Idea Transcript
Social positions and Internet skills A cross-national investigation across European countries
Student:
Roel van Diek
ANR:
s931210
Date:
27-06-2012
Course:
Master Thesis Sociology
Supervisor:
Dr. E.J. (Erik) van Ingen
Reviewer:
Dr. C.M.C. (Ellen) Verbakel
Abstract In this paper, it was examined whether Internet policies affect the relationship between social positions and Internet skills. Based on policy reports by the ITU, OECD and UNESCO, the following three recommendations were investigated; improving Internet in education, stimulating inexpensive broadband connections, and investing in information technologies. In this manner, it was evaluated whether Internet policies contribute to an inclusive information society, that is, a society with a relatively equal distribution of Internet skills across social positions. Hypotheses were tested with Eurobarometer data and multi-level analyses. Results showed that people from high social positions have more Internet skills, and that the difference with people from low social positions is larger in countries with higher quality Internet facilities in education. The remaining policy recommendations did not affect the relationship between social positions and Internet skills. Therefore, this study brings the effectiveness of these recommendations into questions, and calls attention to people of low social position in the development of Internet policies.
Table of contents 1. Introduction .................................................................................................................................... 1 2. Theory and hypotheses ................................................................................................................... 3 2.1. The information society ......................................................................................................... 3 2.2. Differences at the individual level ......................................................................................... 4 2.2.1. Technological appropriation model .............................................................................. 5 2.3. Individual differences in Internet skills ................................................................................. 6 2.3.1. Skill types ..................................................................................................................... 6 2.3.2. Explaining differences in Internet skills ....................................................................... 7 2.4. Differences between European countries .............................................................................. 9 2.5. Explaining differences between European countries ........................................................... 11 2.5.1. Internet in education ................................................................................................... 12 2.5.2. Broadband costs ......................................................................................................... 13 2.5.3. Information technology expenditure .......................................................................... 14 3. Research design ............................................................................................................................ 17 3.1. Sample ................................................................................................................................. 17 3.1.1. Eurobarometer 74.3 .................................................................................................... 17 3.1.2. Sample restrictions ..................................................................................................... 18 3.2. Operationalization ............................................................................................................... 18 3.2.1. Dependent variable: Internet skills ............................................................................. 18 3.2.2. Individual-level variables ........................................................................................... 20 3.2.2.1. Education ........................................................................................................... 20 3.2.2.2. Occupation ......................................................................................................... 20 3.2.3. Country-level variables .............................................................................................. 21 3.2.3.1. Internet in education .......................................................................................... 21 3.2.3.2. Broadband costs ................................................................................................ 21 3.2.3.3. Information technology expenditure ................................................................. 21 3.2.4. Control variables ........................................................................................................ 22 3.2.4.1. Individual level .................................................................................................. 22 3.2.4.2. Country level ..................................................................................................... 22 3.3. Methods ............................................................................................................................... 22 3.4. Grand-mean centering ......................................................................................................... 23
4. Results .......................................................................................................................................... 25 4.1. Illustrating European differences ........................................................................................ 25 4.2. Multi-level regression analysis ............................................................................................ 30 4.2.1. The null-model ........................................................................................................... 31 4.2.2. Random intercepts and random slopes models .......................................................... 31 4.2.2.1. Individual level effects (Model I) ...................................................................... 31 4.2.2.2. Country level effects (Model II) ........................................................................ 33 4.2.2.3. Diverging effects (Model III) ............................................................................ 34 4.2.2.4. Cross-level interactions (Model IV & V) .......................................................... 35 4.2.3. Estimating the models with the exclusion of non-users ............................................. 38 5. Conclusion and discussion ........................................................................................................... 40 5.1. Conclusion ........................................................................................................................... 40 5.1.1. Answering the research questions .............................................................................. 40 5.1.2. Main conclusions ........................................................................................................ 41 5.1.3. Contribution to digital divide research ....................................................................... 42 5.2. Discussion ........................................................................................................................... 43 5.2.1. Alternative Internet policies ....................................................................................... 43 5.2.2. Limitations .................................................................................................................. 44 5.2.3. Recommendations for future research ........................................................................ 45 6. References .................................................................................................................................... 48
1. Introduction The Internet has received widespread attention among scholars since the medium became widespread during the 1990s, and it is argued to be ‘a key tool of economic development‘ (Andrés, Cuberes, Diouf & Serebrisky, 2007). It is at the core of modern society, often referred to as ‘information societies’ (Castells, 2001), and many authors point to the importance of equal usage opportunities (van Dijk, 2006; DiMaggio, Hargittai, Neuman, & Robinson, 2001; DiMaggio, Hargittai, Celeste & Shafer, 2004; James, 2003; Kagami, Tsuji & Giovanni, 2004; Norris, 2001; de Ruijter, Gaay Fortman & Seters, 2003). Several international institutions expect that, if the current distribution of communication technology remains as it is, major disparities, both within and between countries, will increase (ITU, 2010b; OECD, 2001; UNESCO, 2010). Thus, the Internet is considered as a new source of inequality.
But which advantages does the Internet entail? At the individual-level, efficient, or ‘instrumental’ use of the Internet can translate itself into real-life resources (DiMaggio, 2008; van Ingen & Lin, forthcoming; Shah, Kwak, & Holbert, 2001; Shah, McLeod & Yoon, 2001; Sum, Mathews, Pourghasem & Hughes, 2008; Valenzuela, Park & Kee, 2009). Personal contacts can be maintained and created (e.g., Facebook.com), important job-information can be found (e.g., Monsterboard.nl), as well as product information (e.g., Tweakers.net), and equally important, government services are increasingly available on the Internet (Eurostat, 2012a). In this sense, Internet users enjoy benefits over those not using the Internet and, among the online population, advanced users enjoy benefits over the less skilled users. At the country-level, several authors argue that the growth of the developed world during the 1990s was, at least in part, the result of advances made in the ICTsector, especially developments related to the Internet (Castells, 2001; Norris, 2001).
Scholars concerned with Internet inequalities often call attention to the ‘digital divide’ (DiMaggio et al., 2001), which commonly refers to ‘the gap between those who do and those who do not have access to new forms of information technology’ (van Dijk, 2006, p. 221). Initial attempts to understand the digital divide investigated personal or cultural differences between people with and without Internet access (Anderson et al., 1995). More recently, however, scholars improved our insight by bringing attention to inequalities among the online population as well (DiMaggio et al., 2004). More specifically, people can vary considerably with respect to the skills necessary to operate the medium effectively (van Deursen & van Dijk, 2011). In this respect, it is found that the Internet is used in different ways, for different purposes, with more positive outcomes for people 1
from higher social positions (Hargittai, 2008, 2010; Recabarren, Nussbaum & Leiva, 2007), pointing at the existence of a ‘usage gap’ (Bonfadelli, 2002). Between European countries, the proportion of advanced users was found to vary considerably (Brandtzeag, 2011; Ortega, Menéndez, & González, 2007), indicating that some countries have relatively large numbers of people with high Internet skills, while other countries lack behind in this respect.
However, research which specifically examines cross-national differences in Internet skills is scarce. (DiMaggio et al., 2001, 2004). Furthermore, how Internet skills are distributed across social positions, and how countries differ with respect to this distribution is even less the object of scientific studies. This is unfortunate, because this could shed light on policy measures aimed at a more inclusive information society. More specifically, some countries have Internet use highly placed on the political agenda (e.g., Sweden), while other countries have a less active government in this respect (e.g., Italy). Similarly, some countries have a relatively equal distribution of Internet skills across social positions (e.g., the Netherlands), while other countries show relatively large inequalities in this respect (e.g., Ireland). It seems plausible that countries with more active ‘Internet policies’ are those with the most equal distribution of Internet skills across social positions. However, as various authors argue, disparities in Internet skills may result from more traditional inequalities (Kristofic, 2007; Norris, 2001; Willis & Tranter, 2011). To be specific, modern technologies cost money, and operating the Internet requires knowledge and motivation. These conditions are more often met by people from higher social positions, and impede the already disadvantaged groups in society to be fully included. Therefore, closing the usage gap, or the digital divide in general, may require more than policies purely focused on the Internet.
Understanding which people lack the necessary Internet skills to be included in the ‘information society’ (Castells, 2001), and which country characteristics contribute to the development of such skills is necessary in order to form effective policies aimed at closing the usage gap, and the digital divide in general. The aim of this study, therefore, is to contribute to the development of digital divide research, by examining cross-national differences in Internet skills. Additionally, policy recommendations are investigated aimed at closing the so-called usage gap (Bonfadelli, 2002). To be specific, it is examined whether (1) social positions are of influence on Internet skills, (2) the relationship between social positions and Internet skills differs across European countries, and (3) to what extent these differences can be attributed to Internet policies.
2
2. Theory and hypotheses In this section the theoretical background for this study will be provided. First, it will be outlined what is understood by the information society. This is necessary in order to fully comprehend the Internet as a new source of inequality. Second, the most important digital divide research findings are outlined, using the technological appropriation model proposed by van Dijk (2005). Third, literature regarding Internet skills is discussed, leading to expectations about the influence of socioeconomic status on such skills. Fourth, research findings considering cross-national differences in Europe are outlined, which leads to the expectation that the relationship between social positions and Internet skills is different across European countries. Fifth, it is hypothesized how this relationship depends on three country characteristics, using policy-oriented reports from the International Telecommunications Union (ITU), the United Nations Educational, Scientific and Cultural Organization (UNESCO), and the Organization for Economic Co-operation and Development (OECD).
2.1. The information society The information society is often contrasted to the industrialized society on two aspects in order to clarify its meaning. First, the importance of information is emphasized, which is often indicated by the proportion of the working-force employed in service-related occupations. More specifically, in information societies people earn their money with ‘gathering, producing, processing, and selling information’ (Ultee, Arts & Flap, 2003, p. 203). This is in contrast to industrialized societies, with relatively more people employed in manufacturing and construction related occupations (Ultee et al., 2003). Second, the importance of ICT, and the Internet in particular, has increased substantially. It is argued that, in order to compete or ‘keep up’ in society, or as a society as a whole, it is necessary to be connected to the Internet network. McDaniel (2004), for example, argues that ‘heavy reliance on inferior technologies reduces competitiveness for companies operating in open markets, and for individuals limits their options for information and communication’ (p. 161). Therefore, being excluded from the most dominant, and widely used ‘stream of information’ could lead to considerable disadvantages (Castells, 2001).
In other words, information societies are characterized by a focus on information processing capabilities, and ‘the Internet [...] at the heart of the new socio-technical pattern’ (Castells, 2001, p. 265). More specifically, an individual can maintain or appropriate a position in society, when it has access to certain information valuable to others, or when it is able to process or evaluate 3
information. With most organizations being structured around the Internet, and information primarily communicated online, social positions will be determined, in part, by the individual’s ability to operate the medium effectively. Thus, in order to be included in the information society, it is necessary to have Internet skills.
Various authors, however, point the existence of the so-called ‘knowledge gap’ (Price & Zaller, 1993; Viswanath, Kosicki, Fredin & Park, 2000). This concept is often used to describe how people from different social positions differ in terms of media use. The main argument is that people from lower social positions are more likely to use media for entertainment purposes, while people from higher social positions use the media relatively instrumental. Consequently, people from higher social positions gain more information through the media, which they can use for their own benefit (Price & Zaller, 1993; Viswanath et al., 2000). Similarly, some authors (Bonfadelli, 2002; van Dijk, 2005) argue that the Internet has led to a usage gap ‘between people of high social position, income, and education using advanced computer and Internet applications for information, communication, work, business, or education and people of low social position, income, and education using simpler applications for information, communication, shopping, and entertainment’ (van Dijk, 2005, p. 130). Both ‘gaps’ are explained by inequalities in terms of human (e.g., intelligence, prior knowledge), social (e.g., relevant social contacts), and private capital (e.g., money), and both gaps predict that people from higher social positions, who are advantaged already, benefit the most from the media, and the Internet in particular (Wei & Hindman, 2011). Therefore, it is not evident that the usage gap, or digital divide at large, can be closed at all, as it may result from more traditional inequalities (Bonfadelli, 2002).
2.2. Differences at the individual level Understanding how people develop Internet skills requires a theoretical framework. The technological appropriation model, proposed by van Dijk (2005), provides such a framework. Furthermore, the model helps to clarify which achievements have been made in digital divide research during the last two decades. In the model, Internet access is approached as ‘a process with many social, mental and technological causes and not as a single event of obtaining a technology’ (van Dijk, 2006, p 225), distinguishing between motivational, material, skills, and usage access.
4
2.2.1. Technological appropriation model Before someone can develop Internet skills, he or she needs to be motivated to get access to computer and Internet technology. Research concerned with this stage has shown that there are people unwillingly unconnected to the Internet, the ‘have-nots’, and people willingly unconnected, the ‘want-nots’ (Finn & Korukonda, 2004; Rojas, Straubhaar, Roychowdhury & Okur, 2004). Those with sufficient resources and motivation to get Internet access are likely to be connected, leading to material access.
The second stage, material access, has received the most attention in digital divide research. The most recent results indicate that education, age, and occupation are important dividing lines in this respect (Eurostat, 2005; Lengsfeld, 2011). More specifically, higher educated, young people with high incomes have the largest probability of having Internet access. Among those with Internet access, not everybody has the same ‘amount’ of Internet available in his or her social or physical environment. This is usually expressed with the term ‘autonomy of use’ (Hargittai, 2008), and is measured by looking at how many connection points are available to an individual. For example, a university student with Internet access in his room and at school, has more autonomy of use than a comparable student with access only at the school library.
With Internet access, the individual is able to pass on to the third stage, that is, skills access. This concerns the development of certain skills to operate the Internet. The last decennia, this stage has received increased attention among scholars, and research has shown that those from higher social positions, and those with more autonomy of use, develop more skills (van Deursen & van Dijk, 2011; DiMaggio et al., 2004; Hargittai, 2008, 2010). Furthermore, it appears that men have higher Internet skills than women, indicating that the gender-gap is not fully closed yet (Hargittai, 2010). These findings suggest that, while the initial focus of digital divide research was on material access, variations in skills have to be taken into consideration in order to fully understand the digital divide. What is understood by Internet skills is outlined more thoroughly in the next paragraph, as this is the main concern of this study.
The fourth stage, usage access, reflects the actual use of the Internet. Of course, the Internet can be used in a variety of ways, and, generally, it is found that socio-economic status increases capitalenhancing activities, suggesting the existence of a usage gap (Bonfadelli, 2002; van Dijk, 2004, 2005, 2006; Shah, Kwak & Holbert, 2001; Wei & Hindman, 2011). Moreover, it seems that those 5
equipped with more skills have a broader usage pattern, and their use is relatively instrumental (van Deursen & van Dijk, 2011; van Dijk, 2004; Wei, 2012). These findings indicate that people from higher social positions are better capable of using the Internet for ‘real-life’ goals.
Summarizing, most digital divide research has been concerned with material access, and although this brought attention to large inequalities in this respect, it has been accused of approaching the Internet to simplistic (Brandtzaeg, 2010, 2011; van Dijk, 2006; DiMaggio et al., 2001, 2004; Shah, Kwak & Holbert, 2001). By focussing purely on access, it is argued that the extent too and the way in which the Internet is used, as well as the benefits derived from it, are not taken into consideration. Consequently, the magnitude of the digital divide is seen as being underestimated when focussing purely on access inequalities. For example, gender may not be a dividing line with respect to Internet access anymore. However, men were found to be considerably more skillful in using the Internet than women, which implies a digital divide, although distinct in nature (Hargittai, 2008, 2010). Concluding, the technological appropriation model (van Dijk, 2005) goes beyond the focus on material access, and brings differences in Internet skills between people from different social positions into consideration.
2.3. Individual differences in Internet skills Various research findings indicate that Internet skills are not equally distributed across the population. More specifically, people from higher social positions seem better capable of using the Internet for their own benefits compared to people from lower social positions (Bonfadelli, 2002; van Deursen & van Dijk, 2010, 2011; van Deursen, van Dijk & Peters, 2011; van Dijk, 2004; DiMaggio, 2008; Hargittai, 2008, 2010; van Ingen & Lin, forthcoming; Recabarren et al., 2007). The question remains, however, why socio-economic status has a positive influence in this respect. In order to clarify this, it is necessary to outline a) what is understood by Internet skills, and b) how basic skills differ from advanced skills.
2.3.1. Skill types Van Deursen & van Dijk (2011), elaborating on Steyaert (2000), distinguish four types of skill, which have a ‘sequential and conditional nature’ (p. 895). Operational skills are required first, which facilitates handling basic software and hardware problems. This refers to activities such as turning on a computer, opening a website, and executing a search operation. Subsequently, formal skills are developed. This means understanding the structure of the Internet, and ‘requires the skills 6
of navigation and orientation’ (van Deursen & van Dijk, 2011, p. 895). This is reflected by the ability to keep sense of orientation when navigating within or between websites and search results, as well as using more sophisticated means of navigation, such as embedded hyperlinks (van Deursen, 2010).
Subsequently follow information skills, and this reflects the ability to ‘search, select, and process information in computer and network sources’ (van Dijk, 2006, p. 228). Specifically, this means being able to choose which website is most suitable for the desired information, as well as evaluating the source of this information. Finally, strategic skills can be developed, which reflect the capacity to ‘use the Internet as a means of reaching particular goals and for the general goal of improving one’s position in society’ (van Deursen & van Dijk, 2011, p. 895). This implies being able to focus on a specific goal, while making decisions in order to gain benefits from this goal (e.g., finding an inexpensive holiday-trip).
Operational and formal skills, which can be considered as basic skills, enable the individual to use the Internet, and research has shown that people from higher social positions are advantaged in this respect (Hargittai, 2010; van Deursen & van Dijk, 2011). The more advanced skills, that is, information and strategic skills, reflect the ability to assess online content, and the latter type specifically refers to the ability to transform online content into real-life resources. Research has found that, similar to basic skills, socio-economic status positively relates to advanced skills (Hargittai, 2008, Recabarren et al., 2007; van Deursen & van Dijk, 2011), indicating large skill differences between people from different social positions.
2.3.2. Explaining differences in Internet skills But why are people from higher social positions more skillful in using the Internet? In the literature, several factors can be found in order to explain this relationship, which relates to someones information processing abilities (Grabe, Lang, Zhou & Bolls, 2000), relevant social contacts (Bonfadelli, 2002, p. 68), incentives (Ettema and Kline, 1977), and available resources (Guichard, 2003).
Firstly, higher educated people ‘are better at encoding information from audio-visual media...’ (Grabe et al, 2000, p. 20). As advanced skills require such encoding, or processing of
7
information (Parelta, 2003), it can be expected that high levels of Internet skills are found among higher educated people.
Secondly, people from higher social positions ‘are integrated in broader social and/or local networks that function as additional interpersonal information resources’ (Bonfadelli, 2002, p. 68). Moreover, high income and educational settings are structured around the Internet, while this is less likely the case with respect to low income, and lower educational settings. As a result, people from lower social positions receive less support when encountering problems with the Internet, considerably impeding the development of Internet skills (Rojas et al., 2004).
Thirdly, people have to be motivated to process information (Ettema and Kline, 1977). In order to transform online information into real-life resources, relatively high levels of information processing is involved. It can be argued, therefore, that people have to be motivated, or need an incentive, in order to develop advanced skills. Generally, people with high incomes have occupations that force them to use the Internet (e.g., e-mail), and efficient use of the medium often leads to considerable ‘time-savings’. The same holds for higher educational settings. People from lower social positions, on the other hand, are less often confronted with the requirement of efficient Internet use. It can be argued, therefore, that people from higher social positions have a larger incentive to develop Internet skills.
Fourthly, having Internet access at home requires a considerable private investment, pointing at the importance of income (Guichard, 2003). Higher income groups often have more advanced equipment, and multiple devices connected to the Internet. This increases their autonomy of use, which, in turn, increases their level of Internet skills (Hargittai, 2008). People from lower social positions, on the other hand, often cannot afford the same equipment, resulting in less Internet skills.
Summarizing, the positive relationship between socio-economic status and Internet skills can be attributed to four factors. In general, people from higher social positions have (1) better information processing capabilities (Grabe, 2000; Parelta, 2003), (2) more relevant contacts (Bonfadelli, 2002; Rojas et al., 2004), (3) a larger incentive to use the Internet, and (4) higher incomes (Guichard, 2003), which all contribute to the development of Internet skills. Based on these theoretical considerations, and the previously discussed research findings, it is expected that: 8
(h1a) Education is positively related to Internet skills.
(h1b) Occupation is positively related to Internet skills.
It should be noted, that occupation is treated as a proxy for income in this study. This is due to missing data on income, while rather extensive information is available with respect to occupation. It is assumed that people from a higher occupational ‘class’ generally have higher incomes. Therefore, the mechanisms outlined to explain differences in Internet skills between people from different occupational classes focus on the influence of income in this respect. In section 3, this issue will be discussed more thoroughly.
Figure 1. Conceptual model hypotheses 1a and 1b.
2.4. Differences between European countries Substantial differences also exist between countries, both in terms of Internet access (Castells, 2001; Norris, 2001), and usage patterns (Brandtzaeg, 2010, 2011; Montagnier & Wirthmann, 2011; Ortega, Menéndez, & González, 2007). Investigating over 170 countries, Norris (2001) found that the relationship between Internet access and economic development is strong, and suggests that the Internet ‘represents one more disparity reflecting the poverty of those living in developing nations’ (p. 55). He also points at two findings which highlight the importance of other country characteristics. Firstly, once a country reaches a certain level of prosperity, more economic development does not necessarily lead to more Internet use. Secondly, some countries (including Slovakia, Slovenia, Poland, and Estonia) were found to have higher levels of Internet access than would be expected according to their level of economic development. Both findings indicate that other factors, possibly at the country level, may be accountable for cross-national differences in Internet use. 9
Ortega and his colleagues (2007), using a cluster analysis, identified five Internet user types, that is, ‘laggards, confused and adverse, advanced users, followers, and non-Internet users’ (p. 6). Examining 15 E.U. member states, they concluded that the most advanced users can be characterized as high educated men, employed in the service sector, and living in urban areas. Among the less advanced user types they have found a higher proportion of women, unemployed, lower educated, and people living in rural areas. Furthermore, they found considerable differences between countries, pointing at the existence of a North-South divide in Europe. For example, Greece, Portugal, Spain, and Italy were found to have a high share of non-Internet users (over 50%), accompanied by a small proportion of advanced users (below 10%). On the other hand, countries such as the United Kingdom, Finland, Denmark and Sweden were found to have a high share of advanced users (over 25%), while having relatively few non-Internet users (below 30%).
Brandtzaeg (2011), elaborating on Ortega et al. (2007), also identified five user types, that is, ‘nonusers, sporadic users, entertainment users, instrumental users, and advanced users’ (p. 129). Across the five European countries investigated, he found that each user type can be identified in every country, but with considerable differences in their distribution. In Spain, for example, few instrumental or advanced users exist, while in Norway, these user types represent a considerable proportion of the online population. Trying to explain these user types, he found that Internet access and age are important predictors, but that these factors vary in their explanatory power across countries. For example, in Spain and the United Kingdom, age and gender were found to be important predictors of being an advanced user, while in Sweden and Norway, only access was found be a significant predictor.
Montagnier and Wirthmann (2011) investigate the intensity of Internet use across 19 European countries. Performing logistic regressions country by country, they concluded that the influence of various explanatory variables differed considerably across countries. Firstly, income was found to have a stronger influence in Finland and Norway, when compared to Hungary. Secondly, education was found to have a stronger influence in Portugal, when compared to the European average. Thirdly, in Sweden, unemployed persons were found to have a higher odds for being an intense Internet users, than those employed. Fourthly, Estonian women have a higher probability of being an intensive Internet user than men. These findings suggest that the dividing lines are different across countries, and that it is useful to take into account country characteristics in order to fully understand the digital divide. 10
Concluding, it seems that, across European countries, the proportion of advanced users differs considerably. More specifically, Northern European countries show the most advanced Internet behavior, while the Southern European countries lag behind (Brandtzeag, 2010, 2011; Ortega et al., 2007). Moreover, the influence of various explanatory variables were found to be different (Brandtzeag, 2011; Montagnier & Wirthmann, 2011), including income and education. Therefore, it is expected that:
(h2a) The relationship between education and Internet skills differs across European countries.
(h2b) The relationship between occupation and Internet skills differs across European countries.
2.5. Explaining differences between European countries Recent cross-national studies indicate that the relationship between social positions and Internet skills is different across European countries, and this may be attributable to factors at the country level. Similarly, DiMaggio et al. (2004) argue that ‘digital inequality reflects not just differences in individual resources, but also the ways in which economic and political factors make such differences matter’ (p. 47). Regarding inequalities in Internet skills, few cross-national studies exist that investigate the influence of such country characteristics. However, the International Telecommunications Union (ITU), the United Nations Educational, Scientific and Cultural Organization (UNESCO), and the Organization for Economic Co-operation and Development (OECD) have recommend several policy measures to stimulate Internet use, designed to ‘reduce the digital divide’ (OECD, 2001), or aimed at a ‘inclusive information society’ (ITU, 2010b; UNESCO, 2010). These general goals suggest that, besides a positive effect on society at large, the recommendations are aimed at reducing inequalities within countries as well.
The OECD, ITU and UNESCO all stress the importance of Internet access in education and affordable broadband connection fees. Additionally, it is suggested that quality ‘e-government’, which refers to the quality of online government services, can be a meaningful measurement. Since English is the most dominant language on the Internet, stimulating local content is also suggested to get people online, especially the less educated. This study will focus on the first two measures, because the latter two are difficult to compare cross-nationally. Thereafter, the influence of 11
information technology (IT) expenditure is discussed, because investments in IT’s can be considered as the ‘driving force’ behind the first two measures.
2.5.1. Internet in education Schement (2003) found that Internet access at public schools increased dramatically in the United States during 1994 and 1999, and he argues that this was the result of an effective policy program by the Clinton Administration. However, it is argued that providing sheer access is not enough (van Dijk, 2006; ITU, 2010b; UNESCO, 2010). In order to increase Internet skills among learners, teachers with sufficient technological knowledge are needed (ITU, 2010b, UNESCO, 2010). Furthermore, with respect to developed countries, the importance of broadband Internet is stressed, because slow connection speeds are not suitable for educational purposes (ITU, 2010b). Therefore, when Internet access in education is examined cross-nationally across Europe, it is necessary to consider the quality of ICT facilities (e.g., knowledge of teachers, hardware/software quality), besides Internet access in general.
Thus, how does quality Internet access in education increase Internet skills among individuals? Firstly, it gives people from all social positions the opportunity to get acquainted with the Internet, and to receive support when having difficulties operating the medium. Furthermore, when students are being stimulated to use the Internet for educational purposes, this could increase their perceived benefits of the medium. More specifically, if useful information is found, this could lead students to conclude that the Internet may be useful in other domains of life. This provides an incentive to further develop Internet skills. Consequently, it can be expected that, in such a context, people will develop more Internet skills, when compared to a country with less advanced Internet facilities in education. Thus, quality Internet access in education increases familiarity and support for all students, which allows them ‘to acquire new sets of skills required for the information society’ (ITU, 2010b, p. 29). Therefore, it is expected that:
(h3a) In countries where Internet use is stimulated in educational settings, people will have more Internet skills on average.
Additionally, it may be that Internet in educational settings is relatively beneficial for people from lower social positions. To be specific, people from lower social positions less often have Internet access at home, and are less likely to receive social support when encountering problems with the 12
Internet (Rojas et al., 2004). In this sense, providing schools with Internet is of special importance to those without Internet at home, as this enables them to get acquainted with the medium. Therefore, it can be argued that, in countries where Internet use is stimulated in educational settings, smaller barriers exist for people from lower social positions to get online. Therefore, it is expected that:
(h3b) The relationship between education and Internet skills is weaker in countries that stimulate Internet use in education than in countries that do not.
It should be noted that the latter expectation is less evident when it relates to higher educational settings. To be specific, people who reach tertiary education are often those from higher social positions. It can be argued, therefore, that hypothesis 3b is valid only when it applies to Internet access in primary, and secondary schools. Both the ITU (2010b) and UNESCO (2010), however, stress the importance of Internet in tertiary education as a means to a more inclusive information society.
2.5.2. Broadband costs The second country characteristic concerns the general accessibility of the Internet, which is reflected by affordable connection fees, and the price of ‘up to date’ software and hardware. Strover (1999) argues that rural areas in the U.S. experienced higher connectivity rates, due to less competition among Internet service providers, resulting in fewer households online. Similarly, both the OECD (2001), and the ITU (2010b) have stressed the importance of stimulating competition among Internet service providers (ISPs), because this generally leads to lower connection rates, resulting in more people online. Taken into account that people with broadband-connections have more Internet skills than those with narrowband-access (van Dijk, 2006), it is important to consider the cost of a broadband connection.
But how does affordable broadband Internet translates itself into more Internet skills among individuals? Many current online applications require a broadband connection. When broadband connections are inexpensive, more people will be able to use these up-to-date applications. In a country with expensive broadband connections, on the other hand, people will have a narrower
13
usage pattern in general, and it seems plausible that this decreases the average Internet skill level (Broadband Commission, 2010). Therefore, it is expected that:
(h4a) In countries where broadband connections are more expensive, people will have less Internet skills on average.
Additionally, it could be that inexpensive broadband Internet primarily benefit people from lower social positions. A cheaper connection reduces the required income to have a broadband connection. This, in turn, enables more people from low income groups to have broadband Internet, therewith increasing the probability that advanced Internet skills are developed. Relatedly, people from the same social position are often integrated into similar intermediate groups (Berkman, Glass, Brisette & Seeman, 2000; Rubin, 2012; Ultee et al., 2003). Therefore, it can be argued that, under the condition that higher broadband penetration increases Internet skills among people from lower positions, more support can be given to those in a similar social position, when problems are encountered with the Internet. In this sense, inexpensive broadband connections are relatively beneficial for people from lower social positions. Therefore, it is expected that:
(h4b) The relationship between occupation and Internet skills is stronger in countries with expensive broadband connections than in countries with inexpensive broadband connections.
2.5.3. Information technology expenditure The third country characteristic concerns the level of information technology (IT) expenditure. This indicates how developed a country is with respect to the Internet. More specifically, a high IT expenditure means that the government, corporations, and consumers together spend a relatively large amount of money on Internet related technologies. Therefore, it can be argued that countries with a relatively high level of IT expenditure place more importance on the Internet, and that advanced technologies will be more widespread, when compared to countries with a low IT expenditure (McDaniel, 2004). It seems plausible that this affects people’s Internet skills in, at least, three ways.
First, the level of IT expenditure reflects the importance of the Internet as perceived by the government. More specifically, countries with high IT expenditure are expected to stimulate 14
Internet use more actively, compared to countries with low IT expenditure. Consequently, in countries with an active Internet policies, people will use the Internet more frequently, leading to higher Internet skills.
Second, basic Internet infrastructure requires a large investment. In countries with high IT expenditure there are more organizations or stakeholders involved bearing the costs. Furthermore, there is more competition among ISPs, and often, there is less reliance on inferior technologies. These factors all contribute to less expensive broadband connections, and to a more widespread use of the Internet in general (McDaniel, 2004), which, in turn, contribute to the development of Internet skills.
Third, as most educational settings are financed by the government, it can be assumed that higher IT expenditure positively influences Internet in education. Formulated differently, Internet at schools requires an investment, and, in this sense, suggests dependence on governments. Therefore, the availability of Internet in schools will be a reflection of the government’s perceived necessity of the medium, which is expressed by the level of IT expenditure. Therefore, it is expected that:
(h5a) In countries with relatively high IT expenditure, people will have more Internet skills on average.
Additionally, it seems plausible that IT expenditure primarily benefits people from lower social positions. As argued, relatively high levels of IT expenditure lead to less expensive broadband prices, and to better Internet facilities in educational settings. As people from lower social positions are expected to benefit the most from both inexpensive broadband connections and Internet in education, it is hypothesized that:
(h5b) The relationship between education and Internet skills is weaker in countries with relatively high levels of IT expenditure.
(h5c) The relationship between occupation and Internet skills is weaker in countries with relatively high levels of IT expenditure.
15
Figure 2. Conceptual model hypotheses 4, 5 and 6.
16
3. Research design In this section the research design is outlined. First, the sample characteristics of the Eurobarometer are outlined, and is argued that the Eurobarometer 74.3 is sufficient to answer the research questions at hand. Second, the operationalization of the dependent, independent, country, and control variables is outlined respectively. Third, the procedure of the multi-level regression analysis is explained, and the issue of multicollinearity is examined. Fourth, it is explained what grand-mean centering entails, and why this procedure is necessary for the interpretation of the results. Finally, the descriptive statistics are presented in Table 1.
3.1. Sample 3.1.1. Eurobarometer 74.3 The Eurobarometer 74.3 (2010) is used in order to examine the relationship between social positions and Internet skills across European countries. This cross-sectional survey is part of the Eurobarometer, a program launched in the early 1970s, which is aimed at mapping the attitudes of European citizens with respect to various topics (see http://ec.europa.eu/public_opinion/ for more information).
The Eurobarometer 74.3 is used because of three reasons. First, it is a large-scale, cross-national survey, representative for almost all E.U. member states. More specifically, face-to-face interviews were conducted in people’s home in the national language, and respondents were selected through a multi-stage, random probability design. In total 26,574 respondents were included in the sample, with around 1,000 respondents per country. Second, it contains detailed questions about Internet use, while most large-scale surveys do not. To be specific, respondents were asked to indicate which types of activities they perform online, while most surveys only include questions about the frequency of use. Thus, the Eurobarometer 74.3 is one of the few large-scale, cross-national surveys that enables a more detailed investigation of cross-national differences in Internet use. Third, it is argued that the Internet is a ‘moving target’ (DiMaggio et al., 2001). The Internet, being a relatively new medium, changes continuously due to an increasing number of applications and users. Therefore, ensuring that the findings represent how things are at the moment requires ‘up-to-date’ data. The Eurobarometer 74.3, being obtained in November-December 2010, meets this requirement.
17
3.1.2. Sample restrictions Malta and Cyprus are excluded from the analysis, because of missing data on IT expenditure. Northern-Ireland and East-Germany, which are independent ‘countries’ in the original dataset, are added to the United Kingdom and West-Germany respectively. As a result, 25 countries are included in the analysis, which are; Austria (AT), Belgium (BE), Bulgaria (BG), Czech Republic (CZ), Germany (DE), Denmark (DK), Estonia (EE), Spain (ES), Finland (FI), France (FR), United Kingdom (GB), Greece (GR), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), Luxembourg (LU), Netherlands (NL), Poland (PL), Portugal (PT), Romania (RO), Sweden (SE), Slovenia (SI), and Slovakia (SK).
Additionally, two restrictions have been made. First, students are excluded from the analysis, because it is unknown at what age they complete their full-time education. Second, people under the age of 18, as well as people older than 90 years are excluded. This is necessary because few respondents were present in these age groups. Thereby, missing values were counted with respect to all variables. Respondents with missing values on at least one variable are excluded from the analysis, resulting in a total of 22,956 respondents.
3.2. Operationalization 3.2.1. Dependent variable: Internet skills The Eurobarometer 74.3 contains several questions about Internet use. Firstly, respondents were asked to indicate whether they use the Internet ‘every day’, ‘two or three times a week’, ‘once a week’, ‘one or two times a month, ‘less’, ‘never’, or have ‘no access’ at home, work, or somewhere else. Secondly, respondents that use the Internet at one place at least, were asked to indicate whether they use it for ‘sharing pictures, video’s etc.’, ‘social network sites’, ‘e-shopping’, ‘keeping a blog’ , ‘instant messaging’, ‘peer-to-peer software’, ‘telephone- or video calls’, ‘installing plug-ins’, ‘sustaining a website’, ‘online banking’, ‘government services’, and ‘online software’. Some respondents indicated, spontaneously, that they use ‘other’ applications, or that they ‘don’t know’ which specific activities they perform online.
In order to measure Internet skills, a proxy is constructed as follows. First, respondents that never use the Internet, and those without access score ‘0’. Second, for those who use the Internet, the total number of online activities are summed, providing a proxy of Internet skills. Third, some respondents indicated that, besides the activities provided, they use the Internet in other ways as 18
well. For those respondents, an extra activity is added on top of the original score. Fourth, respondents indicating that they ‘don’t know‘ which activities they perform get a score of ‘1’. Fifth, one respondent had a score of 13, meaning that she uses all applications in the questionnaire, and has indicated, spontaneously, that she uses other web-applications as well. This respondent is added to the people indicating that they use 12 online applications. Resulting is a 12-point scale, ranging from 0 (non-users) to 12 (12 activities).
It can be questioned, however, whether the term ‘Internet skills’ is the most suitable one. First, it can be argued that, instead of skill differences, differences in diversity or scope of use are measured. More specifically, since the total number of activities are summed, no attention is given to the difficulties of these activities (e.g., sustaining a website requires more skill than instant messaging). However, former research suggests that the two are empirically not as distinct as they are conceptually different. Wei (2012), for example, found that skills increase with a broader usage pattern. In addition, Hargittai (2008, 2010) found that instrumental use, which requires more skill, increases with a broader usage pattern as well. Considering that one objective of this study is to evaluate policy recommendations on their effectiveness to contribute to a more inclusive information society, it makes more sense to address the issue of Internet skills. To be specific, skills have to be improved in order to include disadvantaged groups in the information society, and skills have to be the aim when policy recommendations are formulated (van Dijk, 2008; ITU, 2010b). It is argued, therefore, that the term ‘Internet skills’ is suitable.
Second, it can be questioned whether this scale only measures differences in Internet skills, or whether it primarily captures differences between users and non-users. Therefore, the multi-level models are also estimated, while excluding the non-users (9,726). The result is a 11-point scale, ranging from 1 (1 activity) to 12 (12 activities). This ‘user-model’ (N1=13,990) is compared to the original results, and the differences are discussed in section 4. For now, it can be said that there a few differences, indicating that the 12-point scale, ranging from non-users (no skills, no benefits), low users (low skills, small benefits), to advanced users (high skills, large benefits), is a robust measurement.
19
3.2.2. Individual-level variables 3.2.2.1. Education Education is measured with the question ‘How old were you when you stopped a full-time education?’. Some respondents did not complete a full-time education (121), did not know (321), refused (110), or were still studying (2224). Respondents that stopped studying before the age of 13, as well as people indicating that they have not completed a full-time education, are put together in one category. Respondents that stopped studying above the age of 30 are put together as well. Resulting is a scale, ranging from ’13 or younger’ to ’31 or older’, which is used to measure educational differences. Respondents which did not know, refused, or were still studying are considered as missing values, because it is unclear at which age they have completed, or will complete, a full-time education.
3.2.2.2. Occupation In order to measure income differences between occupations, the current or last occupation will be used. The question used is ‘What is your current occupation?’. Respondents indicating that they are retired or disabled, unemployed or responsible for the household were asked for their last occupation. For those respondents without paid work at the time of survey, the last occupation is used as an indicator. Based on the EGP schema (Rose, 2005), originally developed by Erikson, Goldthorpe and Portocarero (1979), the categories are recoded to a 7-point scale. These are unemployed (1), unskilled manual workers (2), skilled manual workers (3), supervisors, small selfemployed, and farmers (4), routine non-manual workers (5), professionals, higher grade (6), professionals, self-employed, higher grade (7).
According to Rose (2005), these categories, or classes, can be used ‘as far as overall economic status is concerned’ (p. 5). More specifically, each upward step on the scale leads to increases ‘in terms of greater long term security of income, being less likely to be made redundant; and a better prospect of a rising income over the life course’ (Rose, 2005, p. 5). Therefore, the EGP schema is suitable to measure income differences, which is the factor under consideration in this study. Furthermore, the scale approaches a linear relationship with both education and Internet skills.
20
3.2.3. Country-level variables 3.2.3.1. Internet in education In order to measure the quality of Internet facilities in educational settings, data from the Global Competitiveness Report (2010), presented by the Word Economic Forum (WEF), are used. To provide, otherwise unavailable, cross-national comparable data, the WEF conducts the ‘Executive Opinion Survey’. The respondents in this sample are mostly drawn from ‘recognized economics departments of national universities, independent research institutes, or business organizations’ (WEF, 2010, p. 58). The Executive Opinion Survey is divided into thirteen sections, and respondents evaluate various issues on a 7-point scale, primarily in the domain of politics, economics, and techology. The question at hand is asked in the section ‘innovation and technology’, and is formulated as follows: ‘How would you rate the level of access to the Internet in schools in your country?’ (1=very limited, 7=extensive). This measurement, therefore, provides an indication of the quality of Internet access across all educational levels.
3.2.3.2. Broadband costs As a proxy of broadband costs, data from the International Telecommunications Union (ITU, 2011) is used. This measurement is calculated on the basis of a fixed broadband Internet tariff for a monthly usage of 1 Gigabyte, measured in US$ (corrected for purchasing power) as a percentage of monthly GNI per capita (ITU, 2011, p. 54). Thus, a higher score on this variable indicates a more expensive broadband connection. This measurement is part of the ‘ICT Price Basket’ (IPB), which is often used to evaluate the affordability of information-and communication technology for consumers cross-nationally.
3.2.3.3. IT expenditure In order to measure development in Internet related technology, the expenditure on IT per capita is used. This includes expenditure for IT hardware, equipment, software and other services (Eurostat, 2012c). Eurostat provides IT expenditure as a percentage of GDP. This percentage is used to calculate the IT expenditure in euro’s as a total amount of GDP per capita. Thus, a high score on this variable indicates a relatively high expenditure on Internet technologies per capita. The score for each country is divided by 1,000 and then entered in the analysis.
21
3.2.4. Control variables 3.2.4.1. Individual-level Two individual characteristics that might explain inequality within and between countries must be taken into account. Age has been found to be related to Internet use, and therefore is included as a control variable. Furthermore, it approaches a linear relationship with Internet skills. Additionally, controlling for gender is necessary, because men have more Internet skills (Hargittai, 2010), as well as higher incomes (Eurostat, 2012b).
3.2.4.2. Country-level At the country-level, GDP per capita will be included as a control variable. It seems plausible that GDP affects the variables of interest. That is, it increases the average Internet skill in a country, as well as the quality of Internet facilities in educational settings. Moreover, a larger GDP is associated with a higher educated population, as well as a higher income on average. Thus, it seems plausible that the level of economic development in a country, measured in GDP per capita, is related to several explanatory variables incorporated in the model. In order to control for the possibility of a spurious relationship, it is therefore necessary to include GDP. The data are derived from Eurostat (2012d), and the score for each country is divided by 1,000.
3.3. Methods The assumption is that individual, and country characteristics can explain differences in Internet skills. In order to analyze individual, country and cross-level interaction effects simultaneously, a multilevel analysis will be carried out. The first step of the analysis is to estimate the null-, or empty-model, which shows how much of the variability in Internet skills can be explained by country clustering. More specifically, the null-model enables the calculation of intra-class correlation (ICC), which indicates how much of the variability in Internet skills occurs between countries. Secondly, the random intercept model with the individual-level variables will be estimated. This model will be compared to the empty-model, and gives insight in the amount of variance explained by the individual-level variables at both the individual, and country level. Thirdly, the country-level variables will be added to the random intercept model. Likewise, the model is compared to the empty-model in order to examine the explained variance at both individual, and country level. Fourth, the random slope model will be estimated, in which it is tested whether the effect of education and occupation varies across countries. Here, the model will also be compared to the empty-model to examine the explained variance at both the individual, and 22
country level. Fifth, the random slope model with cross-level interaction effects will be estimated, so that it can be investigated to what extent the relationship between social positions and Internet skills depends on country characteristics.
Thus, first is tested whether Internet skills differ between European countries, based on the nullmodel. Following are the random intercept models, in which it is examined to what degree variation in Internet skills can be attributed to both individual and country differences. Thereafter, the random slope model, in which is it is examined whether the effect of social positions is different across countries. Finally, the cross-level interaction models, in which it is tested whether Internet in education, broadband costs, and IT expenditure affect the relationship between socio-economic status and Internet skills.
Additionally, it was tested whether multicollinearity exists between the explanatory country level variables. Three separate regression models were estimated, with each of the country variables as the dependent variable. IT expenditure and broadband costs explain 24,9% of the variance in Internet in education. IT expenditure and Internet in education explain 43,7% of the variance in broadband costs, while Internet in education and broadband costs together explain 53,1% of the variance in IT expenditure. Therefore, it is concluded that the country level effects are not affected by multicollinearity.
3.4. Grand-mean centering All explanatory variables, with the exception of gender, are grand-centered around their means. Grand-mean centering implies that, for each respondent, the overall mean score is subtracted from the original score. As a result, the respondent’s score on a variable represents the difference with respect to the sample mean. This procedure enables a more straightforward interpretation of the random intercepts and random slopes models, since ‘the variances for the intercept and slopes [can be interpreted] as the expected value when all explanatory variables are equal to zero’ (Heck, Thomas & Tabata, 2010, p. 114). Consequently, the intercept represents the average score on Internet skills, when all other variables are at their mean values. Similarly, in the cross-level interaction models, the effect of education and occupation can be interpreted as the effect on Internet skills in the ‘average’ European country. In Table 1, the descriptive statistics are presented.
23
24
4. Results In this study, it is examined whether one’s social position determines his or her level of Internet skill. More specifically, it is hypothesized that both education and occupation positively influence Internet skills. Furthermore, it is investigated whether IT expenditure, broadband costs, and Internet in education affect this relationship. First, various figures are presented, which clarify how European countries differ with respect to a) their average level of Internet skill, b) the Internet policies of interest, and c) the relationship between education and Internet skills. Thereafter, the multi-level regression models are presented, in which the hypotheses are tested.
4.1. Illustrating European differences
Figure 3. Average Internet skills per country, N=22,956. Source: Eurobarometer 74.3 (2010).
In Figure 3, the average level of Internet skill per country is shown. The European average (EU) is 2.49, and it can be seen that substantial differences exist across Europe. Denmark (DK), the 25
Netherlands (NL), and Sweden (SE), with a score of 4.29, 4.18, and 3.80 respectively, are the leading countries. Portugal (PT) and Romania (RO), with a score of 1.05, and 1.23 respectively, have the lowest level of average Internet skill. To be specific, in Portugal and Romania, people perform one online activity on average. In Denmark, on the other hand, people perform four online activities on average. In general, it seems that the southern and eastern European countries score below average, while the northern countries have a rather diverse Internet usage pattern. It should be noted, however, that Estonia (EE), Great-Britain (GB), and Germany (DE) do not fit this overall pattern.
Figure 4. Average Internet skills per country (Internet users only), N=13,990. Source: Eurobarometer 74.3 (2010).
In figure 4, it is shown how Internet skills are distributed across European countries among the online population. To be specific, the figure shows the average score of Internet skills per country for people who have indicated that they use the Internet at home, at work, or at someplace else (N=13,990). It can be noticed that the average is considerably higher, and there seem to be less 26
differences between countries. This indicates that, especially in the less ‘skilled’ countries, there is still a large proportion of non-users. The correlation of 0.881 (p