Internet Usage and Children's Skills Development

• In Event: Digital technologies in education: student use, skill development and instructional contexts

Proposal

1. Introduction
The effects of the digital technologies on our lives are universal. Children in the digital era may be more influenced by digital technologies than we realize. This study aims to investigate the effects of digital technology on children’s cognitive and non-cognitive abilities from an economic perspective. Moreover, the role of parenting styles in this process would be explored.
Research findings on the effects of digital technology vary to a very great extent on the development of skills among children. Some researchers show that using digital technology can enhance the cognitive abilities of a child, while others argue that it partially supports cognitive development but deteriorates academic performance in children(Johnson, 2010; Giacomo et al., 2017). Bulman and Fairlie (2016), in a review of the literature, found that older studies had a tendency to report positive effects of digital technology on children's outcomes in development and education, with the minority tending to indicate negative impacts. However, recent studies tend to show the effects to be statistically insignificant or small.
Existing studies have reached different conclusions, and the role of parenting styles in children's Internet use has been insufficiently explored. This article further explored the role of the Internet on the skill development of cognitive children from an economic perspective.
2. Empirical Approach
2.1 Empirical Model
The linear value-added model (VAM) was widely used as a kind of education production function in literatures. Todd and Wolpin (2003) derived VAM from cumulative achievement function, Eq.(1) shows the basic cumulative achievement function.
A_it=A_t [X_i (t),F_i (t),S_i (t),α_i0,ε_it] (1)
In Eq.(1), A_it is achievement for the child i at time t. X_i (t),F_i (t) and S_i (t) are the inputs from individual, family and school for child i though year t. α_i0 is the endowment of child i. ε_it is error term.
Regarding the items of Eq.(1) as additively separable, and the parameters are considered age-invariant, based on the derivation pattern of Todd and Wolpin (2003), we obtained a “lagged-score” VAM in Eq.(2).
〖Ability〗_ifst=β_0+β_1 〖Internet〗_ifst+β_2 〖Ability〗_(ifst-1)+β_3 X_ifst+β_4 F_ifst+β_5 S_ifst+ε_ifst (2)
The 〖Ability〗_ifst is the cognitive or non-cognitive ability of child i, who is in family f, attending school s, in year t. 〖Internet〗_ifst is the variable of internet usage (time). X_ifst, F_ifst and S_ifst are vectors of individual, family and school indicator variables. ε_ifst is error term.
〖Ability〗_ifst=β_0+β_1 〖Parenting〗_ifst*〖Internet〗_ifst+β_2 〖Parenting〗_ifst+
〖β_3 〖Internet〗_ifst+β_4 Ability〗_(ifst-1)+β_5 X_ifst+β_6 F_ifst+β_7 S_ifst+ε_ifst (3)

Eq.(3) is used to find the role of parenting style. 〖Parenting〗_ifst is a dummy variable of the parenting style type of the parents of child i.
2.2 Data Source and Variables Setting
2.2.1 Data Source
The individual and domestic microdata proposed to be used in this paper comes from China Family Panel Studies (CFPS) 2010, 2014, 2016 and 2018. As a large micro tracking survey data, the systematic probability sampling method adopted by CFPS ensures its nationwide representativeness. Additionally, CFPS includes the data from levels of region, family and individual, which can provide sufficient data support for this paper.
2.2.2 Variables
(1) Dependent variables
Cognitive ability . It was measured by math test and literacy test in CFPS. We standardized the values of the two tests using the Min-max normalization method and mapped the value range to the interval [0, 10].
Non-cognitive ability. It was measured according to the following five questions from the CFPS interviewer observation: the degree of cooperation of the respondent to the survey; the respondent's level of hospitality; the respondent's interest in the survey; the credibility of the respondent's answer; and the respondent's language expression ability. We conducted the factor analysis for these 5 questions and standardized the value into the interval [0, 10].
(2) Independent variable
1. Whether children access the Internet. 2. Hours of Internet use per day.
(3) Control variables and mechanical variable
Based on the literature and available data, this thesis proposes to control for variables at the individual, household, and school levels. Individual level: age, gender, ethnicity, hukou, urban, educational stage, whether the child is raised by his or her grandparents. Household level: the cognitive and non-cognitive abilities of parents, parents' years of education, household income per capita. School level: whether the child is in a key class, whether the child is in a key school or private school.
Parenting style. Based on two dimensions, responsiveness and demandingness, parenting style was divided into four types: authoritative, authoritarian, permissive, and neglectful. Six questions were used to measure parenting style. They are about parents' attitudes towards their children's behavior and school affairs. The method of factor analysis and standardization were used to generate parenting style variables . In this paper, parenting style is dummy variable, if the parenting style belongs to this type, the value is 1, otherwise it is 0.
2.3 Results
According to the estimation results of VAM, we found the positive relationships between internet usage and math and word test scores, and negative relationships between time spent online and math and word test scores. Moreover, to find the nonlinear relationships, we added the square term of internet usage time, and found the potential “U” shape relationship between internet usage time and math test. For non-cognitive ability, we found a positive correlation between Internet use and non-cognitive ability scores. However, we didn’t find some significant effects of internet usage time.
When parenting style is considered, we found that the authoritarian parents could promote the positive effects of internet usage, and using internet could relieve the negative effects brought by neglectful parents. Authoritative parents aggravate the negative impact of internet usage time on cognitive ability.
There is a positive effect of internet usage time on non-cognitive ability for the children who have authoritarian parents. However, it is negative for the children who have permissive parents, the internet usage time weakened the positive effect of permissive parents on non-cognitive ability.
3. Conclusion
This study reveals the relationship between internet usage and children's cognitive and non-cognitive abilities. Internet use generally improves abilities, but excessive time online can harm cognitive outcomes. In addition, parenting style plays a significant role in these relationships.

Author

• Haojian Dui, Renmin University of China, Harvard University