The Roots of Tomorrow s Digital Divide: Documenting. Computer Use and Internet Access in China s Elementary

Transcription

1 The Roots of Tomorrow s Digital Divide: Documenting Computer Use and Internet Access in China s Elementary Schools Today YIHUA YANG, LINXIU ZHANG, XIAO HU, QINGHE QU, FANG LAI, YAOJIANG SHI, MATTHEW BOSWELL, SCOTT ROZELLE1 Working Paper 234 March 2012 reapchina.org/reap.stanford.edu 1 Yihua Yang, School of Philosophy and Society, Jilin University. Linxiu Zhang (corresponding author), Center for Chinese Agricultural Policy, Institute for Geographical Sciences and Natural Resource Research, Chinese Academy of Sciences. Xiao Hu, Qinghe Qu and Fang Lai, Center for Chinese Agricultural Policy, Institute for Geographical Sciences and Natural Resource Research, Chinese Academy of Sciences. Yaojiang Shi, School of Economic Management, Northwest University of Xi an. Matthew Boswell and Scott Rozelle, Rural Education Action Project, Freeman Spogli Institute, Stanford University. We would like to acknowledge Quanta Computing, Dell Computers, ADOC 2.0, Adobe Systems, Tianhua Shidai, Bowie Lee, the TAG Foundation, and Mary Ann Milias St. Peters for their generous support for REAP s Technology and Human Capital theme area. The hard work of dozens of volunteers from the Chinese Academy of Sciences, Northwest University of Xi an, Qinghai Minorities University have made this paper possible.

2 Abstract: The goal of this paper is to explore the nature of China s digital divide with a focus on differences in access to computers, learning software, and internet at school and home among different groups of elementary school-aged children in China. To meet our goal, we document the digital divide, focusing mainly on computer ownership, use and internet access. Second, we also seek to understand the quality of access to computers and their use. The digital divide is examined in four different dimensions: a.) between students in urban public schools and students in rural public schools; b.) between students in rural public schools and students in private migrant schools; c.) between migrant students in urban public schools and migrant students in private migrant schools; and d.) between students in Han-dominated rural areas and students in rural areas that are inhabited by ethnic minorities. Using data from a set of large scale surveys in schools in different parts of the country, we find the gap between computer and internet access of students in rural areas and urban public school students is extremely wide. The gap widens further when comparing urban students to students from minority areas. The gap is less wide when comparing computer access and access to teaching of the most basic computer skills across urban and rural public schools. However, the divide is still large between urban and rural schools when examining the quality of computer instruction and access to learning software. Migration itself does not appear to eliminate the digital divide. Only when migrant families are able to enroll their children into urban schools does the divide substantially narrow. If the digital divide in elementary schools today is a harbinger of employment, education, and income inequality tomorrow, China needs to seriously address this issue in the near future. 1

3 The Roots of Tomorrow s Digital Divide: Documenting Computer Use and Internet Access in China s Elementary Schools Today Over the past two decades, information and communication technology (ICT) has spread all over the world. The expansion of ICT has been characterized by the rapid development of hardware, software and affordable internet connections. 2 Increasingly powerful ICT devices have influenced the provision of many services, such as education and health, and have also been linked to the ability of individuals to find employment. 3 Because of this, ICT has become an important contributor to economic growth. 4 ICT is also associated with making individuals competitive and productive. 5 However, if ICT is available to only some groups of individuals in a society, but not the others (henceforth, a phenomenon that we call the digital divide), the resulting disparity in access to ICT would likely lead to income inequality in a country and poverty for those individuals without access. The source of the inequality would arise from several dimensions. It is well known that differences in access to health and educational services can lead to overall inequality in a country. 6 Differences in access 2 Timothy Bresnahan and Manuel Trajtenberg, General Purpose Technologies, Engines of Growth?, Journal of Econometrics 65(1), (January 1995), pp Cigdem Aricigil Cilan, Bilge Acar Bolat and Erman Coşkun, Analyzing Digital Divide Within and Between Member and Candidate Countries of European Union, Government Information Quarterly 26(1), (January 2009), pp Organization for Economic Co-operation and Development, OECD Rural Policy Reviews: China, (22 May 2009), available at: The World Bank, Information and Communications for Development Global Trends and Policies, (1 January 2006), available at: www-wds.worldbank.org/external/default/wdscontentserver/wdsp/ib/2006/04/20/ _ /Rendered/PDF/359240PAPER0In101OFFICIAL0USE0ONLY1.pdf. 5 María Rosalía Vicente and Ana Jesús López, Assessing the Regional Digital Divide Across the European Union-27, Telecommunications Policy 35(3), (April 2011), pp Paul Attewell and Juan Battle, Home Computers and School Performance, The Information Society: An International Journal 15(1), (January-March 1999), pp. 1 10; Mark Warschauer, A Teacher s Place in the Digital Divide. National Society for the Study of Education Annual Yearbook 106(2), (December 2007), pp

4 to employment are also closely related to inequality. 7 At the aggregate level it has been shown that there is a negative relationship between inequality and growth. 8 In this way it can be argued that the seriousness of the digital divide in a country has important implications for a county s growth path. There is a large literature that has examined the nature of the digital divide in both developed and developing countries around the world. For example, in America 80 percent of the homes of those earning over $75,000 have access to internet. In contrast, only 25 percent of America's poorest households can get online at home. 9 Similarly, according to Canadian Internet Use Survey, the digital divide still exists between the higher- and lower-income households in Canada. 10 Ninety one percent of people who earn more than $91,000 per year regularly access the internet. In contrast, this figure is only 47 percent for the people who make less than $24,000 per year. In India the rate of access to the internet of urban households in 2008 was ten times that of rural ones. 11 Like elsewhere in the world, government officials and researchers have been interested in documenting and measuring the digital divide in China. For example, it was reported in a study from the mid-2000s that the penetration of internet lines was more than three times higher in urban areas (27.4 percent of urban households) than in rural areas (7.2 percent). 12 Comparison of computer ownership between urban and 7 International Telecommunication Union, World Telecommunication/ICT Development Report: Measuring ICT for Social and Economic Development (March 2006), available at: 8 Roland Benabou, Heterogeneity, Stratification and Growth: Macroeconomic Implications of Community Structure and School Finance, American Economic Review 86(3), (June 1996a), pp Norris Dickard and Diana Schneider, The Digital Divide: Where We Are, (1 July 2002), Retrieved March 8, 2007, from: 10 Canadian Internet Use Survey, The Daily, (November 2008), available at: 11 Sumanjeet Singh, Digital Divide in India: Measurement, Determinants and Policy for Addressing the Challenges in Bridging the Digital Divide, International Journal of Innovation in the Digital Economy 1(2), (April-June 2010) 12 China National Bureau of Statistics, China National Statistical Yearbook, (Beijing, China: China State Statistical Press, 2007). 3

5 rural areas shows even greater inequalities. Research teams also found similarly large discrepancies in ICT access between rich and poor households. 13 While there are a number of official statistics and academic papers on the digital divide in China, judging their accuracy is nearly always difficult. Reports rarely give details on the coverage of surveys (e.g., the report by Chinese Information Center). Sampling frames are almost never specified. 14 The sources of information of the reports/papers are frequently absent (e.g., the paper by Wang, 2001). In most cases information on ICT is not analyzed (for example, the correlates of access to ICT are rarely, if ever, discussed) and many of the statistics in the papers are reported in aggregate form only. In one of the best written (in our opinion) and most carefully documented papers in the published literature, the entire sample contains only five villages. Given this absence of high quality, verifiable and detailed information on China s digital divide, the overall goal of this paper is to document the gaps in access to ICT in China. To meet this goal, we have two specific objectives. First, we document the digital divide, focusing mainly on computer ownership, use and internet access. Second, we also seek to understand the quality of access to computers and their use. We will look at the digital divide in four different dimensions: a.) between students in urban public schools and students in rural public schools (the urban-rural digital divide); b.) between students in rural public schools and students in private migrant schools (the rural-migrant digital divide); c.) between migrant students in urban public schools and migrant students in private migrant schools (the urban-migrant digital divide); and d.) between students in Han-dominated rural areas and students in rural areas that are inhabited by ethnic minorities (the Han-ethnic minority digital divide). 13 Wang Wensheng, Bridging the Digital Divide Inside China, (2001), Working Paper. 14 For example, Xia Jun, Linking ICTs to Rural Development: China's Rural Information Policy. Government Information Quarterly 27(2), (March 2010), pp

6 The ambitiousness of this paper almost necessarily means that there are limitations to our work. First, our empirical evidence on the digital divide is only for students in elementary schools. While this admittedly is only one segment of China s population, it is well known that access to ICT during childhood is a strong predictor of expertise in ICT in adulthood. 15 In this sense it is almost as if we are also looking at the future of China s digital divide (and perhaps income inequality). Second, while our samples in China s urban and rural public schools and migrant private schools are relatively large, the samples are not representative of China as a whole. This means that we must be careful in claiming external validity of our results. Finally, our results are almost fully descriptive in nature. Nowhere in the paper do we attempt to offer definitive determinants analysis regarding the digital divide or offer evidence on the impact on other outcomes (for example, educational attainment or educational performance). To achieve our objectives, the rest of the paper is organized as follows. In the next section we describe the data. In the subsequent section we examine the results by looking at the nature of the four types of digital divides: urban-rural; rural-migrant; urban-migrant; Han-minority. The final section contains a discussion of the results and concludes. Data One of the most important tools required when analyzing the digital divide is high quality data. These data are needed on a regional basis for key regions of the country, that is: urban; rural; migrant; and ethnic minority areas. However, despite the 15 Helene P. Baouendi and Judith D. Wilson, Computer Education in the Primary and Middle Schools in the People s Republic of China: the Late1980 s, Education and Computing 5(4), (1989), pp

7 efforts devoted to this issue by scholars (e.g., Wang, 2001), there is still an absence of school-based ICT data that is disaggregated on a region by region basis. In this paper we have access to data on school-aged students in elementary schools in different regions of China. Specifically, in 2009 and 2010 our research group conducted surveys of students in four sets of elementary schools: urban public schools (which included both students with urban household registration, or hukou henceforth urban students; and students whose parents were migrants, but, who were going to urban public schools henceforth, migrant students in urban public schools); rural public schools (rural students); migrant private schools (migrant students in private migrant schools); and rural public schools in ethnic minority areas (ethnic minority students). The data used in the study were collected during four separate enumeration efforts. We will describe the data sets in the rest of this section. Urban Public School Data (Urban Students and Migrant Students in Public Schools) The Urban Public School data set collected data on two types of students. In the nine schools in the sample, around ten percent of the students had Beijing hukou. 16 This means that at least one of their parents was a Beijing resident (had a Beijing hukou). In the rest of the paper we call these students urban students. The other ninety percent of the students did not have Beijing hukou. Their parents also did not have Beijing hukou and were typically working in Beijing as migrant workers. In the rest of the paper we call these students migrant students in urban public schools. 16 China s hukou household registration system classifies China s citizens as either rural or urban residents. Without an urban hukou migrants and their families have limited access to urban public services, including housing, healthcare, social security, and education. Since the mid-2000s, if there is room in urban public elementary schools, migrant students are allowed to matriculate and are treated as urban students in terms of tuition, fees and most other services while in school. See Barry J. Naughton, The Chinese Economy: Transitions and Growth, (Cambridge, MA: MIT Press, 2007). 6

8 The first step in collecting the Urban Public School data set was to select the schools to include in the sample. To do so, we randomly selected nine public schools in the suburban areas of Changping, Haidian and Chaoyang Districts of Beijing. These schools were chosen purposely in the outskirts of Beijing suburbs so they would contain a mix of urban students and migrant students in urban public schools. While the schools themselves were chosen randomly from suburban schools, they are not representative of Beijing, in general. According to interviews with officials in the Chaoyang and Haidian bureaus of education, compared to other public schools across Beijing, the public schools represented in the Urban Public School dataset can best be classified as schools that are in the lower-middle tier of schools in Beijing. These are schools that perform poorer and have students whose parents are less well off financially than a typical public school in Beijing. The schools are clearly below average in Beijing. Hence, this needs to be kept in mind when interpreting the results. We are really looking at relatively poor urban students, which means that we will likely be looking at a lower bound of the digital divide (that is, the divide would be even wider if we chose urban students from the best schools in Beijing). However, as urban public schools in one of China s richest province-level administrative districts, all the schools in this dataset receive public funds and thus cannot be considered to be under-resourced. In a report written using this data set and comparing these schools to a set of randomly selected private migrant schools, the facilities, teachers and curriculum for the urban schools were of much higher quality than those of the private migrant schools. 17 Because urban public schools are free and of higher quality, migrant parents generally only send their children to private migrant schools when there is no room in local urban public schools. Although there is no 17 Lai Fang, Luo Renfu, Zhang Linxiu, Huang Xinzhe and Scott Rozelle, Does Computer-Assisted Learning Improve Learning Outcomes? Evidence from a Randomized Experiment in Migrant Schools in Shaanxi, (2012), REAP Working Paper. 7

9 formal sorting rule by which migrant children get into urban schools and which do not, it appears as there may be a systematic difference in the type of migrant family that is successful in getting their child into migrant school. In other words, migrant families that can successfully get their child into urban schools appear to be different from those that enroll their children into private migrant schools. Hence, we should consider the difference between these two types of families when interpreting the results of our study. According to Lai et al., 18 migrant students in urban schools score significantly higher on standardized exams than migrant students in private migrant schools. The parents of migrant students in urban schools also have higher levels of education and have been in Beijing for a longer period (that is, they have more work experience) than parents of migrant students in private migrant schools. Because of these systematic differences we must be careful when comparing the digital divide between these two types of families. The type of school may be one factor, but, inherent differences in the characteristics of the families might also play a role. The second step in the data collection process was to visit the urban schools and undertake a survey. Our survey focused on the 1458 fourth grade students in the nine urban schools in our sample. Among all fourth grade students in the urban schools, 154 of them were urban students and 1304 of them were migrant students. The students involved in our survey were given a questionnaire that included the following questions: whether students use computers at school; how many times students use computers at school every week; the nature (quality) of the computer classes (computer skills they learn in computer class; and whether the classes were often cancelled); whether students use computers at home for study; whether the student s family owns a computer; whether the student has access to internet at home; etc. All 18 Lai Fang, Liu Chengfang, Luo Renfu, Zhang Linxiu, Ma Xiaochen, Bai Yujie, Brian Sharbono and Scott Rozelle, Private Migrants Schools or Rural/Urban Public Schools: Where Should China Educate Its Migrant Children?, (2011), Reap Working Paper. 8

10 of these questions were related to computer access and use, the quality of computer education and access to the internet. Rural Public School Data (Rural Students) The data on rural public schools come from Shaanxi Province. In Shaanxi, around 63% of the population lives in rural areas. 19 The incidence of rural poverty in Shaanxi in 2005 was 2.9 times higher than the national average. 20 Shaanxi also has had one of the slowest rates of poverty reduction in rural China since In using a sample of rural students from poor rural public schools we need to realize that the urban-rural digital divide will be overstated when compared to a completely random sample of rural students (because access to ICT technologies is almost certainly better in better-off rural areas). The site of the rural public school survey was chosen to be in Ankang Prefecture. The prefecture is located immediately south of Xi an, the capital of Shaanxi province. All of the counties are located in the Qingling Mountain Chain. Ankang prefecture is home to one of the poorest groups of counties in China. Of the seven counties in Ankang prefecture, there are four counties that are nationally-designated poor counties. 22 We chose to do the study in these four counties. 19 China National Bureau of Statistics, China National Statistical Yearbook, (Beijing, China: China State Statistical Press, 2005). 20 Organization for Economic Co-operation and Development, OECD Rural Policy Reviews: China, (22 May 2009), available at: 21 Ravallion Martin and Chen Shaohua, China s (Uneven) Progress Against Poverty, Journal of Development Economics 82(1), (January 2007), pp In 1994, the Chinese government launched a poverty-reduction initiative under the "8-7 Plan" with the major objective of raising per-capita income to 500 yuan (in 1990 prices) within seven years. This plan targeted 592 designated poverty counties in the country. Poverty reduction, especially in China's west, remains a formidable challenge today. When we say poor county in this paper, we mean that the county is designated by China s government as being a poor county in the poverty reduction program. 9

11 The survey team next chose sample schools and the students in the schools. To do this, we first obtained a list of all schools in each county. We then narrowed this list to include only elementary schools that contained six full grade levels (wanxiao). From the full list of wanxiao in the four sample counties we randomly selected 72 schools. The 72 schools contained 2666 third-grade and fifth-grade students. The rural public school ICT survey was launched in February All the third and fifth-grade students in each of the sample schools were included in the survey. The questionnaire was nearly identical to the urban public school survey. The questions on access to computers and internet were the same. Because of limitations on the length of the survey that we could give during class time, there were fewer questions on how computers were used in class. Private Migrant School Data (Migrant Students in Private Migrant Schools) The sample of private migrant schools in Beijing was chosen from a sampling frame that covered almost all private migrant schools in the city. Creating the list was a challenge. Unlike public schools, no official list of Beijing migrant schools is available. To collect a comprehensive list of migrant schools in Beijing we contacted all educational and research institutes and non-profit organizations in the greater Beijing area that might have contact information for Beijing migrant schools. We then called each school to confirm that the school was still open. During each phone call we also asked the principal of each school if there were any other schools in their area. By proceeding in this fashion we believe that we were able to establish as complete a database of Beijing migrant schools as possible (certainly more complete than any other existing database). A total of 230 elementary schools were on our list. 10

12 We selected our sample schools using the comprehensive list that we assembled. To focus our study on districts where migrant schools are most concentrated (which made implementation somewhat easier by reducing inter-school transportation time), we restricted our sample to the three districts in Beijing. These districts are among the areas of Beijing most densely populated by migrants and migrant schools. Of the 230 schools in the database, 69 schools were in these three districts. We then proceeded to exclude schools that had only one class in the third grade (that is, there was only one grade 3 class instead of two or more grade 3 classes in the school). We did this as part of our strategy for another study. 23 By doing so, our sample private migrant schools are representative of all large private migrant schools in Beijing. In total, 43 schools met the criteria of having two or more grade 3 classes. A total of 4,103 students in 98 classes of 43 Beijing migrant schools were surveyed. As in the urban and rural schools, all third-grade students in the sample schools were provided with questionnaires. The content of the surveys in the private migrant schools were identical to that of the rural schools. Rural Minority Public School Data (Ethnic Minority Students) The data on rural minority public schools come from Qinghai Province. Qinghai is a province in northwest China whose population has a high percentage of ethnic minorities relative to most other provinces in China. The minority population of Qinghai province accounts for 47 percent of the total population. 24 Tibetans account for 24 percent; Hui account for 15 percent; Tu account for four percent and other 23 The study was on the impact of Computer Assisted Learning on educational outcomes see Lai Fang, Luo Renfu, Zhang Linxiu, Huang Xinzhe and Scott Rozelle, Does Computer-Assisted Learning Improve Learning Outcomes? Evidence from a Randomized Experiment in Migrant Schools in Shaanxi, (2012), REAP Working Paper. 24 China National Bureau of Statistics, China National Statistical Yearbook, (Beijing, China: China State Statistical Press, 2010). 11

13 ethnic groups, including Salar, accounts for another four percent. Qinghai Province also is the second poorest province in China. Because most minority communities are rural and located in remote mountainous regions and have lower levels of education, the sample area is home to some of the poorest people in China. 25 The first step that we took to obtain the dataset on minority public schools was to select the prefecture and counties for our sample of rural, public minority schools. Our survey in Qinghai province was implemented in October 2011 in three counties in Haidong Prefecture. This prefecture has six counties within its jurisdiction. We chose three counties in Haidong prefecture, Xunhua, Hualong and Huzhu counties as our sample counties as these three counties are ethnic minority autonomous counties primarily populated by four ethnic minorities (Tibetans, Hui, Tu and Salar). The sample of schools and students were drawn from a comprehensive list of all schools in the three counties. Because the project that was subsequently run after our survey focused on using computer assisted learning software to provide third and fourth-grade students remedial tutoring in Chinese language and math, most Tibetan communities were not included in the survey (the level of Chinese language skills among Tibetan third and fourth graders was too low to make appropriate use of the computer assisted learning program). From a subset of 52 rural minority schools (in Hui; Tu and Salar communities) in the three counties in our sample, we randomly selected 26 schools to be included in our final sample. In total, there were 2587 third and fourth grade students in our sample rural minority public schools. As in the rural and private migrant schools, we included a block in our survey examining access to ICT. The survey questionnaires in Qinghai were identical to those in Shaanxi and migrant communities in Beijing. 25 Zhang, M and S. Li Discussion on Floating Population of Ethnic Minorities in the Process of Urbanization of Qinghai Province. Theory and Perspective. 12

14 Four School Types/Five Student Types Table 1 summarizes the datasets that are used in this study. The table includes summary information on school type, the types of students in each type of school, the location of the sample and the number of sample schools and number of sample students in each of the datasets. In total, we surveyed third, fourth and fifthgrade students in 150 elementary schools in Beijing, Shaanxi and Qinghai. Results In this section we present descriptive evidence on four different digital divides. The first subsection examines the urban-rural digital divide. The second subsection presents information on the rural-migrant digital divide. The third subsection describes the public migrant-private migrant digital divide. The fourth subsection looks at the Han-ethnic minority digital divide. Urban-Rural Digital Divide When comparing a number of variables between urban students and rural students especially those variables associated with in the school rooms of the students, our data perhaps surprisingly show that the digital divide, while present, is not extreme (Table 2, rows 1 to 4). In urban public schools 88 percent of students say that they use computers at school (column 1). Conditional on using computers at school, all students (100 percent) used the computers at least once per week and had 40 minutes of computer class time or more. Students reported that computer classes were never cancelled or replaced by non-computer classes. In other words, computer class time in urban public schools is quite regularized. 13

15 Use of computers by rural students in rural public schools is also fairly regularized although at levels somewhat lower than that in urban public schools (Table 2, rows 1 to 4, column 2). 69 percent of rural students in our sample used computers at school. Of those that used computers at school, between 72 and 78 percent had computer class at least once each week and computer classes were at least 40 minutes long. Computer classes in rural public schools were almost never cancelled (only 2 percent of the students reported so). Given the fairly regularized use of computers in both urban and rural schools, the urban-rural digital divide (in the school rooms) is fairly modest (Table 2, rows 1 to 4, columns 3 and 4). When comparing the use of computers at school the digital divide is only 19 percentage points, or a ratio of 1.3:1. When comparing the regularity and frequency of computer classes, the gap is only 22 to 28 percentage points (ratios from 1.3:1 to 1.4:1). The urban-rural digital divide is still fairly narrow (though widening a bit) when we examine skills learned in computer class (Table 2, rows 5 to 9). All of urban students (100 percent) learn the basics of computer operations how to turn the computer on and off; how to use the keyboard, etc. (column 1). Though not all of the rural students, most of thems (67 to 84 percent) have learned basic computer operation skills such as how to use a computer (column 2). The gaps are between 16 and 33 percentage points (or ratios of 1.2:1 to 1.5:1). The urban-rural digital divide at school becomes wider when examining more complicated skills of computer operations such as using educational software and learning about computer hardware (Table 2, rows 10 and 11). When we asked students whether they used any type of educational software or computer-assisted learning software in class, 90 percent of urban students said they did (column 1). In contrast, 14

16 only 36 percent of rural students had ever used a computer with software that was designed to enhance learning (column 2). There was a 54 percentage point gap (or ratio of 2.5:1) between urban and rural students (columns 3 and 4). Likewise, there was a substantial gap (51 percentage points or 2.3:1 ratio) in students that learned about computer hardware (e.g., the basic components of a computer and their purposes row 11). Hence, when examining the nominal use of computer (as in rows 1 to 4) and when examining the very basic fundamentals of computer use (as in rows 5 to 9) the gap between urban and rural does not appear to be large. However, when looking at patterns of computer use that complement or enhance learning of core school subject matter and the rudiments of computer hardware there is clearly a urban-rural divide among the students in our sample schools. The urban-rural digital divide is even wider when we examine the student use of computer and internet at home (Table 2, rows 14 and 15). In the case of urban students, 80 percent lived in homes that owned some type of computer. In 73 percent of homes, urban students were able to have access to the internet. In contrast, only ten percent of the households of rural students owned some type of computer (row 14). Only five percent of students could access to the internet at home (rows 15). The gaps between the urban and rural students in terms of access to ICT at home varied from 70 to 68 percentage points (column 3). The ratios of urban to rural students in the access to ICT services at home ranged from 8:1 to 14.6:1. Indeed, according to our analysis (and assuming our sample is somewhat representative of the situation across China), there is clearly a wide urban-rural digital divide in China. Some of these gaps also almost certainly make China s digital divide one of the largest in the world. In summary, in this section we have identified that there is a significant digital divide between urban students and rural students. The divide, in fact, does not appear 15

17 to be wide when looking at nominal computer use at school (or the very basic uses of the computer). However, when examining patterns of computer use that complement or enhance learning of core school subject matter as well as learning about computer hardware and when examining the use of computers/internet in homes of students, there is quite a substantial gap between urban and rural students. If the digital divide among school-aged children is a predictor of educational performance and employment opportunities later in one s life, the current magnitudes of China s urban-rural digital divide imply that at least part of China s alarming urban-rural income gap may be expected to persist. Rural-Migrant Digital Divide China is in the middle of one of the biggest episodes of rural to urban migration that the world has ever witnessed. 26 This transfer of population from rural to urban areas is likely to continue in coming decades. As the rural to urban migration continues, is it possible that the disparities described in the section above may correct themselves as more and more people from rural areas settle down in urban areas? When looking at the rural-migrant divide, the answer seems to be no, at least not completely. In fact, our data suggest that insofar as the digital divide is concerned, rural students appear to be better off in some dimensions when compared to migrant students in private migrant schools (that is, one type of migrant student in our surveys, the other being migrant students in urban public schools). This is in contrast to the urban rural divide in which urban students were substantially better off than their rural counterparts. We examine this issue in this section. In the next section we examine the case of migrant students that are in public schools. 26 Scott Rozelle, Li Guo, Shen Minggao, Amelia Hughart and John Giles, Leaving China's Farms: Survey Results of New Paths and Remaining Hurdles to Rural Migration, China Quarterly 158, (June 1999), pp

18 When examining the rural-migrant digital divide, we see that rural students in rural public schools have systematically better access to computers in school than migrant students in private migrant schools (Table 3, rows 1 to 4). According to our data, 69 percent of rural students use computers in school while only 15 percent of migrant students in private migrant schools do (row 1, columns 1 and 2). The gap is 54 percentage points or ratio of 4.6:1. The same is true for the regularity of the use of computers in school and the amount of time spent in computer classes (rows 2 and 3). Conditional on having and using computers in their schools, rural students have computer class more often (72 percent versus 34 percent a gap of 38 percentage points or ratio of 2.1:1) and for longer periods of time per week (78 percent versus 54 percent a gap of 24 percentage points or ratio of 1.4:1) than when compared to migrant students in private migrant schools. Computer classes are also cancelled more frequently (in only two percent of the schools for rural students versus in 20 percent of the schools for migrant students in private migrant schools). In short, even in terms of nominal use of computers, the rural-migrant digital divide is wide. Interestingly, there is not much difference in what is learned in ICT class between rural schools and private migrant schools (Table 3), possibly because students learned very little from ICT classes in both types of schools. In terms of the basics of using computers, both rural students and migrant students in private migrant schools are almost equal (rows 5 to 9). The same is true for patterns of computer use that complement or enhance learning of core school subject matter regarding both software and hardware. In fact, migrant students have a slight, though not significant, advantage over rural students (rows 10 and 11). However, it is important to note that these statistics are conditional on the students actually having a computer class. If one were to look at the unconditional statistics (that is not conditional on having computer 17

19 classes) of what migrant students in private migrant schools learned relative to those students in rural public schools, it is clear that rural students learn much more than migrant students. The evidence is mixed when examining the rural-migrant digital divide in and around the homes of rural and migrant students (Table 3, rows 14 and 15). In the case of migrant students, 42 percent of them own some type of computer at home. In contrast, only 10 percent of rural students own computers at home. In the case of access to internet at home, only five percent rural students in rural public schools can access the internet at home, whereas 37 percent of migrant students in private migrant schools can (rows 14, 15). The gaps are 32 percentage points or a ratio of 1:4 and 32 percentage points or ratio of 1:7 in favor of migrant students. Clearly being in Beijing even as a migrant for some reason has stimulated households to buy computers and install internet access. We do not have information to precisely explain the observations above. It may be a matter of availability (internet lines are more convenient to install in metropolitan areas such as Beijing). It may be a matter of norms (so many families around us have a computer, we need one too). It many be that migrants have higher incomes than households in rural areas. Or, it may be some combination of the above factors with other reasons. In summary, in this section we show the digital divide between rural students and migrant students in private migrant schools is modest. However, in some key dimensions the divide favors rural students. In other words, when rural students migrate with their parents to the urban areas and enter private migrant schools, the digital divide does not automatically narrow down in all dimensions. So is there any way to narrow down the digital divide? Another option for the migrant students (at least for those fortunate to live in a school district with an urban public school that has 18

20 room for the children of migrants) is to enter urban public schools. Is it possible that when they do so that the digital divide between urban students and rural students in China will begin to narrow? This question is explored in the next section. Public Migrant Private Migrant Digital Divide While it is clear from the data and analysis in the previous section that the mere act of moving to the city does not eliminate the urban-rural digital divide, our data also allow us to examine the access to ICT for different types of migrants. In fact, when examining Table 4 (columns 1 and 2), as a first step we can first compare access to ICT for urban students and migrant students in urban public schools (as opposed to migrant students in private migrant schools). This comparison will allow us to first understand the access to ICT for migrant students in urban public schools (rows 1 to 13). When doing so, we find that urban students and migrant students in urban public schools have remarkably equal access to ICT. Of course, since they are in the same school (and mostly in the same classes in those schools), the digital divide inside schools (almost by definition) is zero. For urban students and migrant students in urban public schools, the digital divide at home is not wide (80-70=10 percentage points for computer ownership and 73-58=15 percentage points for internet access rows 14 and 15), either. These data suggest that one of the ways to narrow down the digital divide in China is getting students into urban public schools. We then proceed to examine the digital divide between migrant students in urban public school and migrant students in private migrant schools. We call this the urban public migrant-private migrant divide. When we examine the variables related to in the school rooms of the students, the digital divide between the migrant students in urban public schools and those in private migrant schools appears to be wide (Table 4, 19

21 rows 1 to 4). In urban public schools, 90 percent migrant students say that they use computers at school. Conditional on using computers at school, all students (100 percent) use the computers at least once per week and had 40 minutes of computer class time (column 1). At the same time, according to the student responses to the relevant survey questions, the computer classes were almost never (one percent) replaced by other classes. That means computer class time in urban public school for migrant students is quite regularized. However, regularized use of computers by migrant students in private migrant schools is much lower (column 2). Only 15 percent of migrant students in private migrant schools used computers at schools, 34 percent and 54 percent had computer class at least once each week and these computer classes were at least 40 minutes long. Among all migrant students in migrant schools, 20 percent of them reported that the scheduled computer classes were frequently replaced by other classes. Hence, the digital divide at school between migrant students in urban public schools and migrant students in private schools is fairly wide (rows 1 to 4, columns 3 and 4). The public migrant-private migrant digital divide is also wide when we examine variables associated with anything more than nominal use of computers inside schools. Although given use of computers in schools (that is, comparing all urban public schools and the small share of private migrant schools with computers) most students learn how to perform rudimentary functions on the computer (Table 4, rows 5 to 9), when comparing the use of educational software and instruction on understanding computer hardware, there is a wide gap (rows 10 and 11). Among migrant students in urban public schools, 90 percent of them used software that complemented or enhanced learning of core school subject matter and also focused on teaching about computer hardware. In contrast, less than half (49 percent) of migrant students in 20

22 private migrant schools either used educational software or learned about computer hardware in school. Finally, the public migrant-private migrant digital divide is also fairly wide in terms of computer use and internet access at home (Table 4, rows 14 and 15). According to our data, there are two substantial gaps. Migrant students in urban public schools have higher computer ownership (70 percent) than migrant students in urban private schools (42 percent). Migrant students in urban public schools also have more common internet access (58 percent) compared to migrant students in urban private schools (37 percent). The gaps for this dimension of the digital divide range from 1.6 to 1.7 times. In summary, in this section we have identified the public migrant-private migrant digital divide is significant. The digital divide between migrant students in urban public schools and those in private migrant schools appears to be widest when we examine the computer use at school. Hence, schools clearly play an important role in the digital divide between China s rural and urban areas and among different types of students. Hence, if China is serious about narrowing the digital divide, it may want to carefully redesign the ways that computers are used in all schools, including rural schools and private migrant schools. Alternatively, if China s educators committed themselves to getting every child that is in the city into urban schools, it would go a long way to narrowing China s digital divide. Han Ethnic Minority Digital Divide On account of many possible factors language, culture, geographical isolation, etc. it may be that even though rural minority students are all in rural public schools they have differential access to ICT than the typical Han students in rural public 21

23 schools. In this section we examine the differences in access to ICT between students in rural public schools (in the Han-dominated province of Shaanxi) and rural minority students in rural minority public schools (in Qinghai Province). If there is a digital divide, we call this the Han-Ethnic Minority digital divide. Of course, if there is a Han-Ethnic Minority digital divide then the urban-minority divide shall be even larger (since the urban-rural digital divide, as seen above, is already large). When examining the Han-Ethnic Minority digital divide, we find that students in rural public schools have better access to computers at school than minority students in rural minority public schools (Table 5, row 1). When we asked the students in rural public schools whether they use computers at school, 69 percent of the students said that they did. In contrast, only 16 percent of minority students in rural minority public schools claimed they did. This suggests, of course, that there is a great gap (53 percentage points or a ratio of 4.3:1) between rural Han students and minority students insofar as their use of computers in school is concerned. Furthermore, conditional on using computers at school (row 2-4, column 3-4), rural Han students have computer class more often (72 percent versus 16 percent--a gap of 56 percentage points) for longer periods of time per day in the class room (78 percent reported having computer classes last for 40 minutes or longer in Han schools versus 16 percent in the rural minority schools a gap of 62 percentage points) than rural minority students. Computer classes are also cancelled more frequently (only two percent for rural students versus 16 percent for rural minority students reported frequent cancellation of computer classes). Therefore, in terms of computer use and computer class at school, the Han-Ethnic Minority digital divide is significant. When looking at the skills and knowledge learned in the ICT class (conditional on having regular computer class), we find that the gap between students in rural 22

24 public schools and students in rural minority public schools also appears to be wide (Table 5, rows 5 to 13). In fact, in almost all categories of computer use inside schools (that have computers), the patterns are similar in rural public schools and rural minority schools. Like their counterparts in Han schools, students in rural minority schools are not able to use learning software. Clearly, there is an urgent need for policy makers to improve the use of computers (once they are in schools) in all rural public schools, those in Han areas and even more so for those in rural minority areas. The Han-Ethnic Minority digital divide also exists in the home of the students. More Han students in rural public schools (10 percent) have computers at home than do minority students in rural public schools (only six percent rows 14 and 15). Han rural students also have more access to internet (five percent) than minority rural students do (only two percent). More importantly, whereas we can still claim that there is a digital divide between Han rural students and minority rural students in their homes (the gap ranges from 1.7:1 to 2.5:1), given that computer ownership and internet access in the home of Han rural public school students is also extremely low, the absolute digital divide is not wide. In summary, on the whole we can conclude from this section that the Han-Ethnic Minority digital divide is not the widest (e.g., compared to the Urban-Rural digital divide). But, this is not good news. It is true that rural students in Han areas are in schools that have much more access to computers. However, once computers are installed in either rural schools or rural minority schools, there is little use of educational software or instruction about computer hardware. The access to ICT at home for both rural students in Han areas and rural minority students is extremely low. Access to computers and internet at home is almost non-existent. 23

25 Conclusions In order to understand the digital divide among different groups of elementary school students in China (which, we believe, may predict tomorrow s overall digital divide), we empirically examined three gaps: the difference in access to ICT in the school rooms of the students, what is learned in ICT class, and in the homes of the students. Using our own data on more than 10,000 students in Beijing, Shaanxi and Qinghai, we measured four dimensions of China s digital divide: the urban-rural digital divide, the rural-migrant digital divide, the public migrant-private migrant digital divide and the Han-ethnic minority digital divide. The results show the digital divide among students in different regions are generally quite wide. The order of access to ICT is for the most part clear. Urban students in urban schools have quality access to ICT in both schools and at home. Interestingly, migrant students in urban public schools are not far behind. There is ambiguity about who is next. In terms of access to computers and quality software, rural students in rural public schools are better than migrant students in private migrant schools. Migrant students, however, have more access to computers and internet than rural students at home. The worst in all categories are minority students in rural public schools. Given the order urban student migrant student in urban public schools migrant student in private migrant school rural student minority student, the most severe gaps (perhaps as severe as any gap in the world) are between urban students and minority students. Indeed, the 80 percent to six percent (= 74 percentage point difference) gap in computer ownership and the 73 percent to two percent gap in internet access indicate that the ratios for access to ICT at home range from 13:1 to 36:1. 24