1

The effect of Consumer’s Socio-Economic stratum on Complexity Expectations of

New Technological Products

Andres Barrios†

Sonia M. Camacho†

Carlos A. Trujillo†

Jose Antonio Rosa††

Bogotá, Colombia

Working paper. A version of this paper is under review for the Journal of Business Research.

This manuscript is being shared through the University of Wyoming Sustainable Business

Practices as a courtesy. Please do not cite without permission from the authors.

† Universidad de los Andes – School of Management

†† University of Wyoming

Send correspondence to Andres Barrios (abf@adm.uniandes.edu.co)

2

Abstract

The adoption of new technological products is a salient issue in current marketing

activities. As low-income consumers from emerging economies are increasingly targeted

by technology, it becomes necessary to understand the effect of socioeconomic stratum on

the way technology is adopted. We contribute to this topic by proposing and testing a

scheme of the way socioeconomic stratum influence product complexity expectations, a

construct that has been found to be an important determinant of technology adoption (e.g.,

Wood & Moreau, 2006). Using regression analysis and structural equation modeling we

show that socioeconomic stratum influence complexity expectations through consumer’s

knowledge, Self-esteem and subjective beliefs of capabilities.

3

1. Introduction

As technology (cell phones, internet) is being quickly spread to subsistance markets, it is

of salient importance for marketers to understand the particularities of the psychological

mechanisms that underly technology adoption in these segments. One way to apprach the

problem is to explore the way in which socio economic level affects known patterns of

behavior. Knowing these effects allows marketing strategies to go beyond the obvious

income barriers and to incorporate insights about the behavior of subsistence consumers.

Such knowledge truly adds to the long term success of technology adoption with

consideration for the consumer well being. Under this framework, this research mainly

contributes to the understanding of the cognitive barriers that subsistence consumers face

before being ready to adopt new technologies. In addition, we contribute, albeit to a lesser

extent, to knowledge in the adoption proccess of technology, by outlining the particularities

and contingencies that technology companies may face when targeting subsistence

segments. (see Venkatesh, Davis and Morris, 2007, for a recent review of the new

directions of research on technology adoption).

Our work focuses on getting insights into the mechanisms that underly the effect of socio

economic level on the cognitive structures that precede technology adoption. Specifically,

we focus on a relevant construct known as expectations of complexity, which has been

found to be a determinant precedent of adoption (Rogers, 1996). It is rather obvious to

expect that subsistance consumers face more barriers than other segments when adopting

and using new technologies. However, there is little knowldege regarding the detailed

4

mechanisms, psychological and sociological, that shape the negative effect of a low socio

economic context. We argue that we need that knowledge in order to help businesses in the

development of correct strategies that are directed to the real causes of subsistence

consumers barriers to new markets. With this goal in mind, we aim at setting a general

framework that provides an overall map of the main relationships that intervene in the

effect of socio economic level on complexity expectations of new technology.

The paper is organized in the following way. We first conducted a literature review aimed

at identifying psychological variables that are both influenced by socio economic level and

determinants of complexity expectations. From this exploration we extracted some

relationships and then we went further and developed four hypotheses that were necessary

in order to complete a general conceptual scheme of the effect of socio economic stratum

on complexity expectations, specifying the connections among the mediating variables that

operationalize such effect. Once the conceptual model was established, we performed a

quasi experimental procedure to test it. We analyze the results in two steps. First, we sought

empirical evidence of the pairwise relationships. Then, based on those results and the

theoretical arguments, we rebuilt the conceptual map of relationships and tested it through

structural equation modeling. Results show that our scheme is an adequate representation of

the general effect of socio economic level on expectations of complexity, providing insights

on the mechanisms that produce the effect. Finally, implications are analyzed for

marketing, cultural perspectives and policy making.

5

2. Complexity expectations and socio economic stratum

Different approaches have been taken to explain how new technologies are adopted. One of

the most widely used framework focuses on the adopters’ characteristics, according to the

chronological adoption of a new product. This way, a consumer might be classified as

early, average or late adopter (Dickerson and Gentry, 1983; Rogers, 1962). This widespread

approach is very useful for marketing purposes to the extent the researcher is able to find

particular market segments that should be targeted at different moments throughout the

product development. This view, however, says nothing about the internal process a given

consumer performs in order to adopt a new product, regardless of the relative moment of

adoption in comparison with other consumers. There is research on technology adoption

that focuses on the adoption processes, looking in particular for behavioral or judgmental

variables that are related to it (e.g., Bruner and Kumar; 2005; Davis et al 1989). Under this

framework, it has been found that the early experience with a product affects definite

adoption (Shih and Venkatesh, 2004). As part of the early experience with a product,

complexity expectations have been found to be a determinant of the adoption and diffusion

of new products and technologies (e.g. Rogers, 1996; Thompson, Hamilton and Rust, 2005;

Wood and Moreau, 2006). Complexity expectations are consumers’ ex-ante predictions

about ease of use, learning time, and learning difficulty of new products (Wood and

Moreau, 2006). The positive or negative confirmation of these expectations (i.e., the

technology is less or more complex than expected) provides a point of reference that

influences the evaluation of the innovation during its early use, which in turn exerts an

effect on final adoption.

6

Even though the effect of complexity expectations on the adoption process has been the

object of different studies (e.g., Shih and Venkatesh, 2004, Wood and Moreau 2006), little

research has examined the variables that shape such expectations (Moreau, Lehmann and

Markman 2001). For instance, there are no studies that address how complexity

expectations differ among consumers depending on their market and personal context. In

this work, we contribute to the understanding of complexity expectations by linking it with

a contextual variable such as socioeconomic stratum of consumers.

It has been established that consumer’s market context serves as frame for the development

of her cognitions, emotions and intentions (Hill and Gaines, 2007), revealing a connection

between her sense of self and the ability to consume (McCraken, 1986; Zukin and Maguire,

2004). Within a general notion of market context, one of the variables that shape individual

differences is socioeconomic stratum (Boardman and Robert, 2000; Wheatley, Chiu and

Stevens, 1980). Consumer’s socioeconomic stratum contribute to shape beliefs, attitudes

and behaviors (Lewis, 1959; Hill and Stephens, 1997; Jones and Lou, 1999) and it has been

widely accepted that there is a relation between consumption behavior and consumer’s

social class (Coleman, 1983; Levy 1966; Schaninger, 1981). Considering this frame, our

research aims to determine whether a contextual variable like socioeconomic status

influences complexity expectations. More specifically, we hypothesize that the

socioeconomic conditions that surround the consumer, determine certain judgments and

types of knowledge, which in turn produce a cognitive response that shapes the expected

complexity of new technological products. In order to operationalize this notion, we

7

reviewed the literature using a two step strategy: We first looked for some potential

determinants of consumer’s complexity expectations, and then we determined whether

consumer’s socioeconomic status may exert an effect on those determinants. In the

following sections, we provide details on what the literature contains regarding these

relationships, and then we go further proposing a number of hypotheses that complete the

scheme.

3. Determinants of complexity expectations:

The notion of complexity expectations can be generally understood as a construct whose

constituents are related to subjective beliefs and cognitions regarding the relationship of a

person with an object. In this sense, we selected from the literature a set of variables that

capture such beliefs and cognitions and therefore may constitute determinants of the

complexity expectation construct. These are: knowledge of domains (Moreau, Lehmann

and Markman, 2001), previous categorization (Moreau, Markman and Lehmann, 2001)

subjective beliefs of capabilities (Beloff, 1992), and Self-esteem (Twenge and Campbell,

2002).

Complexity expectations and knowledge: The cognitive side of expectations is related with

consumer´s knowledge. Bettman and Park (1980) stressed the role of previous knowledge

in the formation of expectations, pointing out that consumers place new information in a

context based on their past and present events. Such mapping process contributes to form

expectations about a new product. Other researchers emphasize such view stating that

8

complexity expectations are constructed using the consumer’s relevant prior knowledge

(Oliver and Winter, 1987; Wood and Moreau, 2006). Note that this type of knowledge is a

collection of previous experiences without a specified classification structure.

The role of knowledge can also be understood according to some kind of classification.

Previous research has established that consumer’s prior knowledge about related product

categories may be classified according to the domain to which the information belongs: it

could be related whether to the primary domain, when the basic category is directly

associated to the new product or to the supplementary domain, when a superordinate

category is recalled. This process facilitates product learning (Gregan-Paxton and John,

1997), by allowing the consumer to transfer extant knowledge from familiar to novel

domains (Yamauchi and Markman, 2000). Furthermore, it has also been argued that

marketers may facilitate categorization–based transfer of knowledge by giving consumers a

plausible category label. The activation of such category would then influence behavior by

eliciting previously formed preferences that apply to the new product. By the same

mechanism, the primed category may also modify consumer’s expectations of product

features (Moreau, Markman and Lehmann, 2001).

In the present research we test the robustness of these findings by treating expectations of

complexity as one of the subjective judgments that is affected by consumer previous

knowledge of categories, either primary or supplementary. In this sense we hypothesize that

complexity expectations are reduced due to the improved accessibility to knowledge related

to either type of category.

9

Complexity expectations and Self-assessed capabilities: There is no explicit account in the

literature, to our knowledge, of a connection between the subjective beliefs of own

capabilities and the construct of complexity expectations. However, we suggest that such

relation exists and has a key role in the determination in the construction of the expected

complexity of a product. We base this hypothesis on the idea that the subjective concept of

own capacity generates a general sense of difficulty/easiness for any task a person faces.

Such relationship is coherent with the well established notion that perceived self efficacy is

a determinant regulator of cognitive processes (e.g., Bandura, 1989; Gist and Mitchell,

1992). This general judgment of difficulty/easiness can be conceptualized, in a very

specific expression, as the expectations of complexity. This means that the consumer uses

her sense of general efficacy to confront the product and the task ahead. As a result of this

match, a complexity expectation judgment is issued. This way, two consumers with

identical knowledge of a product should display differences in their expectations of

complexity due to different subjective capabilities.

Summarizing, we have put together from the literature a set of cognitive and behavioral

variables that may constitute significant determinants of expectations of complexity. These

are domains of knowledge, related categories and self assessed capabilities. These variables

are expected to be the ones that channel the effect of socio economic stratum on complexity

expectations. Thus, in the following section we explore the evidence regarding the

relationship of these variables with socio economic aspects.

10

4. Effects of socio economic stratum

In this section we discuss the effects of socio economic stratum on the different variables

and psychological constructs that we previously identified as determinants of complexity

expectations. However, we must note that socio economic stratum is the reflection of more

specific socio economic variables, such as income, education and place of residence among

others. In this research we do not discriminate the effects of these variables separately.

Instead, we measure them and then calculate an individual overall socio economic stratum.

The focus of our conceptualization and analysis is the effect of this aggregated variable,

which can be understood as the aggregate effect of the actual causal agents (i.e., Income,

education, etc.)

Socio economic stratum (SES) - Knowledge of domains and categories: The widespread

characterization of low-income consumers as uneducated and limitedly experienced in

important areas of social interactions have made them seem unprofitable and unattractive

for marketers. This causes a kind of exclusion from marketing activities (Alwitt and

Donley, 1996; Bauman, 1998; Boyce, 2000; Hamilton and Catterrall, 2005). Such isolation

affects the formation of knowledge about products. Thus, even though researchers have not

explicitly established a direct relationship between socioeconomic status and knowledge of

technology domains, some have found that prior knowledge about products is formed

through exposure to external sources of information (Gregan- Paxton and Roedder,

1997).This means that socio economic level becomes a proxy for low exposure to the types

of marketing promotions that serve to educate consumers in the capabilities and functions

11

of new technologies. Therefore, we suggest that marketing exclusion constitutes a learning

obstacle for low-income consumers. This naturally decreases their knowledge on both

technology domains and product categories.

Socio economic stratum (SES) - Self-assessed capabilities: The relationship between

socioeconomic stratum and subjective capabilities is treated from different perspectives in

the literature. For instance, Alwitt and Donley (1996) describe some societies where low

SES people are expected to underperform on intellectual tasks. Furthermore, some studies

have found that there exists a stereotypical view of poverty, in the sense that some believe

poverty is actually caused by lack of capabilities (Becker, 1997). Some authors (e.g.,

Croizet & Claire, 1998; Rist 1970) argue that such conceptions about low SES groups can

be strong enough to produce a stereotype that is somehow adopted by the poor. This leads

low SES groups to underperform in different tasks even in the absence of a discriminating

treatment (Crocker & Major, 1989; Steele, 1997). Complementary, such stereotype is

believed to influence the subjective estimates of intelligence (Beloff, 1992). This way, we

expect that SES and Self assessed capabilities are positively correlated.

Socio economic stratum (SES) - Self-esteem – Self assessed capabilities: As it was

explained above, it has been established that SES affects subjective capabilities through the

formation of certain negative beliefs about intelligence. However, there is an additional

component of that relationship that is more affectively laden due to the mediation of Self-

esteem. Consumer self esteem is an important determinant of behavior that has been found

to be related to socio economic stratum. It could be defined as “the overall affective

12

evaluation of one´s own worth, value or importance” (Blascovic and Tomaka, 1991).

Twenge and Campbell (2002) found a small but statistically significant positive

relationship between Self-esteem and SES, emphasizing that income, education and

occupation were the most salient aspects of SES that drive the effect on Self-esteem.

Subsequently, Self-esteem would affect Self assessed capabilities in the following way:

Self-esteem is an overall Self-evaluation (Boardman and Robert, 2000) composed by

different dimensions including an “academic” facet (Marsh, 1986, 1990). It has been

established that such overall evaluation of the Self is some kind of aggregation of the single

dimension assessments. However, if the evaluation of one of the dimensions is not firmly

established and supported by clear evidence, then the directionality of the evaluations may

be reversed. This means that the overall evaluation would be the driver of the weakly

developed dimension evaluation (Rosenberg, Schooler, Schoenbach and Rosenberg, 1995).

Given that low SES people have little evidence and experience regarding their capabilities,

as well as clear biases, such evaluation is weakly supported. We hypothesize that this is

indeed the case and therefore, that a consumer with a low Self-esteem will likely have low

Self-assessed capabilities.

5. Summary of relationships and hypotheses.

We have so far explained the relationships that are likely to play a role in the overall effect

of socioeconomic stratum on expectations of complexity. Initially, socio economic stratum

influences these variables, and through them expectations of complexity is then affected.

These variables are: Consumer self esteem, consumer knowledge of product domains,

13

consumer knowledge of related categories and Self assessed capabilities. Since not all these

relationships are well established in the literature we developed specific hypotheses to

complement the structural conceptualization.

H1: Information about product categories, primary or supplementary reduces complexity

expectations of the new product.

H2: There is an inverse relationship of consumer’s Self-assessed capabilities and

expectations of complexity.

H3: Consumer’s Self-esteem positively influences Self-assessed capabilities.

H4: Socio economic stratum positively influences knowledge of technology.

Figure 1 summarizes the complete scheme. The dotted lines denote our hypotheses and

solid lines denote the relationships extracted from the literature. Variables represented by

squares are observable and variables represented by ovals are non observable constructs.

14

Figure 1. Conceptual scheme of effects.

6. Test of the conceptual scheme

In order to test the scheme and to provide empirical support for it, we performed a quasi

experimental procedure. The basic idea was to expose a group of participants from different

social levels to a new technological product and ask them to perform some task with it. As

we shall explain, we did a pre test in order to select the most appropriate product. Then, at

different stages of the experiment, we collected the necessary data to evaluate all variables

in the model.

Choice of innovation stimulus: The product selection was based on two requirements: (1) it

should be a product unavailable in the local market, and therefore the participant should not

Socio

economic

stratum

Self assessed

capabilities

Expectations

of technology

Self esteem

Knowledge

of domains

15

have used it, nor even know specifically how to use it. (2) It should be a product that the

consumer may be able to classify in his primary or supplementary domain of knowledge.

We ran a pre test with 180 participants from different socio economic strata in Bogota,

Colombia. We basically showed participants some information about products and then

we asked them to say what they thought the product was and whether they knew it. In order

to test the second requirement we developed two treatments for each product. In one

treatment the participant was given a picture of the product and a description of some

secondary functions of it. In the second treatment we added to that information the main

function of the product. If the second requirement was fulfilled, participants in the first

treatment should describe the product using a supplementary (or superordinate) category,

whereas participants in the second treatment should describe the product using the primary

category. The product that produced the best results was an e-book. The primary category

mostly used by participants was indeed “digital book reader” (31%); and the most frequent

supplementary category was PDA (personal digital assistant) (46%). These results provided

us in addition, with the categories that we should use as the primary and supplementary

base domains in the subsequent main procedure.

Method

The procedure followed in the quasi experiment was performed one participant at a time, in

a closed office with the experimenters. It consisted of four phases, chronologically ordered

in the following way: First we identified respondent’s socioeconomic stratum (SES); then

we applied a series of instruments in order to measure the mediating variables, namely,

participants’ Self-esteem, Self-assessed capabilities and knowledge of technology domains.

16

Up to that point all participants went through the same procedure. In the third phase we

performed a category knowledge manipulation. Note that product categories were pre-

tested during the choice selection phase, and two (e-book reader and PDA) were used in the

main study to assess generalizabilty across new types of technology. Participants were told

that the product was either a PDA or an e-book. This should not affect complexity

expectations. Finally, respondents were asked to respond the expectations of complexity

questionnaire and afterwards to complete a task with the product. Before leaving,

participants were paid a flat fee of 6, 9 or 14 USD accordingly to the socio economic

stratum.

Measurement instruments.

Socio economic stratum (SES): Participants filled out a set of demographic questions in

order to establish their socioeconomic stratum. We used the assumptions and procedure of

Hollingshead and Redlich (1967) , according to which, socioeconomic status is determined

by: “(1) the existence of a class status structure in the community, (2) that class status

positions are determined mainly by a few commonly accepted symbolic characteristics and

(3) that characteristics symbolic of class status may be scaled and combined by the use of

statistical procedures so that a researcher can reliably and meaningfully stratify the

population” This way, following their procedure the demographic variables used to

determine SES were: residential stratum, monthly income, educational scale and parents’

educational scale.

17

Self-esteem: We used a standard method for assessing the individual Self esteem that

conveniently suited our research procedure. This was the Coopersmith Self-Esteem

Inventory (1967). It consists of a group of 25 questions that evaluates participants’ attitudes

towards themselves in four areas: (1) social self-peers, (2) home-parents, (3) school / work

(4) general-self.

Beliefs of Capabilities: To assess participant’s belief of capabilities, we adapted the

methodology used by Rammstedt and Rammsayer (2002), which asks subjects to assess

their score in the seven aspects of intelligence based on Thurstone’s primary mental

abilities (verbal comprehension, verbal fluency, arithmetic ability, memory, perceptual

speed, inductive reasoning and spatial visualization), and in the types of intelligence

proposed by Gardner (1983) (linguistic, logical-mathematical, spatial, musical, body-

kinesthetic, interpersonal and intrapersonal intelligence).Since this instrument was too long

for our purposes, the adaptation we conducted was the following: Some weeks before the

experiment we applied the test to 64 people different than those of the main experiment.

We performed a factor analysis to determine which of the 14 original items of the test

should be included. We selected surrogated items following the procedure used by Hair et

al (1998): We took the items with a factor loading of 0.7 or above in each of the factors of

the rotated solution and we used those items as the constituents of the reduced version of

the test. These are the items finally included in the adapted Self-assessed capabilities test:

verbal fluency, arithmetic ability, interpersonal intelligence, linguistic intelligence and

spatial intelligence. The test presented a short description of each item and asked subjects

to assess their subjective level in each item using a seven point scale.

18

Knowledge of technology domains: Participants’ knowledge was measured using a

questionnaire that included 6 different true/false statements. We used a separated

questionnaire for each category domain: digital book reader (primary) and PDA

(supplementary). Then, participants answered whether the statements were true or false or

whether they didn’t know the answer (an option used to avoid guessing) (Johar, Jedidi and

Jacobi, 1997; Moreau, Lehmann and Markman 2001). The sum of the correct answers

provided the measure of knowledge.

Complexity expectations: Following Wood and Moreau (2006) and Rogers (1996)

procedure, participants indicated on three seven point scales how difficult to use they

expected the product to be, how long it would take to learn to use it, and how much of a

challenge its use would be.

Sample

The sample consisted of 266 participants from Bogota, Colombia, representing most socio

economic strata of the city. We hired a trained recruiter, experienced with several

marketing research companies, to find the people in the different neighborhoods and bring

them to the university campus in small groups.

7. Results

Descriptive information: The sample had a mean age of 39 years old, with a SD of 16 years

ranging between 19 and 84 years. 50.8% were male and 49.2% women. Since SES was a

19

composite measure of four items, we obtained a continuous variable ranging from 1 to 4.

We obtained information for all SES levels. Table 1 displays descriptive statistics for the

main variables.

Table 1. Descriptive statistics of main variables.

N Min Max Mean SD

1. Socioeconomic Status (SES) 266 .67 4.00 2.11 .85

2. Self-steem 266 24.00 100.00 79.98 12.43

4. Knowledge of domains 266 .00 10.00 3.61 2.82

6. Expectations of complexity 266 1.00 7.00 3.91 1.26

3. Capabilities (belief) 266 1.20 6.60 5.10 1.07

We analyzed the information in two steps: In the first one we took the individual

relationships among the different variables in order to seek empirical support for each of

the relationships explained previously. However, this analysis is not sufficient to test the

whole conceptual model. The reason is that several variables (knowledge, Self esteem and

subjective capabilities are both dependent and independent. Therefore, step two of the

analysis builds on the results of step one by constructing a structural equation model of the

whole system of relationships in order to validate its behavior as a system. Such structural

model is hence supported by the theoretical argumentations and the empirical results of the

regression analyses.

Individual relationships: We first performed multiple linear regressions to assess the effects

of socio economic stratum on the different intermediate variables. Table 2 summarizes the

results.

20

Table 2. Multiple linear regressions of the effect of SES

Dependent variableSocioeconomic

StatusAge Self-esteem R

2

Self-esteem .699** -.08 .54

Capabilities (belief) .266** -.252** .491** .72

Knowledge of

domains .354** -.311** .32

** p < .01.

Standardized Betas

We found that SES exerts a highly positive significant effect on Self esteem, capabilities

beliefs and knowledge. That supports H4 (positive effect of SES on knowledge) and

confirms and replicates the results already documented in the literature. In addition, Self

esteem display a significant positive effect on capabilities belief, providing suport for H3.

Note that we also included in the analysis the effect of age, which we found to be a

significant determinant of subjective capabilities and knowledge. This means that younger

people know more about technology, which is not surprinsing, and also that younger people

tend to show high subjective capabilities. This may influence complexity expectations and

should be taken into account in the subsequent analyses. In summary, regressions provided

support for the expected effects of SES on knowledge, Self esteem and subjective

capabilities. In addition, the relationship between Self esteem and subjective capabilities

was also supported.

We moved forward to analyse the effect of knowledge and subjective capabilities on

expectations of complexity. This analysis was performed using two procedures. In the first

21

one we performed a multiple linear regression including knowldege of domains and

subjective capabilities as regressors of expectation of complexity. We also included age,

due to the effect found in the previous analyses. In the second analysis, we seperately tested

the effect of the category priming manipulation on complexity expectations using Anova.

Table 3 summarizes the results of the regression analysis.

Table 3. Linear regression of knowledge and subjective capablities on expectations of

complexity

Dependent variableCapabilites

(belief)

Knowledge of

domainsAge R

2

Expectations of

complexity -.55** -.25** -.03 .47

** p < .01.

Standardized Betas

The regression analysis revealed a significant negative effect of knowledge and subjective

capabilities on expectations of complexity. These results provide support for H2 and

replicate the effect of knowldege according to the literature. Age displayed no effect on

expectations of complexity. In addition, as we expected the category information

manipulation had no effect on expectations of complexiy (t= 0,5, p = 0,6).

Structural model: The analyses performed in the previous section provided initial empirical

support for most of the relationship described as the basis of our conceptualization of the

effect of socio economic stratum on expectations of complexity. However, the structure of

the model renders the regression analysis insufficient to draw insightful enough

22

conclusions. The main feature of our scheme is the dual role of some variables (knowledge,

self esteem and subjective capabilities), so that they work both as dependent and

independent (see figure 1). In addition, several of our variables are non observable

constructs. These characteristics lead to the use of structural equation modeling as a highly

appropriate technique (Hair, Black, Babin, Anderson and Taham, 2006) to evaluate and test

the complete model as a whole. The input of regression analyses then became a source of

empirical support for the conceptual scheme depicted in figure 1 that helps to trasform the

conceptual map into a testable path diagram. The final path diagram produced with Amos

software 16.0 is shown in figure 2.

23

Figure 2. Path diagram of the effect of Socio economic stratum on expectations of

complexity.

24

This path diagram produced a Goodness of Fit Index (GIF) of 0,87 with a comparative fit

index (CFI) of 0,93. The root mean square error of approximation (RMSEA) was 0.08 and

the Chi-square was 316, significant at the 0,01 level with 115 degrees of freedom. In

addition, the squared multiple correlation of expectations of complexity was 0.7, meaning

that the model captured 70% of this construct´s variance. Given the characteristics of the

model complexity (16 observed variables and 37 estimated parameters) and sample size

(266), the model shows an adequate fit, following Hair et al., (2006); Carmines and McIver

(1981) and Browne and Cudeck (1993) criteria for model adequacy assessement. Thus, we

can state the we succesfully produced a conceptual scheme that captures a significant extent

of the effect of SES on complexity expectations. Appendix 2 contains correlations of

measurement variables and appendix 2 contains information on regression coefficients of

both measurement and structural models. Finally, we must note that the model is

appropriate but not excellent. Gooodnes of fit is a little below of a desirable 0.9 and

RMSEA is right on the threshold. This is probably due to the fact that some variance of

expectations of complexity (30%) remains unexplained and the complexity of the

psychological constructs we used. Further research can address these questions using the

baseline we are providing in this work.

8. Discussion

We were able to define a conceptual model that adequately captures the main mechanism

by which expetactions of complexity are reduced as the socio economic level of people

increases. More specifically, we showed that there are three salient variables through which

25

this occurs, namely, knowldege of technology domains, self esteem and subjective beliefes

of own capacities.

Marketing implications: These results contribute to the understanding of the particular

barriers that subsistence consumers face when they start to participate in new markets, in

particular those that involve technological developments. Moreover, this research remarks

the fact that the barriers faced by subsistance consumers go beyond the purchasing power

issues, entering deeper psicological aspects. Our work emphasizes the notion that socio

economic level is a contextual variable that exerts significant effects on a wide range of

psychological constructs that are determinants of observable consumer behaviors.

Understanding these effects is useful for social and behavioral scientists and marketing

practitioners. In this sense, our results provide insights that can be applied to innovative

market segmentation strategies, as well as to develop better framed marketing

comunications of technological products, aimed at acting against the negative

psychological influence of a low socio economic context. For instance, there could be

market segments based on different degrees of consumer knowledge of technological

categories and based on that, three or four communication strategies could be developed,

varying their degree of educational content. Similar approaches may be taken for self

esteem based or capabilities based segmentation strategies. Based on our results, the bottom

line of any strategy targeted to subsistence consumers is that explicit tactics should be

implemented in order to reduce complexity expectations, and this may me accomplished by

incorporating knowledge, self esteem and subjective capabilities as the key strategic objects

of interventions.

26

Note that our research not only confirms the influence of knowledge of domains on the

technology adoption process (Gregan-Paxton and John, 1997; Moreau, Markman and

Lehmann, 2001; Gragan-Paxton, 1999; Yamauchi and Markman, 2000), but also provides

evidence on the relation between consumer knowledge of domains and economic status.

From the marketing perspective, finding this two way link makes salient the consequences

of what some researches (e.g, Alwitt and Donley, 1996; Also, Hamilton and Catterrall,

2005; Boyce, 2000) have called the exclusion of the low income from some marketing

strategies. Such exclusion creates psychological barriers that need time and properly

conceptualized actions to be overcame.

Effects of culture: This research complements studies of cultural differences as a barrier to

technology adoption. It has been suggested that cultural differences can act as a barrier to

technology adoption (Pohjola, 2003; Erumban and de Jong, 2006). At the same time,

different authors state that the low-income consumer has created a culture based on her lack

of resources (Hill, 2002, Hill and Gaines, 2007; Hamilton and Catterall, 2005). Our

research complements these arguments by providing an account of the way socio economic

stratum constitutes a contextual variable that shapes the cultural characteristics of people.

According to our results, this seems to structurally influence the behavior of individuals.

Furthermore, our results also complements knowledge on what Croizet and Claire (1998)

called the stereotype threat of the low-income consumer, adding details on the

characteristic of such stereotype regarding self esteem effects and subjective capabilities

effects on behavior.

27

Policy. Finally, we would like to comment on policy making implications of our research.

Technology diffusion is usually considered as a desirable goal in order to increase living

standards. Different multilateral organizations are working to reduce what they have called

the “Digital Divide” of the low-income population. There are different studies (e.g., Fife

and Pereira 2002; Erumban and de Jong, 2006) that explore macro economic factors that

determine the characteristics of countries’ adoption rates. They also suggest that socio

cultural factors are salient determinants of such adotion rates. Notwithstanding, the details

of how these socio economic factors operate are not clearly understood, let alone

incorporated in policies. We believe that behaviorally driven research can inform policy

makers about important issues that should be part of policy implementation in order to

maximaze its results. Our research leads us to argue that capabilities beliefs and knowledge

are part of the socio cultural factors that influences technology adoption. This provides two

specific directions towards which policies can be focused. In this sense, programs that

promote the construction of better self percepetion as well as inclusion in the difussion of

knowldge should help to increase aggregated technology adoption rates and help people to

be more functional in a technological society.

References

Alwitt. Donley. The Low-income Consumer: Adjusting the Balance of Exchange, London:

Sage Publications, 1996.

28

Bandura Albert. Regulation of cognitive processes through perceived self-efficacy.

Developmental Psychology 1989; 25(5): 729 – 735

Bauman. Work, Consumerism and the New Poor, Buckingham: Open University Press,

1998.

Becker Saul. Responding to Poverty: The Politics of Cash and Care. London: Longman,

1977 .

Beloff, Halla. Mother, Father and Me: Our IQ. The Psychologist 1992; 5: 309–311.

Bettman James. Park Whan. Effects of Prior Knowledge and Experience and Phase of the

Choice Process on Consumer Decision Processes: A protocol Analysis. Journal of

Consumer Research 1980; 7 (December): 234-248.

Blascovich. Tomaka. Measures of Self-esteem. In: Robinson,. Shaver. Wrightsman, editor.

Measures of personality and social psychological attitudes, NY: Academic, 1991.

Boardman Jason. Robert Stephanie. Neighborhood Socioeconomic Status and Perceptions

of self-efficacy. Sociological Perspectives 2000; 43 (1): 117-136.

Boyce Gordon. Valuing Customers and Loyalty: The Rhetoric of Consumer Focus versus

the Reality of Alienation and Exclusion of (Devaluated) Customers. Critical Perspectives

on Accounting 2000; 11 (6), 649-689.

29

Browne Michael. Cudeck Robert. Alternative ways of assessing model fit. In: Bollen

Kenneth and Long Scott, editors. Testing Structural Equation Models. Sage Focus Editions,

1993. pp 136 – 159.

Bruner Gordon. Kumar Anand. Explaining Consumers Acceptance of Handheld Devices.

Journal of Business Research 2005; 58: 553-558.

Carmines Edward. McIver John. Analyzing models with unobserved variables: analysis of

covariance structures. In: G. W. Bohrnstedt and E. F. Borgotta, editors. Social

Measurement: Current Issues. Beverly Hills, CA: Sage 1981. pp. 65 – 115.

Coleman Richard. The Continuing Significance of Social Class to Marketing. Journal of

Consumer Research 1983; 10 (December):265-280.

Coopersmith S. The Antecedents of Self-esteem. NY. Freeman Press, 1967.

Crocker Jennifer. Major Brenda. Social stigma and self-esteem: The Self-protective

properties of stigma. Psychological Review 1989; 96: 608-360

Croizet Jean-Claude. Claire Theresa. Extending the concept of stereotype threat to social

class: The intellectual Underperformance of students from low socioeconomic

backgrounds. Personality and Social Psychology bulletin 1998; 24 (June): 588 – 594.

30

Davis Fred. Bagozzi Richard. Warshaw Paul. User acceptance of computer technology: A

comparison of two theoretical models. Management Science 1989; 35 (August): 982-1003.

Dickerson Mary. Gentry James. Characteristics of adopters and non-adopters of home

computers. Journal of Consumer Research 1983; 10 (September): 225-235

Erumban Abdul. de Jong Simon. Cross-country differences in ICT adoption: A

consequence of Culture?. Journal of world business 2006; 41 (4): 302-314.

Fife Elizabeth. Pereira Francis. Socio-economic and cultural factors affecting adoption of

broadband access: a cross-country analysis. 2002.

www.marshall.usc.edu/ctm/publications/FITCE2002.PDF, accessed in May 2008

Gist Marilyn. Mitchell Terence. Self-efficacy: a theoretical analysis of its determinants and

malleability. Academy of Management Review 1992; 17 (2): 183-211.

Gregan-Paxton Jennifer. How does prior knowledge influence consumer learning? A study

of analogy and categorization effects. Working paper, Department of Business

Administration, University of Delaware, Newark 1999.

31

Gregan-Paxton Jennifer. John, Deborah. Consumer Learning by Analogy: A Model of

Internal Knowledge Transfer. Journal of Consumer Research 1997; 24 (December):266-

284.

Hair. Anderson. Tatham. Black. Multivariate Data Analysis. NJ. Prentice Hall, 1998.

Hamilton Kathy. Catterrall Mitiam. Towards a Better Understanding of the Low Income

Consumer. Advances in consumer Research 2005; 32: 627-632.

Hill Ronald P. Consumer culture and the culture of poverty: Implications for marketing

theory and practice. Marketing theory 2002; 2 (September): 273-294

Hill Ronald P. Gaines Jeannie. The consumer culture of poverty: Behavioral research

findings and their implications in an ethnographic context. Journal of American Culture

2007; 30 (March): 81-95.

Hill Ronald P. Stephens Debra. Impoverished consumers and consumer behavior: The case

of AFDC Mothers. Journal of Macromarketing 1997; 19 (December): 32-48.

Hollingshead August. Redlich Frederick. Social class and mental illness : a community

study , NY: John Wiley, 1967.

32

Johar Gita. Jedidi Kamel. Jacoby Jacob. A Varying Parameter Averaging Model of On-

Line Brand Evaluations. Journal of Consumer Research 1997; 24 (September): 232-248.

Jones Rachel. Ye. Lou. The culture of poverty and African American culture: An empirical

assessment. Sociological Perspectives 1999; 42: 439-459.

Levy, S. Social Class and Consumer Behavior. In: Newman, editor. On knowing the

consumer, NY: John Wiley and Sons, 1966.

Lewis, O. Five families: Mexican case studies in the culture of poverty. NY: Basic Books,

1959.

Marsh Herbert W. Global self-esteem: its relation to specific facets of self-concept and their

importance. Journal of Personality and Social Psychology 1986; 51: 1224 – 1236

Marsh Herbert W. Influences of Internal and External Frames of Reference on the

Formation of Math and English Self-Concepts. Journal of Educational Psychology 1990;

82: 107-116

McCracken Grant. Culture and consumption: A theoretical account of the structure and

movement of the cultural meaning of consumer goods. Journal of Consumer Research

1986; 13 (June):71-84.

33

Moreau Page. Lehmann Donald . Markman Arthur. Entrenched Knowledge Structures and

Consumer Response to New Products. Journal of Marketing Research 2001; 38 (February):

14-29

Moreau Page. Markman Arthur. Lehmann Donald. “What is it?” Categorization Flexibility

and Consumers’ Responses to Really New Products. Journal of Consumers Research 2001;

27 (March): 489-498

Oliver Richard. Winter Russell . A Frame Work for the Formation and Structure of

Consumer Expectations: Review and Propositions. Journal of Economic Psychology 1987;

8 (4); 469-499

Pohjola Matti. The adoption and diffusion of ICT across countries: Patterns and

determinants. In: D.C. Jones, Editor. New Economy Handbook. San Diego, CA: Academic

Press, 2003.

Rammstedt Beatrice. Rammsayer Thomas H. Self-Estimated Intelligence: Gender

differences, relationship to psychometric intelligence and moderating effects of level of

education. European Psychologist 2002; 7 (December): 275-284

Rist R. Students social class and teacher’s expectations: The self-fulfilling prophecy in

ghetto education. Harvard Educational Review 1970; 40: 411-451.

Rogers. Diffusion of innovations, NY: The Free Press, 1996.

34

Rosenberg Morris. Schooler Carmi. Schoenbach Carrie. Rosenberg Florence. Global self-

esteem and specific self-esteem: different concepts, different outcomes. American

Sociological Review 1995; 60 (February): 141 – 156

Schaninger Charles. Social Class versus income: An empirical investigation. Journal of

Marketing Research 1981; 18 (May): 192-208

Shih Chuan-Fong. Venkatesh Alladi. Beyond Adoption: Development and application of a

Use-Diffusion Model. Journal of Marketing 2004; 68 (January): 59-72

Steele C. A threat in the air: how stereotypes shape the intellectual identities and

performance of women and African Americans. American Psychologist 1997; 52:613-629.

Thompson Debora. Hamilton Rebecca. Rust Roland. Feature Fatigue: When product

capabilities become too much of a good thing. Journal of Marketing Research 2005; 42

(November): 431-442

Twenge Jean. Campbell W. Keith. Self-Esteem and Socioeconomic Status: A Meta-

Analytic Review. Personality and Social Psychology Review 2002; 6 (1): 59-71

35

Venkatesh Viswanath. Davis Fred. Morris Michael. Dead or alive? The development,

trajectory and future of technology adoption research. Journal of the Association for

Information Systems 2007; 8 (April): 268 - 286

Wheatley John J. Chiu John S. Stevens Andrea C. Demographics to Predict Consumption.

Journal of Advertising Research 1980; 20 (December): 31-38.

Wood Stacy L. Moreau C. Page. From fear to loathing? How emotions influences the

evaluation and early use of innovations. Journal of Marketing 2006; 70 (July): 44-57.

Yamauchi Takashi. Markman Arthur. Inference using Categories. Journal of Experimental

Psychology: Learning, memory and cognition 2000; 26 (May): 776-795.

Zukin Sharon. Maguire Jennifer. Consumer and consumption. Annual Review of Sociology

2004; 30: 173-197

36

Appendix 1

Correlation of measurement variables

2.

Sel

f -

este

em4

. Kn

ow

led

ge

of

dom

ain

s

1.1

Ed

uca

tio

nal

scal

e

1.2

Res

iden

tial

stra

tum

1.3

Mo

nth

ly

inco

me

1.4

Par

ents

'

edu

cati

on

al

scal

e

3.1

Ver

bal

flu

ency

3.2

Ari

thm

etic

abil

ity

3.3

Inte

rper

son

al

inte

llig

ence

3.4

Lin

guis

tic

abil

ity

3.5

Spa

tial

inte

llig

ence

6.1

Tas

k

dif

ficu

lty

6.2

Tas

k

lear

nin

g ti

me

6.3

Tas

k

chal

len

ge

6.4

Pro

du

ct

dif

ficu

lty

6.5

Pro

duct

lear

nin

g t

ime

2.

Sel

f-st

eem

4.

Kn

ow

ledg

e o

f do

mai

ns

.389

**

1.1

Edu

cati

onal

sca

le.6

95*

*.5

23

**

1.2

Res

iden

tial

str

atu

m.6

52*

*.4

17

**

.73

2**

1.3

Mo

nth

ly i

nco

me

.546

**

.30

5*

*.6

47

**.5

91

**

1.4

Par

ents

' ed

ucat

iona

l sc

ale

.606

**

.44

4*

*.6

81

**.7

44

**

.39

9**

3.1

Ver

bal

flu

ency

.585

**

.34

5*

*.5

35

**.4

97

**

.31

5**

.54

3**

3.2

Ari

thm

etic

ab

ilit

y.6

03*

*.3

92

**

.64

9**

.52

6*

*.4

17

**.5

57*

*.4

88

**

3.3

In

terp

erso

nal

inte

llig

ence

.417

**

.24

1*

*.2

98

**.2

96

**

.25

3**

.26

0**

.14

8*

.17

3**

3.4

Lin

gu

isti

c ab

ilit

y.6

01*

*.3

44

**

.55

0**

.53

5*

*.3

93

**.4

76*

*.4

07

**

.43

0**

.381

**

3.5

Sp

atia

l in

tell

igen

ce.5

74*

*.3

37

**

.53

6**

.45

0*

*.2

96

**.4

75*

*.4

83

**

.46

0**

.320

**

.48

7**

6.1

Tas

k d

iffi

cult

y-.

53

8**

-.3

88

**

-.5

65

**-.

496

**

-.4

22

**-.

48

5**

-.4

19

**

-.4

41

**-.

165

**

-.4

07

**-.

31

8**

6.2

Tas

k le

arn

ing

tim

e-.

60

2**

-.4

97

**

-.6

33

**-.

533

**

-.4

30

**-.

57

8**

-.4

68

**

-.4

68

**-.

247

**

-.4

53

**-.

44

0**

.71

0*

*

6.3

Tas

k ch

alle

nge

-.4

45*

*-.

32

1*

*-.

49

4**

-.49

2*

*-.

37

0**

-.4

98*

*-.

35

9*

*-.

37

7**

-.18

8*

*-.

30

7**

-.2

85*

*.5

57

**

.51

8**

6.4

Pro

duct

dif

ficu

lty

-.5

26*

*-.

33

4*

*-.

51

3**

-.45

8*

*-.

42

6**

-.4

65*

*-.

37

8*

*-.

42

5**

-.1

57

*-.

40

6**

-.3

89*

*.6

42

**

.57

2**

.438

**

6.5

Pro

duct

lea

rnin

g ti

me

-.5

96*

*-.

46

7*

*-.

62

5**

-.54

2*

*-.

45

9**

-.5

75*

*-.

49

6*

*-.

51

1**

-.22

0*

*-.

45

4**

-.4

30*

*.5

88

**

.70

7**

.447

**

.59

2*

*

6.6

Pro

duct

ch

alle

ng

e-.

47

8**

-.3

49

**

-.5

38

**-.

577

**

-.4

05

**-.

55

4**

-.4

38

**

-.3

85

**-.

176

**

-.3

46

**-.

31

1**

.55

5*

*.5

51

**.7

10*

*.4

00

**

.51

6**

** p

< 0

.01

* p

< 0

.05

37

Appendix 2

Measurement

Model

Standardized

Coefficients

Unstandardized

CoefficientsStandard Error Significance

Socioeconomic

level

Educational scale .895 1.000 .059 .000

Residential stratum .842 1.042 .067 .000

Monthly income .655 .829 .074 .000

Parents' educational

scale¨ .797 1.000 NA

Capabilities

Spatial intelligence .631 1.017 .109 .000

Linguistic ability .661 .923 .095 .000

Arithmetic ability .697 1.232 .121 .000

Verbal fluency¨ .651 1.000 NA

Expectations of

complexity

Product challenge¨ .696 1.000 NA

Product learning

time .794 1.099 .092 .000

Product difficulty .706 1.024 .096 .000

Task challenge .665 .907 .090 .000

Task learning time .849 1.229 .097 .000

Task difficulty .802 1.149 .095 .000

¨ = Loading value fixed at 1.000 for estimation purposes

38

Structural ModelStandardized

Coefficients

Unstandardized

CoefficientsStandard Error Significance

Socioeconomic level

Self-esteem .775 7.801 .565 .000

Socioeconomic level

Knowledge of Domains .545 1.244 .137 .000

Socioeconomic level

Capabilities .658 .532 .066 .000

Self - esteem

Capabilities .381 .031 .005 .000

Capabilities

Expectations of complexity -.750 -.836 .098 .000

Knowledge of domains

Expectations of complexity -.151 -.060 .020 .002

Goodness of Fit = 319.5, df = 115, p =.000; GFI=.87; CFI=.92; RMSEA=.082

�

�

�

�

�

�

χ2