when studying flow in online and computer environments, the concept of this crucial balance is lost in some studies. For example, except for Ghani (1995) and Chen et al. (1998), other studies have treated the impact of perceived challenges and skills separately instead of using the balance of these two (e.g., Novak et al. 2000; Koufaris 2002).
2.4.2 Methodological Issues
In addition to the conceptualization problems aforementioned, there are issues related to methodology and measurement as well. First, there are some problems with data collection. The data collection method employed in a study determines what kind of data is collected. It also determines how close the data collected is to the phenomenon we are trying to study and measure. In studies of flow in the Internet, data collection methods are dominantly self-reported surveys (Novak et al. 2000) and retrospective questionnaires after experiments (Nel et al. 1999). Results from those studies may be informative, but are not adequate for studying flow, which is a situated, conditioned, dynamic, individual experience (Chen et al. 1999). Major limitations of the survey method in studying flow are (1) ignoring the situated, dynamic nature of flow and (2) memory loss and distortion. “Surveying non-situated, generalized factors does not account for the dynamism of each factor and how its fluctuation influence flow” (Finneran and Zhang 2002).
Experiments with questionnaires quickly following them enable researchers to observe the situation more closely and to collect data with less memory loss and distortion, but are not without drawbacks of their own. Although experiments raise concerns of external validity in general, it is especially so in the study of flow, for two reasons. First, there is little consensus on the underlying structure and operationalization of flow and related constructs. Second, flow is such a context-specific phenomenon that the design of the experiments (selected factors, task, and the site used) may have a compound effect. Thus, we may reach an incorrect conclusion by using an experimental approach before we have obtained a clearer understanding of flow using other research methods. For example, Nel et al. (1999) concluded that the Web site type was the leading factor of flow. This result is misguided at best, since the context considered in that study was mainly in terms of site type, ignoring the fact that context consists of so many other uncontrolled and unconsidered factors, such as user intention and site design. Thus, without a solid theoretical foundation and a systematic plan to investigate contextual factors, any experimental design would appear arbitrary. Only one study made an attempt to use the ESM to examine flow in the context of Web surfing (Chen et al. 1998); however why used 5 to 7 minutes as the time interval was not explained. In regard to data collection issues, Hoffman and Novak (1996) called for a comprehensive measurement procedure to include flow, antecedents, consequences, and other related psychological experiences via both qualitative and process tracing measures. Finneran and Zhang (2002) have suggested using qualitative techniques to enhance our
understanding of flow and to ensure validity. In our study we will try to overcome those conceptual and methodological challenges.
Another issue is lack of consistent measurement. We can see researchers tried to build on previous work by reusing scales when it was possible. However, there is still an inconsistency, which can be traced back to inconsistent operationalization of constructs.
2.4.3 Additional Observations
The next two observations are of special relevance to our study. First, in terms of results, although most findings are consistent with the original flow theory, there are a few inconsistencies and contradictions among them. For example, Chen et al. (1998) reported contradictory results. In their study, in-flow surfers enjoyed less and paid less attention than those not in flow. No fully satisfying explanation was given for this result. It was suggested to reexamine the applicability of flow in Internet activities, which is one of the objectives of our study.
Second, there is no comprehensive understanding of the specific activities during which people actually experience flow on the Web (Novak et al. 2003). Some empirical studies of flow in Internet usage only investigated the general experience of Web usage and treated Web activities as one whole activity. This approach warrants reexamination. There are so many different online activities and although these activities share a certain commonality, they are totally different in terms of intention, expectations, challenges, people’s skills, and so on. Chen et al. (1999) shows that people experience flow more in certain Web activities than in others. Treating those activities as though they were the
same neglects the contextual nature of flow. Therefore, studying individual activities may yield more specific guidelines for different kinds of sites. We chose to start with Internet shopping because (1) there is little empirical study done on flow experience in Internet shopping; (2) Internet shopping is a contextually rich activity; and (3) it is quite different from its counterpart in conventional shopping.
In summary, studying flow in online environments is still underexplored, both because flow theory itself is still under development and it is only relatively recently that flow has been introduced into Information Systems research. In response to the call for an extension on their work by going “beyond a retrospective general evaluation of customer experience on the Web to its modeling in specific online situations” (Novak et al. 2000), this dissertation is going to look into online shopping experience in particular and in a close-up manner using both qualitative and quantitative methods.
CHAPTER III
RESEARCH OBJECTIVES AND QUESTIONS
The literature review showed that applying flow to understand the online consumer experience was promising, but that this area of research was in need of further development. This research effort was spent in two aspects of flow research. In Study 1 we conduct a validity study of flow. Study 2 tests a comprehensive flow model in online shopping. By conducting those studies this dissertation continue the endeavor and deepen the inquiry regarding flow in IS and Internet research in several ways, which will be discussed within the next section of Overview. Detailed discussion of sub-studies will follow.