WPbanner5.jpg (10176 bytes)

No. WP-98-01

A longitudinal investigation of personal computers in homes:
Adoption determinants and emerging challenges


Viswanath Venkatesh
Susan A. Brown
The Robert H. Smith School of Business
Kelley School of Business
University of Maryland
Indiana University
College Park, MD 20742
Bloomington, IN 47405
Phone: 301-405-7594
Phone: 812-855-3484
FAX: 301-314-9157
Fax: 812-855-8679
E-mail: vvenkate@mbs.umd.edu
E-mail: suebrown@indiana.edu
August 1998

Center for Social Informatics
SLIS
Indiana University


ABSTRACT

While technology adoption in the workplace has been studied extensively, drivers of adoption in homes have been largely overlooked. This paper presents the results of a nation-wide, two-wave, longitudinal investigation of the factors driving personal computer (PC) adoption among American homes. The findings revealed that innovators and early adopters were driven by a desire to obtain hedonic outcomes (i.e., pleasure) and social outcomes (i.e., status) from adoption. The early majority was strongly influenced by utilitarian outcomes, and friends and family members. The late majority and laggards have not adopted primarily because of rapid changes in technology and consequent fear of obsolescence. A second wave of data collection conducted six months after the initial survey indicated an asymmetrical relationship between intent and behavior among intenders and non-intenders, with non-intenders following more closely with their intent (to not adopt a PC). We present important implications for research on adoption of technologies in homes and the workplace, and also discuss challenges facing the PC industry.


INTRODUCTION

"There is no reason anyone would want a computer in their home."
-Ken Olson, President, Chairman and Founder of Digital Equipment Corporation (DEC), 1977.

Early predictions about the adoption and use of a telephone proved to be false as it became a necessity, rather than a luxury, in most American households. Likewise, early assessments about the adoption and use of personal computers (PCs) in homes are quickly proving to be false (Venkatesh 1996). Increasingly, PCs, powered by the ability to deliver Internet services (e.g., email, the world wide web) and support activities in homes (e.g., household financial planning), are being touted as an innovation that will have impact similar to the telephone. While recent surveys, including a Times-Mirror survey (Kohut 1995), Nielsen surveys, and a survey by members of the academic community (Kraut 1996; Kraut, et al. 1996), indicate that a little over a third of all households have adopted PCs, little systematic research has been conducted to understand the determinants of PC adoption in homes.

Adoption and diffusion of information technologies have been studied extensively in information systems (IS) research at the individual level (e.g., Agarwal and Prasad 1998) and the organizational level (e.g., Cooper and Zmud 1990). The theoretical models employed to study user acceptance, adoption, and usage behavior include the Theory of Reasoned Action (e.g., Ajzen and Fishbein 1980; Davis, Bagozzi, and Warshaw 1989), the Technology Acceptance Model (e.g., Davis, et al. 1989), the Theory of Planned Behavior (e.g., Ajzen 1991; Mathieson 1991), the Model of PC Utilization (Thompson, Higgins, and Howell 1991), the Decomposed Theory of Planned Behavior (e.g., Taylor and Todd 1995), and Innovation Diffusion Theory (e.g., Agarwal and Prasad 1997; Brancheau and Wetherbe 1990; Rogers 1995). These models were applied primarily to explain user (individual) adoption and usage of technologies in the workplace. However, little, if any, research has been done to understand why people do and do not adopt computers for home use. Even in the context of understanding home use of PCs, prior research has studied the role of the home PC as a tool to enable working from home (see Venkatesh and Vitilari 1992). We expect the factors affecting household technology adoption decisions to be different from workplace decisions due to the personal nature of the expense, as well as the goals for technology use. Thus, the first objective of this research is to identify factors that determine household PC adoption.

The extensive research on technology adoption employs two key assumptions: (1) a common core set of salient beliefs drives behavior (e.g., Davis, et al. 1989), suggesting that adopters and non-adopters will simply have different evaluations of the same set of factors, and (2) everyone will eventually adopt (Rogers 1995). However, an emergent perspective in the business press has challenged these two assumptions (see Moore 1991). Therefore, a second objective of this research is to determine the factors influencing home PC adoption among different categories of adopters, and non-adopters.

In understanding the relationship between the salient determinants and subsequent adoption behavior, psychology, marketing, and IS research employ intention as the key predictor of behavior (e.g., Ajzen 1991; Davis, et al. 1989; Morwitz and Schmittlein 1992). Prior research has reported intention-behavior correlations of .18-.84 across a variety of behaviors (see Ajzen 1991 for a review; see also Sheppard, Hartwick, and Warshaw 1988). Though there has been some recent research challenging the predictive validity of user acceptance models (e.g., Straub, Limayem, and Karahanna-Evaristo 1995), there is also evidence supporting the predictive validity of these models (e.g., Taylor and Todd 1995). Thus, the third objective of this research is to examine the nature of the relationship between the intent to adopt and subsequent purchase behavior in the context of household adoption of PCs.

To accomplish these objectives, theories from psychology, marketing and IS were integrated to develop a core set of determinants expected to influence adoption and non-adoption of PCs in homes. To empirically test and refine the theory, we conducted a two-wave, nation-wide telephone survey, consisting of an initial survey and a follow-up survey six months later. The survey included open-ended questions to avoid constraining the results to a priori expectations and thus, allowed us to account for unanticipated determinants (Stone 1978). This was particularly important given the exploratory nature of this work.

PC adoption in homes

In this section, we first discuss the adopter categories used in this research and how they map on to the categories identified by Rogers (1995). Next, we discuss the technology adoption life cycle and the bases for our expectations about differences in drivers of adoption decisions across the different categories of adopters. Finally, we identify the different determinants expected to influence adoption decisions, and the role each determinant plays among different adopter categories and non-adopters.

Adopter Categories

Rogers (1995) defines five categories of adopters: innovators, early adopters, early majority, late majority, and laggards. These categories, derived by partitioning a continuum of innovativeness, represent variability about the mean (i.e., when half of the population has adopted an innovation). The innovators and early adopters together represent 16% of all adopters. The early majority represents the next 34% of the population, followed by the 34% called the late majority. The final 16% of all adopters are laggards. Rogers (1995) acknowledges that this categorization does not explicitly include non-adopters since there is an implicit assumption that everyone will adopt, eventually.

The current level of diffusion of home PCs, which is estimated to be approximately 33-37% (Kraut 1996; Kraut, et al. 1996), will therefore include innovators, early adopters, and part of the early majority. Given the six-month window between the initial and follow-up surveys in this research, we use the terms "early adopters" refer to those who had already adopted at the time of the initial survey (i.e., the innovators, early adopters, and part of the early majority), "later adopters" to refer to those who have not yet adopted but intend to do so in the six month time frame of this research (i.e., an additional part or even the rest of the early majority), and "non-adopters" to refer to those who have not yet adopted, and have no plans to do so in the time frame of this research (i.e., the late majority and laggards). Figure 1 presents the technology adoption life cycle, the adopter categories of Rogers (1995) and the corresponding categories used in this research.


Figure 1: Adopter Categories in the Current Study and Rogers (1995)

a It was not possible, a priori, to separate later adopters from non-adopters. But, based on the data, we will be able to distinguish later adopters from non-adopters.


Psychographic Differences Across Adopter Categories

Prior research on behavior in general, and technology adoption in particular, has tended to assume that there is a common set of determinants of behavior among various segments of the population (e.g., Davis, et al. 1989; Fishbein and Ajzen 1975). Innovation diffusion theory shares this basic assumption and also assumes that everyone will eventually adopt (Rogers 1995). More specifically, the same set of innovation characteristics, such as those identified by Davis, et al. (1989), Moore and Benbasat (1991), and Tornatzky and Klein (1982) (e.g., relative advantage, compatibility), is expected to influence adoption of an innovation across all adopter categories. Emerging views in the business press challenge these two basic assumptions underlying much technology adoption research. For example, Moore (1991) suggests that people in each of the different adopter categories are systematically different from those in the category to their immediate left. These differences across categories are referred to as "cracks in the bell curve" (Moore 1991). This view suggests that innovations that succeed among innovators and/or early adopters may fail among the early majority or late majority since the innovation does not possess the characteristics that appeal to those in these later categories. Further, this implies that not only will factors influencing different categories of adopters be systematically different from each other but also, since not everyone will adopt an innovation, it is quite likely that factors influencing non-adoption will be different from factors influencing adoption.

From a scientific standpoint, very little evidence exists to substantiate claims of psychographic differences across adopter categories and non-adopters. Recently, Agarwal and Prasad (1998) presented personal innovativeness, a possible construct that allows categorization of individuals into different adopter categories, as a moderator of the relationship between user beliefs and intention to adopt. Though they found only partial support for the relationship, the ideas are consistent with marketing research that has identified different profiles of consumer innovators (e.g., Dickerson and Gentry 1983; Holak 1988). Given the current state of knowledge, this research will first seek to theoretically explain potential differences in salient determinants among the different adopter categories used in this research (early adopters, later adopters, and non-adopters). Subsequently, based on the results of the empirical investigation, we will attempt to explain the determinants of adoption as they relate to the different adopter categories of Rogers (1995).

Utilitarian and Hedonic Outcomes

Technology adoption decisions have been typically characterized by a strong productivity orientation. Across the different models, validated and applied in student and workplace settings, constructs related to the use-productivity contingency (e.g., perceived usefulness, relative advantage, job fit, etc.) have emerged as the strongest predictors of adoption and usage behavior (Adams, Nelson, and Todd 1992; Agarwal and Prasad 1997, 1998; Chin and Todd 1995; Davis 1989, 1993; Davis and Venkatesh 1996; Gefen and Straub 1997; Igbaria, Zinatelli, Cragg, and Cavaye 1997; Mathieson 1991; Segars and Grover 1993; Trevino and Webster 1992; Subramanian 1994; Szajna 1994, 1996; Taylor and Todd 1995; Thompson, et al. 1991; Venkatesh and Davis 1996). The current research adapts this rational basis for technology adoption and usage to the context of households via the construct, "utilitarian outcomes," defined as the extent to which using a PC enhances the effectiveness of household activities.

In addition to utilitarian outcomes, we expect household adoption of PCs to be influenced by hedonic outcomes. Consumer behavior research describes hedonic outcomes as the pleasure derived from the consumption, or use, of a product (Babin, Darden, and Griffin 1994; Hirshman and Holbrook 1982; Holbrook and Hirshman 1982). In contrast to the workplace, in a household setting, the entertainment potential of PCs is expected to have a strong influence on the adoption decision. In fact, PC applications have evolved over the past decade, providing more and more opportunities to play with technology. Similar to video games (e.g., Nintendo), PC games (e.g., MYST) are entertaining and render PC use enjoyable (Davis, Bagozzi, and Warshaw 1992; Holbrook, et al. 1984; Malone 1981). Further, they provide an opportunity to escape reality and become absorbed in a new world, thus exhibiting characteristics consistent with a hedonic perspective (Foxall 1992; Lacher and Mizerski 1994). Dating back to the last decade, the importance of hedonic outcomes as a determinant of PC adoption is also borne out by a decline in people’s use of other competing forms of entertainment and fun such as radio, movies, fiction reading, and social activities (Kohut 1995; Robinson and Godbey 1997; see also Vitilari, Venkatesh, and Gronhaug 1985).

The role of utilitarian and hedonic outcomes is also supported by motivation theory. Motivation research suggests that there are two main classes of motivation: extrinsic and intrinsic. Extrinsic motivation pertains to achievement of a specific goal whereas intrinsic motivation is the pleasure and satisfaction derived from a specific behavior (Deci and Ryan 1980; Vallerand 1997). A significant body of research has established extrinsic and intrinsic motivation as primary drivers of behavior in several domains (see Vallerand 1997 for a review). In the context of PC adoption in homes, utilitarian outcomes relate closely to extrinsic motivation, and hedonic outcomes relate closely to intrinsic motivation.

We expect utilitarian outcomes and hedonic outcomes to have a differential impact on early and later adopters. Recent technology adoption research suggests that personal innovativeness, defined as the willingness of an individual to try out a new information technology, will moderate the relationship between perceptions about technologies and intentions to use them (Agarwal and Prasad 1998). Given that approximately one-third of households have adopted PCs, it could be argued that the households that have already adopted are more "innovative" than those that have yet to adopt a PC (Agarwal and Prasad 1998; Rogers 1995). Thus, we expect early adopters to be more strongly influenced by hedonic outcomes, while later adopters will be more strongly influenced by utilitarian outcomes.

Social Outcomes

Social outcomes can be thought of as the social rewards (e.g., "public" recognition) that would be achieved as a result of adopting an innovation (Fisher and Price 1992). This may lead to an elevation in power, knowledge, and/or status if the decision is thought by others to be a good one. Prior research has emphasized the importance of social outcomes as a determinant of behavior (e.g., McCracken 1988). Similarly, innovation literature suggests that the desire to gain status is an important reason for the adoption of an innovation (Rogers 1995). Although PCs have been in existence in one form or another since the 1970s, they are still relatively new to the household arena. In the context of household adoption of PCs, adopters, particularly early adopters, achieve a certain status for being among the first to adopt a PC. They can serve as role models for later adopters by choosing wisely among their innovation options. Their role is to adopt an innovation, evaluate it, and communicate their evaluation to members of a social network, thus decreasing uncertainty for others (Rogers 1995). Clearly, as more and more people adopt an innovation, its status value diminishes. Thus, as followers, later adopters are not likely to place as much emphasis on such social/status outcomes. In sum, we expect social outcomes to be a significant determinant for early adopters but not for later adopters.

Social Influences

Social influences are the extent to which members of a social network influence one another's behavior (Rice, Grant, Schmidt, and Torobin 1990). This influence is exerted through messages and signals that help to form perceptions of the value of a product or activity (Fulk and Boyd 1991; Fulk, Steinfield, Schmitz, and Power 1987; Salancik and Pfeffer 1978). The role of social influence has been well documented in prior research in social psychology and marketing. Research in psychology emphasizes and demonstrates the importance of the role of social influences on behavior (e.g., Ajzen and Fishbein 1980; Fishbein and Ajzen 1975; Ajzen 1985, 1991; Triandis 1977). In IS research, the evidence is mixed regarding the role of social influences on technology adoption in the workplace (e.g., Davis, et al. 1989; Mathieson 1991; Taylor and Todd 1995). While the importance of utilitarian outcomes (such as productivity gains from technology use) diminishes the importance of social influences in the workplace, household decisions tend to be characterized by more of a normative orientation (e.g., Burnkrant and Cousineau 1975). Thus, we can expect household PC adoption decisions to be influenced by the views of relevant others, such as friends and family members (Brunkrant and Cousineau 1975; Childers and Rao 1992; Fisher and Price 1992; Miniard and Cohen 1979).

In trying to understand the differences in the role of social influences among early and later adopters, recall that early adopters are driven strongly by intrinsic motives (i.e., hedonic outcomes and social outcomes). Given that they are the "first ones on the block to adopt," their role is to serve as influencing agents of those who come later. This is consistent with Rogers’ (1995) idea that early adopters convey their evaluation via social networks. Thus, early adopters are not expected to be driven by social influences, but are expected to exert influence on later adopters. However, later adopters being followers can be expected to wait for PCs to be well-settled in the marketplace and supported by positive "word of mouth" from friends and family and other early adopters. Therefore, we propose that the opinions of relevant others (e.g., friends and family) will have a strong influence on later adopters but less of an influence on early adopters.

Barriers to Adoption

Psychology research has shown that the presence of constraints can inhibit both the intent to perform a behavior and the behavior itself (see Ajzen 1991 for a discussion). Specifically, IS research has identified knowledge and resources as barriers to technology adoption intent (e.g., Mathieson 1991; Taylor and Todd 1995) and actual usage behavior (Taylor and Todd 1995). Two specific barriers (constraints), namely knowledge and price (resources), are expected to be relevant in the context of PC adoption in homes. Consistent with prior research, in the context of PC adoption in homes, not possessing the requisite knowledge to use a computer will significantly inhibit adoption. Also, the role of resources as a barrier is expected to be significant given evidence from prior marketing research (e.g., Erickson and Johansson 1985) that price is a factor in many consumer decisions, especially in the case of expensive goods (e.g., Sahni 1994) such as PCs that cost at least $1,000, about five times the cost of many other consumer durable goods such as TVs, VCRs, etc. There is some evidence to suggest that barriers will be more salient when the individual has less control over the barriers (e.g., Schifter and Ajzen 1985), thus indicating a possible differential role for barriers among adopters and non-adopters.

There is general agreement that early adopters tend to be more well educated than later adopters (Brancheau and Wetherbe 1990; Rogers 1995) and more affluent than later adopters (e.g., Feldman and Armstrong 1975; LaBay and Kinnear 1981; Rogers and Shoemaker 1971). Early adopters, who are more well educated, are likely to have had more exposure to computers in general, and thus are potentially more computer-literate than later adopters. Based on the asymmetrical role of barriers, we expect early adopters are not likely to find knowledge to be a significant barrier to adoption, but later adopters and non-adopters will be significantly constrained by knowledge or lack thereof. Similarly, cost will be less consequential to the more affluent (i.e., early adopters) but will be a significant barrier for later adopters, and almost prohibitive for non-adopters. Consistent with the theory of planned behavior (Ajzen 1991), barriers are expected to have an impact on the intent to adopt and actual adoption behavior.

Dependent Variables: Intention and Behavior

Much prior research in IS, innovation diffusion, psychology, and marketing has relied on intention and behavior as key dependent variables (e.g., Agarwal and Prasad 1998; Ajzen 1991; Morwitz and Schmittlein 1992; Taylor and Todd 1995). Intention mediates the effect of key determinants on behavior. In addition to intention, barriers have a direct effect on behavior. As mentioned earlier, prior research has found intention-behavior correlations from .18 to .84 across a wide range of behaviors (see Ajzen 1991 for a review; see also Ajzen 1988; Ajzen and Fishbein 1980; Canary and Seibold 1984; Sheppard, Hartwick, and Warshaw 1988). Based on a meta-analysis of 87 studies, Sheppard, et al. (1988) found a correlation of about .50. Further, constraints and opportunities also have a direct effect on behavior, with correlations ranging from .20 to .78 (see Ajzen 1991 for a review). Typically, intentions predict behavior quite well unless there are constraints beyond the individual’s control that completely overshadow intention (e.g., Schifter and Ajzen 1985). Consistent with the overall pattern of findings in psychology and marketing research, IS research provides evidence to suggest that intention is a fairly good predictor of usage behavior (Davis, et al. 1989; Taylor and Todd 1995) and issues of control have a direct effect over and above intention (Taylor and Todd 1995).

Summary

In this research, we seek to understand and explain PC adoption in homes using the determinants proposed. In addition to its effect on adoption intent, barriers are expected to have a direct effect on adoption behavior. Based on the earlier discussion, a summary of the factors thought to be most relevant for each adopter category is presented in Table 1. As can be seen in Table 1, adopters and non-adopters are not simply at opposite ends of the same continuum of determinants. In other words, the factors salient to early adopters in their decision to adopt are not the same factors salient to later adopters or non-adopters. For early adopters, we believe that hedonic and social outcomes will be most influential in the adoption decision. For later adopters, we believe that utilitarian outcomes and social influences will be most salient. Finally, for non-adopters lack of knowledge and cost will be the key barriers influencing the decision.


Table 1: Factors Proposed as Most Significant by Adopter Category


  Early Adopters Later Adopters Non-Adopters
Utilitarian Outcomes N Y N
Hedonic Outcomes Y N N
Social Outcomes Y N N
Social Influences N Y N
Lack of Knowledge N Y Y
High Cost N Y Y
Y = expected to be significant
      N = not expected to be significant

METHODOLOGY

A market research firm, with no prior affiliation with the authors, participated in the design and conduct of the study. A market research firm was employed for three key reasons: (1) to ensure an unbiased design and execution of the study consistent with industry-wide market research standards, (2) the large scale and scope of the project required significant amount of resources in terms of skilled individuals (e.g., interviewers) and equipment (e.g., random phone dialer) and (3) the researchers were not specifically qualified to hire interviewers, train them effectively, and monitor the interview process per se. The study was designed to capture a cross-sectional snapshot and a dynamic longitudinal picture of the underlying phenomena. In order to accomplish this, data were collected in two waves that were six months apart. In Phase 1, data was collected from over seven-hundred households about their PC adoption and usage behavior. Six months later, in Phase 2, we attempted to contact all Phase 1 respondents for a follow-up survey to understand their changing views and follow-up behavior pattern.

A variety of techniques are available to capture consumer perceptions and attitudes. In order to obtain a nationally representative sample of respondents within a limited time frame and budget, mail and telephone surveys were the most appropriate options. Telephone interviews lend themselves to more flexibility and are equal, if not superior, to other active methods of data collection (Brock 1986; Rogers 1976; Tyebjee 1979). Further, telephone interviews reduce the amount of bias normally associated with an in-person interview due to the elimination of visual cues (Biscomb 1986; Morton and Duncan 1978). Also, the potential for an increased response rate made the telephone interview more appealing, as response rates for random mail surveys tend to be in the 10 to 15 percent range (Steeh 1981). Finally, the ability to elicit open-ended responses and to follow up on those responses using a two-stage questioning methodology allowed us to obtain factors that were not constrained by a priori identification of constructs as in traditional survey research (Babbie 1990; Stone 1978), and also determine the importance of each of the factors.

One drawback of the telephone interview is the potential for nonresponse errors due to inadequate sampling frame specification. In other words, while the population is American households, the sampling frame will include only those households with telephones. Some telephone survey research omits all individuals not listed in telephone directories; the current study has overcome this problem by using random digit dialing (discussed further below). Thus, while some households are omitted from the study (i.e., those without telephones), the impact of their omission will be minimal compared to the increased overall response rate.

Instrument Development

The instrument used in this research featured two broad categories of questions regarding: (a) factors related to PC adoption and usage and (b) demographic variables. Open-ended questions were written to elicit factors driving adoption decisions, future adoption intent, and usage behavior. The questions were evaluated by experts and peers, and minor modifications were made based on their comments and suggestions. A two-stage questioning technique was employed to elicit the importance of each factor. For example, in the first stage, the respondent was asked about the key factors influencing their decision, and if the respondent indicated that 'entertainment' was a factor driving purchase intent, the second stage of questioning asked how important (on a 5-point scale, where 1 was not at all important and 5 was extremely important) 'entertainment' was as a factor. All responses were handled in the same manner, regardless of how extensive of an answer the respondent gave. Additionally, demographic variables were captured using a categorization scheme consistent with the Census Bureau (Day 1996).

Pilot Study

A pilot study was conducted in a large metropolitan area in the Midwest. Participants were randomly selected from the city's telephone directory. The pilot study was used to conduct a preliminary test of the instrument, to solicit comments and suggestions about the instrument from respondents, and to assess the duration of a typical interview. Sixty households completed the pilot study. The instrument was refined following the pilot study to reduce the duration of the typical interview to about 8 minutes in order to minimize refusal rate (see Dillman 1978).

Sample

The sample of households was chosen through the use of a random digit dialing technique. First, the random number generator module of the system generated a random area code from among the valid area codes in the U.S. Next, a random seven-digit number was generated. These two steps were repeated to generate about 15,000 numbers. Finally, the numbers generated were forwarded to a centralized system that accessed and dialed numbers when prompted by the interviewer's terminal. The interviewers were not aware of the specific number dialed. Though this method increases the probability that a dialed number will not belong to a member of the sampling frame (i.e., American households), it was preferred over national telephone directories since it increases the probability of contacting typically underrepresented populations, such as those with unlisted telephone numbers and new listings (Frey 1989). This method of sample selection has been employed in recent surveys by industry leaders such as Nielsen and Times-Mirror (see Kohut 1995; Robinson and Godbey 1997).

Data Collection

Over one-thousand households were contacted and the primary decision maker in the household was invited to participate in the voluntary phone survey during a three-week window in March/April 1997 (Phase 1). The follow-up survey (Phase 2) was completed six months later. The interviews were conducted in different time slots during the day and evening, from 9 a.m. to 8:30 p.m. Monday through Saturday. On average, an interview took just under 9 minutes to complete and interviewers produced an average of 5.1 responses per hour. Twelve interviewers were assigned to this project. They had an average of 3.2 years of interviewing experience, including at least six months of telephone interviewing. They received two eight-hour days of training on this particular instrument. As suggested by Lavrakas (1993), the following monitoring activities were conducted to ensure the validity of the data: (a) supervisors randomly monitored actual interviews and (b) respondents were contacted at random to verify participation in the interview.

Data Coding Procedure

Two individuals, knowledgeable in coding techniques, were employed to code the open-ended responses. The coders were given a brief overview of the instrument and response types. As suggested by Miles and Huberman (1994), the coders were given definitions from prior research for the key constructs: utilitarian outcomes, hedonic outcomes, social influences, social outcomes, and barriers. The coders were instructed to hold out responses that did not seem to fit these definitions. An initial pool of 30 interview responses was chosen at random to be coded. The results of the preliminary analysis were carefully studied by one of the authors, and found to be quite accurate. Following discussion of discrepancies between coders, further clarifying instructions were given. Additionally, items that did not fit into the initial set of definitions were further analyzed. Consistent with Miles and Huberman's (1994) recommendations, this resulted in minor refinement of definitions and additional constructs being identified. Once this preliminary phase was completed, the remaining data were coded by both coders. The entire coding process took six months to complete (i.e., though May 1998).

Statistical Analysis Procedure

We performed some basic analyses to compute the response rate in Phase 1, and the follow-up response rate in Phase 2. The next step was to assess the representativeness of the sample. We determined the characteristics of the U.S. population using the U.S. census data from 1990, adjusted for 1997 using the projections provided by an official report from the Census Bureau (Day 1996). Given that all the categories employed in this research were consistent with the Census Bureau categories, we generated corresponding percentages for the sample.

The open-ended questioning allowed us to gather information from the respondents in an unbiased and non-leading way. The two-stage questioning method complemented the open-ended questions since it allowed us to determine the valence of each of the individual responses (factor affecting adoption/usage) and thus, provided the means to perform statistical analyses on the data. For each of the different salient factors, response frequencies, and means, and standard deviations for the valences were calculated.

RESULTS

Response Rate Analysis

The random digit dialing technique employed in this research caused the sampling frame to include phone numbers that did not necessarily belong to households (e.g., business phones, faxes, etc.). Of the 15,007 phone numbers we attempted to contact in Phase 1, 12,271 were valid phone numbers (i.e., excluding disconnected and invalid numbers). Of these 12,271 numbers dialed, we reached 988 households, 10,145 other numbers (e.g., business phones, faxes, cell phones, pagers, voice-mail, fax, etc.) and received no answer on 1,138 numbers even after four callbacks. Of the 988 households, we were able to contact 743 (75% percent) participated in the study, and 733 completed the entire survey and provided usable responses for an overall response rate of 74.2%. The three components of non-response error in this case are those who could not be reached, those who refused to participate, and those who started but did not complete the survey. While we had little control over these situations, the design and implementation of the study attempted to minimize potential problems of non-response with up to four follow-up calls made to reach numbers that could not be reached when first called. Interviewer effects were assessed by comparing responses across the twelve interviewers. No significant differences were found across the interviewers. Additionally, no patterns were found within interviewer responses.

In Phase 2, a follow-up call was made to all respondents from Phase 1. The overall follow-up response rate was 87.9%. Respondents participating in Phase 1 but not in Phase 2 was attributable to a two primary reasons: inability to reach respondents (e.g., invalid phone number perhaps due to change of residence and no forwarding information, no response or voice-mail despite up to 4 callbacks) and unwillingness to participate.

Sample Characteristics

We examined the characteristics of the sample in order to ensure that the sample was truly representative of American households. Using the procedure described earlier, the population and sample characteristics were computed. Table 2 presents a breakdown of the sample in terms of different characteristics, and compares the corresponding characteristics of the population derived from census data. A comparison of the characteristics of the sample and population indicate that the random sample of households included in this study was highly representative of the population, suggesting that the results from this research are likely to generalize to the population of American households.


Table 2:
Population and Sample Characteristics


Households
Population Characteristics
Sample Characteristics
Family 
     -Married Couples
     -Female – No Husband
     -Male – No Wife
69%
52%
13%
4%
70%
50%
14%
6%
Nonfamily 
     -Female
     -Male
31%
17%
14%
30%
17%
13%
Racial Background
Population Characteristics
Sample Characteristics
White 
       -White -not Hispanic
       -Hispanic
84%
76%
8%
88%
78%
10%
Black 13% 9%
Asian/Pacific Islander 3% 3%
Householder Age
Population Characteristics
Sample Characteristics
15-24 5% 7%
25-34 19% 20%
35-44 24%  25%
45-54 19% 18%
55-64 12% 10%
65+ 21% 20%
Nativity
Population Characteristics
Sample Characteristics
Native U.S. 89% 90%
Foreign Born 11% 10%
Not a Citizen 6% 5%
Region
Population Characteristics
Sample Characteristics
Northeast 19% 20%
Midwest 24% 26%
South 35% 34%
West 22% 20%
Residence
Population Characteristics
Sample Characteristics
Inside Metro Areas 
      -Inside Central Cities
      -Outside Central Cities
80%
31%
49%
76%
39%
37%
Outside Metro Areas 20% 24%
Earnings of Full-time Year Round Workers ($’s)
Population Characteristics
Sample Characteristics
Male 31,864 31,003
Female 24,272 23,650
Households
Population Characteristics
Sample Characteristics
Female – No Husband 13%  14%
Male – No Wife 4% 6%
Nonfamily 31% 30%
Female 17% 17%
Male 14% 13%
Per Capita Income ($’s)
Population Characteristics
Sample Characteristics
All Races  18,545 17,988
White 19,524 19,003
Black 12,523 12,071
Asian/Pacific Islander 18,830 18,225
Hispanic 10,544 10,009

           Notes:

     
    1. All categories used are consistent with those used by the U.S. Census Bureau.
    2. Population characteristics reflect U.S. Census Data for 1990, adjusted for the year 1997 using Day (1996).

Characteristics of Adopters and Non-Adopters

About 33% of all households sampled possessed a PC in Phase 1, indicating that these households were largely early adopters and part of the early majority per Rogers (1995). Innovation research suggests that there are systematic differences between early adopters and late adopters (Rogers 1995). Specifically, early adopters are younger (Assael 1981), more educated, and tend to be more affluent (Rogers 1995) than later adopters. Upon comparing adopters and non-adopters on the basis of these characteristics (Table 3), the results are consistent with prior research (Assael 1981; Rogers 1995).



 
 

Table 3:
Demographic Characteristics of Adopters and Non-Adopters

  Adopters Non-Adopters
Age 32.3 45.1
Education Level About 80% of all heads of
households had graduated college
Nearly 70% of all heads of households 
had graduated high-school, but not college.
Household Income $62,300 $38,442

Data Coding Results

Per the procedures discussed earlier, the data were carefully coded. The inter-coder reliability was .89, which is well above the minimum of .70 identified by Miles and Huberman (1994). As a final step in the coding process, the few remaining discrepancies were resolved by the researchers. The results from the coding process yielded a core set of factors driving adoption intent and behavior. Further, the coding also suggested that most of the major categories of determinants identified in the theoretical development were multi-dimensional. Table 4 presents each of the determinants and their respective dimensions. The main construct is listed and the number in parentheses provides a notation for each dimension. Specifically, there were four dimensions of utilitarian outcomes, one dimension of hedonic outcomes, two dimensions of social outcomes, two dimensions of social influences and five dimensions of barriers. The term "Barriers" was chosen for the five dimensions listed because each of the five was a hurdle to adoption that would not directly contribute to any relevant individual or household outcomes. Thus, while the overarching categories were chosen based on prior research, the breakdown into more detailed dimensions was possible due to the use of open-ended questions. In fact, the data coding process helped identify dimensions that had not been accounted for in prior research, providing further support that the use of open-ended items helped to overcome a priori expectations, resulting in a more complete understanding of the phenomenon.


Table 4

Factor Defintion Code Detailed Factor
Hedonic Outcomes The pleasure derived from PC use H1 Applications for fun
Utilitarian Outcomes The extent to which using a PC enhances the effectiveness of household activities U1 Applications for personal use
U2 
Utility for children
U3 
Utility for work-related use
U4 
Reduced utility due to obsolescence of current PC
Social Outcomes The change in status that coincides with a purchase decision SO1 Status gains from possessing current technology
SO2 Status losses due to obsolete technology at home
Social Influences The extent to which members of a social network influence one another's behavior SI1 Influences from friends and family
SI2 Influence of information from secondary sources (such as news on TV, newspaper, etc.)
Barriers Factors inhibiting adoption B1 Rapid change in technology, and/or fear of obsolence
B2 Declining cost
B3 High cost
B4 Ease/difficulty of use
B5 Requisite knowledge for PC use

PC Adoption: Current Level and Future Projections

Using the data collected in Phase 1, we examined the frequency breakdown of households based on whether or not they possessed a PC. Also, households without PCs were asked about their intent to purchase a PC, and households with PCs were asked about their intent to purchase another PC. Table 5 summarizes these results. The results indicated that a third of all households owned a PC, suggesting a certain extent of stagnation given the results from earlier surveys (in 1995 and 1996). However, given the potential for some sampling error, we do not want to read too much into possible stagnation. On average, households with PCs had owned their current PC for just under two years. Interestingly, nearly two-thirds of those households intended to purchase another PC. Further, about 23% of households not possessing a PC currently (16% of the total sample) intended to purchase in the next six months, suggesting that the number of households with PCs would grow from 33% to almost 50% at the time of the follow-up survey (six months later).


Table 5.  Frequency Breakdown of Households With and Without PCs,
and Their Future Purchase Intent  (n = 733)

Table 5A Households With a PC
(n = 245)

Intend to purchase another PC?  
Yes
In < 6 months:  136
In > 6 months:    24
No 40
Don&rsquo;t know
45
Table 5B Households Without a PC
(n = 488)
Intend to purchase another PC?  
Yes
In < 6 months:  114
In > 6 months:      0
No 304
Don&rsquo;t know
70

Drivers of Adoption and Non-Adoption

Factors Affecting Current Purchase Decision

We conducted a detailed cross-sectional analysis of the determinants of purchase behavior, based on the initial survey (Phase 1) (Table 6). As expected, hedonic and social outcomes were the strongest determinants of purchase behavior of current users (early adopters). In contrast to expectations, the results suggested social influences and utilitarian outcomes were also significant determinants of purchase behavior. Specifically, status impact from possessing current technology was most important, followed by applications for fun, the influence of friends and family members, and applications for personal use. For the non-adopters, social influences and barriers were most significant. Specifically, we identified four drivers of non-adoption: information from secondary sources such as TV or newspapers, rapid change in technology, high cost, and lack of knowledge. As proposed, the factors influencing non-adoption are not simply the lack of factors favoring adoption, suggesting that the underlying decision-making processes of adopters and non-adopters do not lie on opposite ends of the same continuum. Clearly, barriers to entry are more salient to non-adopters, whereas among adopters other factors are salient possibly because adopters have scaled the barriers. The results indicated that the factors driving use among early adopters (Table 7) were somewhat different from the reported reasons for adoption (see Table 6). Interestingly, utilitarian outcomes, which was a marginal factor in the initial purchase decision, had become a driver of usage behavior, suggesting that people may be "discovering" uses for the acquired product.


Table 6.
Factors Affecting Current Purchase Decision (n = 733)



  Adopters
(n =245)
Non-Adopters
(n = 488)
Frequency
M  (SD)
Frequency
M  (SD)
Utilitarian Outcomes
U1: Applications for personal use
37
3.5  (0.7)    
Hedonic Outcomes
H1: Applications for fun
110
3.8  (0.6)    
Social Outcomes
SO1: Status gains (current technology)
146
4.1  (0.6)    
Social Influences
SI1: Friends and family
62
3.7  (0.5)    
SI2: Secondary sources    
320
4.4  (0.6)
Barriers
B1: Rapid change in technology    
287
3.8  (0.5)
B3: High cost     187 3.7  (0.7)
B5: Requisite knowledge     197 3.8  (0.6)

Table 7.
Factors Affecting Current Purchase Decision (n = 245)


 
Frequency
M  (SD)
Utilitarian Outcomes
     U1: Applications for personal use
169

3.9  (0.7)
Hedonic Outcomes
    H1: Applications for fun
112
4.0  (0.8)
Social Outcomes    
Social Influences    
Barriers    


Factors Affecting Future Purchase Intent

We examined factors influencing future purchase intent among (current) non-adopters (Table 8). As mentioned earlier, among the 488 households without PCs, 114 (about 23%) intended to purchase in less that six months (i.e., later adopters), while the rest did not intend or reported "don't know." Consistent with expectations, utilitarian outcomes (applications for personal use) were the key drivers of intenders. In contrast, those who did not intend to adopt were driven primarily by barriers, most notably rapid change in technology.


Table 8.
Factors Affecting Future Purchase Intent Among Non-Adopters (n = 488)


  Intenders
(n =114)
Non-Intenders
(n = 304)
Frequency
M  (SD)
Frequency
M  (SD)
Utilitarian Outcomes
U1: Applications for personal use
100
4.1  (0.6)
43
3.8  (0.5)a
Hedonic Outcomes
         
Social Outcomes
         
Social Influences
SI1: Friends and family
68
3.7  (0.8)    
Barriers
B1: Rapid change in technology    
270
4.0  (0.7)
B2: Declining cost
47
3.8  (0.6)    
B5: Requisite knowledge
51
3.8  (0.5) 37 3.7  (0.6)

a In the case of non-intenders, they reported a lack of application for personal use (i.e., lack of utilitarian outcomes).

Note: The table reports an analysis of intenders and non-intenders. Those who responded with a "Don&rsquo;t Know" (n = 70) were not asked about specific factors due to the uncertainty conveyed by the response.


PC Purchase Behavior: Six Months Later

To understand PC purchase behavior, the data were partitioned into three categories based on intentions expressed in Phase 1: (a) those who intended to purchase, (b) those who intended to not purchase, and (c) those who were uncertain. Among the 488 households in Phase 1 that were non-adopters, we successfully collected responses from 435 households in Phase 2, resulting in a follow-up response rate of 89%. Table 9 presents a frequency breakdown of follow-up purchase behavior among those who intended in Phase 1, and an analysis of the factors driving the follow-up behavior. Only 46 out of 107 (43%) intenders actually purchased a PC in the six-month time frame. Among those who intended to purchase (in Phase 1) and did actually purchase (by Phase 2), the factors influencing the purchase decision were not entirely consistent with those reported in Phase 1 (see Table 8), though being a recipient of a "gift" helped scale the barrier of high cost. Those who did not adopt feared rapid change in technology, coupled with declining costs. Only a very small percentage of those who did not intend in Phase 1 (1%) and reported "Don't Know" in Phase 1 (11%) actually purchased a PC in the six-month period leading up to the follow-up survey (Tables 10 and 11). Thus, a large percentage of the non-adopters (in Phase 1) conformed with their intent of not purchasing a PC in the six-month time frame leading up to Phase 2. Consistent with the reasoning reported in Phase 1 (see Table 6), the key barrier reported in Phase 2 was the rapid change in technology.


Table 9.
PC Purchase Behavior Among Intenders (n = 107a)

Purchased (n = 46)
Did Not Purchase (n = 61)


  Purchased
(n =46)
Did Not Purchase
(n = 61)
Frequency
M  (SD)
Frequency
M  (SD)
Utilitarian Outcomes
U2: Utility for Children
13
4.1  (0.9)    
Hedonic Outcomes
         
Social Outcomes
SO1:  Status gains (current technology)
29
3.9  (0.5)    
Social Influences
SI1: Friends and family
28
3.9  (0.5)    
Barriers
B1: Rapid change in technology    
48
4.1  (0.5)
B2: Declining cost
47
3.8  (0.6) 37 3.7  (0.7)
B3: High costb
25
4.0  (0.6)    

a In Phase 2, we reached 107 out of the 114 intenders in Phase 1.

bSome households scaled the "high cost" barrier via a "gift" from a close friend or family member.
 
 

Table 10.
PC Purchase Behavior Among Non-Intenders (n = 284a)


  Purchased
(n =4b)
Did Not Purchase
(n = 280)
Frequency
M  (SD)
Frequency
M  (SD)
Utilitarian Outcomes
U1: Applications for personal use    
68
3.2  (0.7)c
Hedonic Outcomes
         
Social Outcomes
         
Social Influences
         
Barriers
B1: Rapid change in technology    
208
4.0  (0.8)
B2: Ease/difficulty of use     27 3.8  (0.6)

a In Phase 2, we reached 284 out of the 304 non-intenders in Phase 1.
b The sample size did not allow for an accurate assessment of key factors.
c In the case of those who did not purchase, they reported a lack of application for personal use (i.e., lack of utilitarian outcomes).
 
 

Table 11.
PC Purchase Behavior Among Those Who Were Uncertain (n = 44a)


  Purchased
(n =5b)
Did Not Purchase
(n = 39c)
Frequency
M  (SD)
Frequency
M  (SD)
Utilitarian Outcomes
         
Hedonic Outcomes
         
Social Outcomes
         
Social Influences
         
Barriers
B1: Rapid change in technology    
32
4.0  (0.7)

a In Phase 2, we reached 44 out of the 70 households that were uncertain in Phase 1.
b The sample size did not allow for an accurate assessment of key factors.
c Factors that amounted to frequencies less than 10 are omitted from the table.


Based on their intention in Phase 1 and follow-up behavior reported in Phase 2, Table 12 summarizes the intention-behavior conformance among intenders, non-intenders, and those who were uncertain. Clearly, the pattern of conformance is nearly perfect among non-intenders, and nearly 90% of those who were uncertain chose not to adopt. However, less than half of all intenders conformed with original plan/decision.


Table 12. Intention-Behavior Relationship

Adoption Behavior

Adoption Intent Reported in Phase 2

Phase 1
Yes 
No Conformance
Intenders
(n = 107)
46 61 43%
Non-Intenders
(n = 284)
4 280 99%
Uncertain ("Don&rsquo;t Know")
(n = 44)
5 39 89%a

Note: The above analysis is based on responses gathered in both phases of data collection. As can be expected, the response rate in the follow-up survey was lower than the initial survey, which was 114, 304, and 70 in each of the three categories respectively.

a This percentage was calculated by treating the uncertainty as an intent to not adopt.


Mapping Results On To Rogers&rsquo; (1995) Categories

Given the level of diffusion of PCs at Phase 1 (about 33%) and the six-month window associated with the surveys in this research, we used three categories of adopters: early adopters (i.e., those who had already adopted a PC at the time of Phase 1, thus including innovators, early adopters, and part of the early majority), later adopters (i.e., those who had not adopted at the time of Phase 1 but intended to adopt in the six-month time window of this research, thus including the rest of the early majority), and non-adopters (i.e., those who had not adopted at the time of Phase 1 and did not intend to adopt in the six-month time window of this research, thus including the later majority and laggards). In this section, we report the salient determinants of adoption in the different adopter categories specified by Rogers (1995).

In order to accomplish this, we examined the data associated with those who had already adopted at the time of Phase 1 (see results presented in Table 6). The adopters (n = 245) formed about 33% of the sample. The data was then partitioned on the basis of how long households owned a PC in general (not just the current PC). The first 117 households, which equaled 16% of the sample, represented the innovators and early adopters per Rogers (1995). Nearly all of these households indicated hedonic outcomes and social outcomes were important factors influencing their adoption decision. Interestingly, there were no significant psychographic differences between innovators and early adopters.

Next, we proceeded to understand the key determinants influencing the early majority, the next 34% of the sample. This early majority comprised the remaining 128 (out of 245) households from the previous analysis, and also included all intenders (n = 114) from Phase 1 (see results in Table 8), yielding a total of 242 households (approximately the next 34% of the sample). Thus, these households would be consistent with Rogers&rsquo; (1995) definition of what will serve as the early majority. Adoption among over half the households in the early majority was determined by utilitarian outcomes and/or social influences. Hedonic outcomes and social outcomes were relevant only to about 40 out of the 242 households, and its role was significant only among the very early part of the early majority, suggesting that they (the very early part of the early majority) shared some of the psychographics of the early adopters.

The late majority and laggards comprised those who did not intend to adopt a PC or were uncertain about adoption during the initial survey (Phase 1). While, as expected, these households did not adopt a PC in the six-month time frame of this research, this meant a clear separation of the late majority from the laggards was not possible. However, one barrier emerged very strongly across these non-adopters in Phases 1 and 2 &ndash; the rapid change in technology and the consequent fear of obsolescence. Of the different factors influencing the earlier categories of adopters, the only factor relevant to the non-adopters (i.e., later adopters and laggards) was the lack of utilitarian outcomes.

Figure 2 pictorially presents a summary of the mapping of the results from the present work on to the adopter categories specified by Rogers (1995). Further, the overall pattern of results was also supported when statistically controlling for age, income, and education.


Figure 2: Mapping Adopter Determinants to Adopter Categories of Rogers (1995)


DISCUSSION

The longitudinal study with two waves of measurement, an initial survey and a follow-up survey six months later, yielded several important and interesting insights. First, it helped identify several dimensions within each construct included in the theoretical development, thus leading to a richer and more complete understanding of PC adoption in American homes. Second, contrary to much prior research investigating behavior in general and PC adoption in particular, but consistent with our expectation, the factors salient to those who intended to adopt (or those who already adopted) and those who did not intend to adopt were very different. Different dimensions of hedonic outcomes and social outcomes emerged as key drivers of early adopters&rsquo; first PC purchase. In contrast to early adopters, later adopters were driven primarily by utilitarian outcomes. Third, it was apparent that non-adopters were constrained by barriers, with obsolescence of technology emerging as a critical hurdle that had not been theoretically anticipated. This overall pattern of results was also supported by the second wave of data collection six months later. Finally, Phase 2 (six months after the initial survey) revealed that nearly all non-intenders and almost 90% of those who were uncertain (at the time of the initial survey) did not adopt a PC in the six-month time frame of this research, but less than half of all intenders followed up with their intent to adopt a PC, indicating an asymmetrical relationship between intent and behavior among intenders, non-intenders, and those who are uncertain.

Theoretical Contributions and Implications

This research presents an important step in examining the generalizability of models and theories of technology acceptance and usage behavior applied to the workplace [such as the Technology Acceptance Model (Davis, et al. 1989) and Innovation Diffusion Theory (e.g., Agarwal and Prasad 1998)] to home PC adoption. Specifically, by identifying and developing the role of hedonism, social outcomes and fear of obsolescence in home PC adoption, we depart from a more utilitarian perspective that has characterized the results from the significant body of prior technology acceptance research in workplace settings (e.g., Davis, et al. 1989). The current research also suggests that social influences play a key role in home adoption of PCs. Overall, this pattern of results calls for additional research in three key directions. Future work should examine the generalizability of these findings using a more quantitative approach such as a survey using validated items measuring the various dimensions/constructs. Also, the present work sets up the need for future research to investigate the possible role of hedonic outcomes and social outcomes in organizational contexts. Further work is also necessary to explore the role of hedonism to enhance user acceptance in the workplace. Constructs such as playfulness (Webster and Martocchio 1992, 1993) and enjoyment (Davis, et al. 1992) present starting points for work that aims to leverage hedonism in the workplace to create more favorable user perceptions about technology.

Prior research on innovation diffusion (Rogers 1995) and other technology acceptance models (e.g., Davis, et al. 1989) has tended to view adoption as a continuum. Specifically, the transition from one category of adopters (per Rogers 1995) to the next is assumed to be smooth with each of the earlier categories serving as a reference group for the next. The results, that the psychographics of the different adopter categories are different, challenge this assumption and support the viewpoint of Moore (1991) that there are indeed gaps/cracks in the technology adoption life cycle (bell curve). Specifically, hedonic outcomes and social outcomes were critical to innovators and early adopters suggesting they are enthusiasts/visionaries, utilitarian outcomes were important to the early majority suggesting they are pragmatists, and a fear of obsolescence constrained later adopters and laggards indicating they are conservatives/skeptics (see also Moore 1991).

These findings present very important implications for the future marketing of PCs and other high-tech products (innovations). Efforts aimed to sell any innovation should focus on the factors most salient to the target adoption category, rather than relying simply on characteristics deemed important by innovators and early adopters. This calls for careful attention to the extent of diffusion of the particular innovation (in terms of the technology adoption life cycle). While the present work explained psychographic differences in Rogers&rsquo; (1995) adoption categories, future work is necessary to understand more thoroughly the cracks/gaps in the bell curve (between each of the different categories proposed by Rogers 1995). One approach to investigate this question would be to track an innovation from its infancy through the entire technology adoption life cycle. Such an approach will address one potential limitation of the current work that there may be some retrospective justification in the responses (about factors influencing adoption) among those who had already adopted at the time of the initial survey. However, the possible existence of such biases, which would dictate that individuals tend to justify their original decision based on the current behavior, is alleviated to a certain extent in the current research given that the factors reported to have influenced the original purchase decision were different from factors driving current use. Additional future work is also necessary to examine the generalizability of these ideas to the contexts of individual adoption of technology in the workplace based on personal innovativeness (Agarwal and Prasad 1998), and organizational adoption of technology based on firm-level innovativeness (e.g., Mohr 1969; see also Rogers 1995 for a discussion).

Prior technology adoption research has typically seen the presence of certain factors (e.g., perceived usefulness) as leading to adoption, while a lack of those factors is seen as the cause of rejection. Building on the psychographic differences across adopter categories, this research also suggests non-adoption (or rejection decisions) is based on critical barriers, with factors relevant to adopters being of little or no significance in the decision to reject. One avenue for further exploration in the context of home PC adoption is understanding whether removal of the barriers to adoption makes the reasons for adoption (as identified by current users) more salient to non-adopters in their decision making process. Though the results are situated in the context of PC adoption in homes, the strong pattern of differences between adopters and non-adopters invite the examination of such boundary conditions (using models/theories such as Innovation Diffusion Theory, Technology Acceptance Model, etc.) in the workplace. A similar pattern of results in the workplace would suggest the need to refine current models explaining acceptance and use of information technology in the workplace. In any case, the present research emphasizes the need to carefully consider factors that are salient to non-adopters, especially if the ultimate goal is to increase technology adoption.

A distressing pattern emerged in examining the intention-behavior relationship over time. While about 99% of households indicating that they would not purchase a PC conformed with their original intent, only 43% of households indicating an initial intent to purchase did follow up on their intent. This suggests that intenders and non-intenders are not only different from the standpoint of the decision-making process but also from the perspective of actual follow-up behavior, with intenders being less likely to follow up with their purchase intent. The key implication of this asymmetrical relationship between intent and behavior from the perspective of marketing theory and practice is that it is important to follow up with customers who are intenders since they may still need additional persuasion to actually convert their intention to behavior. Also, importantly, this highlights the need to design interventions, possibly via promotions, that help non-intenders scale key barriers, or at the very least minimize the perception that the barriers exist.

While one of the key factors emerging as a barrier to adoption was the fear of obsolescence, this is a finding subject to volatility due to the declining cost of PCs. As with most consumer purchase decisions, particularly with an expensive product like a PC, cost may play a critical role, possibly by changing the dynamics of the role and importance of other determinants of adoption and non-adoption. For instance, high cost coupled with the rapid change in technology results in a cost-to-useful life ratio that is possibly unacceptable to many consumers. However, if the cost were lower and the perceived useful life higher, cost and/or obsolescence may not be significant factors. Under these conditions, other factors might emerge as significant. Thus, not only does cost represent an important issue for future research to address, but it also represents a key aspect for researchers and practitioners to track and understand, potentially to establish non-linear and/or interaction effects on purchase intent and behavior (see Sahni 1994).

To the household consumer, adopting a PC is a decision about hardware and software. Thus, the results of this research very likely pertain to the combination rather than hardware or software per se. Further research is necessary to examine differences in factors influencing adoption and non-adoption of software in homes, especially given the significant cost differences between hardware and software and the potential for sunk costs on hardware to influence software purchases. The issues related to constant upgrading and potential obsolescence can be expected to be significantly more complex in the case of software adoption. In the case of hardware, upgrading may have key implications only from the perspective of cost and compatibility. Software upgrades, however, may create the need to invest time and effort to learn a new package. Also, from the perspective of daily usage, constant upgrades create situations wherein the end user has to expend significant cognitive energy and effort just to use the software. From a pragmatic standpoint, especially for early adopters, it may also create issues of system incompatibility with a majority of users.

Challenging Conventional Practice in the PC Industry

The findings of this research bear important implications for practitioners. Clearly, adoption and non-adoption of PCs do not represent different ends of the same continuum. Factors influencing those who have adopted PCs to this point can be expected to differ systematically from those who will adopt later, consistent with Rogers (1995). Specifically, hedonic outcomes and social outcomes seem to be more salient to early adopters, compared to more utilitarian reasons and social influences for later adopters. Further, it appears that adoption among later adopters is being inhibited in the current environment. This implies that advertising campaigns designed to capture the business of the latter two-thirds of households will have to be tailored to overcoming factors that have been identified as barriers to adoption. Particularly, the importance being placed by household decision makers on potential obsolescence is critical. This suggests that the continuing trend of rapid innovation that is pervasive in the PC industry is likely to leave households more skeptical and less likely to adopt PCs. The later adopters are less affluent and thus, despite the declining costs, the issue of obsolescence can be expected to continue to be salient to them. This treadmill of upgrading may also have a negative effect from the perspective of learning a new system ever so often.

Cumulatively, these issues may provide a potential explanation for the recent disappointing sales performances of some of the leaders in the PC industry. Cost, coupled with rather rapid obsolescence is essentially keeping a large segment of the population out of the PC market, and consequently, out of the national information loop. One potential solution is to retard the pace of technological innovation, thus challenging the wisdom of the PC industry that "newer is better." It is, however, difficult to argue with newer, better, and cheaper. Thus, while retarding technological advances is an untenable solution, a feasible alternative may be to focus on backward compatibility, thus potentially allaying some fears of obsolescence. For instance, to some consumers the introduction of Windows 98 will be a call to upgrade their systems -- for others, particularly the late majority, it will be just another reason to avoid adopting a PC. Although an upgrade to Windows 98 is not necessary for most people, and likely to result only in a minimal improvement in performance, its existence potentially exacerbates the perception of rapid obsolescence to many consumers.

Very recently, businesses seem to have begun to pay attention to issue of obsolescence. Though not by slowing the pace of technological change, or highlighting backward compatibility, but by attempting to provide "insurance" against obsolescence. For example, Gateway&rsquo;s program termed "Your:)WareSM" ("Gateway" 1998) allows consumers to trade in their Gateway computer for its blue book value any time between the second and fourth years of ownership.Of course, it is not without a catch. Consumers must finance their purchase through Gateway, obtain internet access through Gateway, or purchase a special software bundle in addition to the computer. It remains to be seen if the additional expense will be an acceptable mechanism to ward off the dreaded technology obsolescence. In fact, the actual amount of security that this program provides is likely to be minimal if one were to consider the rapidity with which PC prices have declined in the last few years. The conservative/skeptical late majority and laggards, presumably the target audience for such a program, may quite possibly see through this plan and recognize that its actual value is likely to be quite low, rendering it more of a perceived insurance than a tangible one. In any event, it appears as though fear of obtaining technology that will become obsolete quickly is contributing to the non-adoption of PCs in a majority of American households, and thus keeping them from navigating the information superhighway. If perceived insurance against obsolescence helps to bring along the skeptics, then more companies will pursue it. If, however, we are correct in assuming that savvy consumers will see through the plan, then the challenge facing the PC industry is reducing actual obsolescence.

CONCLUSIONS

In sum, this research on PC adoption in homes presented several key insights for theory and practice. The findings suggested that there were important psychographic differences among the various adopter categories, especially between adopters and non-adopters. While adopters were influenced by different outcomes, utilitarian, hedonic and social, non-adopters were influenced strongly by the fear of obsolescence. The purchase behavior of non-intenders was very much in line with their original decision, whereas less than half of all intenders followed up on their original decision. Based on these findings from the two-wave, nationwide phone survey, we presented important implications for adoption of technologies in homes and workplaces. Finally, key challenges facing organizational practice in the PC industry are outlined.
 

Acknowledgements: We would like to extend our sincere appreciation to Professors Fred Davis, Cheri Speier, and Dan Smith for providing insightful comments on earlier drafts of this paper. We would like to express our deepest sense of gratitude to the coders, Joanne Rypien and Jason Henson. We would also like to thank Tracy Ann Sykes for her editorial comments.
 
 

REFERENCES

Adams, D. A., R. R. Nelson, and P. A. Todd, "Perceived Usefulness, Ease of Use, and Usage of Information Technology: A Replication," MIS Quarterly, 16 (1992), 227-47.

Agarwal, R. and J. Prasad, "A Conceptual and Operational Definition of Personal Innovativeness in the Domain of Information Technology," Information Systems Research, 9 (1998), 204-15.

_______ and _______, "The Role of Innovation Characteristics and Perceived Voluntariness in the Acceptance of Information Technologies," Decision Sciences, 28 (1997), 557-82.

Ajzen, I., "From Intentions to Actions: A Theory of Planned Behavior," in Action Control: From Cognition to Behavior, Springer Verlag, New York, 1985, 11-39.

_______, Attitudes, Personality, and Behavior, Dorsey Press, Chicago, 1988.

_______, "The Theory of Planned Behavior," Organizational Behavior and Human Decision Processes, 50 (1991), 179-211.

_______ and M. Fishbein, Understanding Attitudes and Predicting Social Behavior, Prentice-Hall, Englewood Cliffs, New Jersey, 1980.

Assael, H., Consumer Behavior and Marketing Action, PWS-Kent Publishing Company, Boston, MA, 1992.

Babbie, E., Survey Research Methods, 2nd ed. Wadsworth Publishing Company, Belmont, California, 1990.

Babin, B. J., W. R. Darden, and M. Griffin, "Work and/or Fun: Measuring Hedonic and Utilitarian Shopping Value," Journal of Consumer Research, 20 (1994), 644-56.

Bearden, W. O. and M. J. Etzel, "Reference Group Influence on Product and Brand Purchase Decisions," Journal of Consumer Research, 9 (1982), 183-94.

Biscomb, V. J., "Detailed Planning Makes Telepolling Effective," Marketing News, (1986), 5.

Brancheau, J. C. and J. C. Wetherbe, "The Adoption of Spreadsheet Software: Testing Innovation Diffusion Theory in the Context of End-User Computing," Information Systems Research, 1 (1990), 115-43.

Brock, J. J., "Phone Interviews Equal in-Person Methodology," Marketing News, (1986), 29.

Burnkrant, R. E. and A. Cousineau, "Informational and Normative Social Influence in Buyer Behavior," Journal of Consumer Research, 2 (1975), 206-15.

Canary, D. J. and D. R. Seibold, Attitudes and Behavior: An Annotated Bibliography, Praeger, New York, 1984.

Childers, T. L. and A. R. Rao, "The Influence of Familial and Peer-Based Reference Groups on Consumer Decisions," Journal of Consumer Research, 19 (1992), 198-211.

Chin, W. W. and P. A. Todd, "On the Use, Usefulness, and Ease of Use of Structural Equation Modeling in MIS Research: A Note of Caution," MIS Quarterly, 19 (1995), 237-46.

Cooper, R. B. and R. W. Zmud, "Information Technology Implementation Research: A Technological Diffusion Approach," Management Science, 36 (1990), 123-39.

Davis, F. D., "Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology," MIS Quarterly, 13 (1989), 319-39.

_______, "User Acceptance of Information Technology: System Characteristics, User Perceptions and Behavioral Impacts," International Journal of Man-Machine Studies, 38 (1993), 475-87.

_______, R. P. Bagozzi, and P. R. Warshaw, "User Acceptance of Computer Technology: A Comparison of Two Theoretical Models," Management Science, 35 (1989), 982-1003.

_______, _______, and _______, "Extrinsic and Intrinsic Motivation to Use Computers in the Workplace," Journal of Applied Social Psychology, 22, 1111-1132.

_______ and V. Venkatesh, "A Critical Assessment of Potential Measurement Biases in the Technology Acceptance Model: Three Experiments," International Journal of Human-Computer Studies, 45 (1996), 19-45.

_______ and P. R. Warshaw, "What Do Intention Scales Measure?," The Journal of General Psychology, 119 (1992), 391-407.

Day, J. C., "Projections of the Number of Households and Families in the United States: 1995 to 2010," C. P. R. U.S. Bureau of the Census, Government Printing Office, Washington, D.C., 1996, P25-1129.

Deci, E. L., and R. M. Ryan, "The Empirical Exploration of Intrinsic Motivational Processes," in Advances in Experimental Social Psychology, Ed. L. Berkowitz, Academic Press, New York, 1980, 39-80.

Dickerson, M. D. and J. W. Gentry, "Characteristics of Adopters and Non-Adopters of Home Computers," Journal of Consumer Research, 10 (1983), 225-35.

Dillman, D. A., Mail and Telephone Surveys: The Total Design Method, John Wiley, New York, 1978.

Erickson, G. M. and J. K. Johansson, "The Role of Price in Multi-Attribute Product Evaluations," Journal of Consumer Research, 12 (1985), 195-99.

Feldman, L. P. and G. M. Armstrong, "Identifying Buyers of a Major Automotive Innovation," Journal of Marketing, 39 (1975), 47-53.

Fishbein, M. and I. Ajzen, Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Addision-Wesley, Reading, MA, 1975.

Fisher, R. J. and L. L. Price, "An Investigation into the Social Context of Early Adoption Behavior," Journal of Consumer Research, 19 (1992), 477-86.

Foxall, G. R., "The Behavioral Perspective Model of Purchase and Consumption: From Consumer Theory to Marketing Practice," Journal of the Academy of Marketing Science, 20 (1992), 189-98.

Frey, J. H., Survey Research by Telephone, 2nd ed. Sage Publications, Newbury Park, CA, 1989.

Fulk, J. and B. Boyd, "Emerging Theories of Communication in Organizations," Journal of Management, 17 (1991), 407-46.

_______, C. W. Steinfield, J. Schmitz, and J. G. Power, "A Social Information Processing Model of Media Use in Organizations," Communication Research, 14 (1987), 529-52.

"Gateway Introduces Your:)WareSM, a Revolutionary Way to Buy a PC," Web page, May 1998, Available at http://www.gateway2000.com/home/aboutgw/pressreleas/pr052798/.
Home: About Gateway: About Gateway: Press Releases: Press Releases: pr052798.

Gefen, D. and D. W. Straub, "Gender Differences in the Perception and Use of E-Mail: An Extension to the Technology Acceptance Model," MIS Quarterly, 21 (1997), 389-400.

Hirschman, E. C. and M. B. Holbrook, "Hedonic Consumption: Emerging Concepts, Methods and Propositions," Journal of Marketing, 46 (1982), 92-101.

Holak, S. J., "Determinants of Innovative Durables Adoption an Empirical Study With Implications for Early Product Screening," Journal of Product Innovation Management, 5 (1988), 50-69.

Holbrook, M. B., R. W. Chestnut, T. A. Olivia, and E. A. Greenleaf, "Play As Consumption Experience: The Roles of Emotions, Performance, and Personality in the Enjoyment of Games," Journal of Consumer Research, 11 (1984), 728-39.

_______ and E. C. Hirschman, "The Experiential Aspects of Consumption: Consumer Fantasies, Feelings, and Fun," Journal of Consumer Research, 9 (1982), 132-40.

Igbaria, M., N. Zinatelli, P. Cragg, and A. L. M. Cavaye, "Personal Computing Acceptance Factors in Small Firms: A Structural Equation Model," MIS Quarterly, 21 (1997), 279-305.

Kohut, A., Technology in the American Household, Times-Mirror Center for People and the Press, Washington, DC, 1995.

Kraut, R., "The Internet@Home," Communications of the ACM, 39 (1996), 32-35.

_______, W. Scherlis, T. Mukhopadhyay, J. Manning, and S. Kiesler, "The Home Net Field Trial of Residential Internet Services," Communications of the ACM, 39 (1996), 55-63.

Labay, D. G. and T. C. Kinnear, "Exploring the Consumer Decision Process in the Adoption of Solar Energy Systems," Journal of Consumer Research, 8 (1981), 271-78.

Lacher, K. T. and R. Mizerski, "An Exploratory Study of the Responses and Relationships Involved in the Evaluation of, and in the Intention to Purchase New Rock Music," Journal of Consumer Research, 21 (1994), 366-80.

Lavrakas, P. J., Telephone Survey Methods Sampling, Selection, and Supervision, 2nd ed. Sage Publications, Newbury Park, CA, 1993.

Malone, T. W., "What Makes Computer Games Fun?" Byte, (1981), 258-76.

Mathieson, K., "Predicting User Intentions: Comparing the Technology Acceptance Model With the Theory of Planned Behavior," Information Systems Research, 2 (1991), 173-91.

McCracken, G., Culture and Consumption: New Approaches to the Symbolic Character of Consumer Goods and Activities, Indiana University Press, Bloomington, IN , 1988.

Miles, M. B. and A. M. Huberman, Qualitative Data Analysis: An Expanded Sourcebook, Sage Publications, Inc., Thousand Oaks, California, 1994.

Miniard, P. W. and J. B. Cohen , "Isolating Attitudinal and Normative Influences in Behavioral Intentions Models," Journal of Marketing Research, 16 (1979), 102-10.

Mohr, L. B., "Determinants of Innovation in Organizations," American Political Science Review, 63 (1969), 111-26.

Moore, G. C. and I. Benbasat, "Development of an Instrument to Measure the Perceptions of Adopting an Information Technology Innovation," Information Systems Research, 2 (1991), 192-222.

Moore, G. A., Crossing the Chasm: Marketing and Selling Technology Products to Mainstream Customers, HarperCollins Publishers, New York, 1991.

Morton, R. and D. Duncan, "A New Look at Telephone Interviewing Methodology," Pacific Sociological Review, 21 (1978), 259-73.

Morwitz, V. G. and D. Schmittlein, "Using Segmentation to Improve Sales Forecasts Based on Purchase Intent: Which 'Intenders' Actually Buy?" Journal of Marketing Research, 29 (1992), 391-405.

Rice, R. E., A. E. Grant, J. Schmitz, and J. Torobin, "Individual and Network Influences on the Adoption and Perceived Outcomes of Electronic Messaging," Social Networks, 12 (1990), 27-55.

Robinson, J. P. and G. Godbey, Time for Life: The Surprising Ways Americans Use Their Time, The Pennsylvania State University Press, University Park, PA, 1997.

Rogers, E. M. Diffusion of Innovations, 4th ed., The Free Press, New York, 1995.

_______ and F. F. Shoemaker, Communication of Innovations, Free Press, New York, 1971.

Rogers, T., "Interviews by Telephone and in Person: Quality of Responses and Field Responses and Field Performance," Public Opinion Quarterly, 40 (1976), 51-65.

Sahni, A., "Incorporating Perceptions of Financial Control in Purchase Prediction: An Empirical Examination of the Theory of Planned Behavior," Advances in Consumer Research, 21 (1994), 442-48.

Salancik, G. R. and J. Pfeffer, "A Social Information Processing Approach to Job Attitudes and Task Design," Administrative Science Quarterly, 23 (1978), 224-53.

Schifter, D. B. and I. Ajzen, "Intention, Perceived Control, and Weight Loss: An Application of the Theory of Planned Behavior," Journal of Personality and Social Psychology, 49 (1985), 843-51.

Segars, A. H. and V. Grover, "Re-Examining Perceived Ease of Use and Usefulness: A Confirmatory Factor Analysis," MIS Quarterly, 18 (1993), 517-252.

Sheppard, B. H., J. Hartwick, and P. R. Warshaw, "The Theory of Reasoned Action: A Meta-Analysis of Past Research With Recommendations for Modifications and Future Research," Journal of Consumer Research, 15 (1988), 325-43.

Steeh, C., "Trends in Nonresponse Rates 1952-1979," Public Opinion Quarterly, 45 (1981), 40-57.

Stone, E. F., Research Methods in Organizational Behavior, Goodyear Publishing Company, Santa Monica, California, 1978.

Straub, D., M. Limayem, and E. Karahanna-Evaristo, "Measuring System Usage: Implications for IS Theory Testing," Management Science, 41 (1995), 1328-42.

Subramanian, G. H., "A Replication of Perceived Usefulness and Perceived Ease of Use Measurement ," Decision Sciences, 25 (1994), 863-74.

Szajna, B., "Software Evaluation and Choice: Predictive Validation of the Technology Acceptance Instrument," MIS Quarterly, 18 (1994), 319-24.

_______, "Empirical Evaluation of the Revised Technology Acceptance Model," Management Science, 42 (1996), 85-92.

Taylor, S. and P. A. Todd, "Understanding Information Technology Usage: A Test of Competing Models," Information Systems Research, 6 (1995), 144-76.

Thompson, R. L., C. A. Higgins, and J. M. Howell, "Personal Computing: Toward a Conceptual Model of Utilization," MIS Quarterly, 15 (1991), 124-43.

Tornatzky, L. G. and K. J. Klein, "Innovation Characteristics and Innovation Adoption-Implementation: A Meta-Analysis of Findings," IEEE Transactions on Engineering Management , EM-29 (1982), 28-45.

Trevino, L. K. and J. Webster, "Flow in Computer-Mediated Communication: Electronic Mail and Voice Mail Evaluation and Impacts," Communication Research, 19 (1992), 539-73.

Triandis, H. C., Interpersonal Behavior, Brooks/Cole, Monterey, CA, 1977.

Tyebjee , T. T., "Telephone Survey Methods: The State of the Art," Journal of Marketing, 43 (1979), 68-78.

Vallerand, R. J., "Toward a Hierarchical Model of Intrinsic and Extrinsic Motivation," Advances in Experimental Social Psychology, 29 (1997), 271-360.

Venkatesh, A., "Computers and Other Interactive Technologies for the Home," Communications of the ACM, 39 (1996), 47-54.

_______ and N. P. Vitalari, "An Emerging Distributed Work Arrangement: An Investigation of Computer-Based Supplemental Work at Home," Management Science, 38 (1992), 1687-706.

Venkatesh, V. and F. D. Davis, "A Model of the Antecedents of Perceived Ease of Use: Development and Test," Decision Sciences, 27 (1996), 451-81.

Vitalari, N. P., A. Venkatesh, and K. Gronhaug, "Computing in the Home: Shifts in the Time Allocation Patterns of Households," Communications of the ACM, 28 (1985), 512-22.

Webster, J. and J. J. Martocchio, "Microcomputer Playfulness: Development of a Measure With Workplace Implications," MIS Quarterly, 16 (1992), 201-26.

_______ and_______, "Turning Work into Play: Implications for Microcomputer Software Training," Journal of Management, 19 (1993), 127-46.