Journal of Men's Health,2023,19(12):11-22 DOI:10.22514/jomh.2023.125
Original Research

The effect of Internet use on males’ body mass index and overweight: evidence from China

Weigang Liu1, Jian Liu2, Jens-Peter Loy1, Yanjun Ren2,3,*,

1Department of Agricultural Economics, Kiel University, Kiel, SH 24118, Germany

2Department of Agricultural Markets, Leibniz Institute of Agricultural Development in Transition Economies, Halle, ST 06120, Germany

3College of Economics and Management, Northwest A&F University, 712100 Xianyang, Shaanxi, China

*Corresponding Author(s):yanjun.ren@nwafu.edu.cn (Yanjun Ren)

History Submitted: 25 October 2022 | Accepted: 03 February 2023 | Published: 30 December 2023
Copyright:  ©2023  The Author(s). Published by MRE Press.
This is an open access article under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/).

Collapse table of contents

Abstract

The effects of increasing Internet use in developing economies have attracted various attention from politicians and academic researchers, while the literature investigating the effects of Internet use on nutrition and health is insufficient. In particular, research on China, with its large overweight population, is still not clear and complete. This study aims to evaluate the effect of Internet use on body mass index (BMI) and overweight for males in China and shed light on the underlying mechanisms of the effect. To address the non-random distribution problem of Internet use and control for unobservable factors that might bias the estimates we are interested in, an endogenous switching regression model is applied, an econometric method in which counterfactual analysis is used to obtain the average treatment effect of Internet use on BMI and overweight. The study sample is chosen from longitudinal research data from the China Health and Nutrition Survey, with 12,846 observations covering 5 waves. Estimation results indicate that Internet use could reduce BMI and the probability of being overweight by 3.626% and 33.963%, respectively. Heterogeneity analysis shows that Internet use has a greater effect on males living in urban areas and those with higher education. We also reveal that reducing total energy intake and macronutrient intake and increasing dietary knowledge levels and the time spent on exercising serve as important mechanisms through which Internet use can have an effect on BMI and overweight. This study shows that Internet use has a beneficial effect on BMI and overweight of males in China. Therefore, internet platforms may be an effective way to regulate males’ macronutrient intake and BMI and overweight. More importantly, using Internet platforms to circulate information about healthy diets may contribute to improving dietary knowledge and the prevalence of overweight.

Keywords:Internet use;Males;BMI;Overweight;Macronutrient intake
PDF(1.49 MB)|EndNote (RIS)|BibTeX|RefMan|RefWorks

Cite this article

Weigang Liu, Jian Liu, Jens-Peter Loy, Yanjun Ren. The effect of Internet use on males’ body mass index and overweight: evidence from China.Journal of Men's Health,2023,19(12):11-22 DOI:10.22514/jomh.2023.125

1. Introduction

Over the past two decades, the widespread use and development of the Internet worldwide have undergone significant changes, which have profoundly affected people’s lifestyles. According to the latest World Bank statistics, the proportion of individual Internet users to the total population worldwide has increased dramatically from 1.32% in 1996 to approximately 60% by 2020. Of these, 29 countries have an Internet population of over 90% of the total population [1], and some articles now suggest that the number of Internet users could potentially reach 2.7 billion [2]. The Internet has contributed to the reform of government services and the rapid development of e-government, which allows citizens to participate in state affairs through e-government platforms [3, 4]. The Internet has led to the growth of e-commerce, reducing the cost of customer acquisition and making millennials primarily online shoppers [5, 6, 7]. The Internet has facilitated a shift in the media industry, with online media becoming the focus of attention, and a decline in the use of traditional media [8]. The Internet has transformed the way people socialise, with online interaction making distance a less barrier to communication. All these changes mean that modern people are increasingly inseparable from the Internet, with it playing a greater role in their lives in general and in their daily work in particular, where they rely on the Internet to access information and communicate [9, 10].

Extensive research has investigated the impact of Internet use on the economic outcomes and well-being of its users. In contrast, little attention has been paid to its impact on nutrition-related health. Li and Zhou [11] find that Internet use enhances parent-child contact and increases subjective well-being among older adults; Hong and Chang [12] find that the Internet increases the incomes of forestry growers and thus increases farmers’ subjective well-being. At the same time, studies have clearly shown that Internet use significantly increases the non-farm incomes of farming households in rural areas of developing countries [13, 14]. More recently, it has been found that Internet use may be associated with nutritional outcomes, but conclusions about the relationship between Internet use and BMI are inconsistent. Many studies indicate that Internet use has either no significant effect or a positive effect on overweight and obesity [15, 16, 17]. For instance, Beleigoli et al. [18] find that each additional hour of Internet use is associated with an 8% increase in the odds of being overweight or obese. Another study suggests that Internet use significantly improves dietary habits regarding healthy foods, such as fruits and vegetables [19], and increases contact with sources of health information, which could consistently improve people’s health status [17, 20, 21, 22]. For instance, an empirical study in Serbia, a developing country, shows that more frequent Internet access is likely to result in BMI decrease [23].

Whether the above findings hold true in the transitional economy of China, especially for males, is still unclear. Therefore, this study aims to assess how and to what extent Internet use contributes to the nutritional outcomes of males in China. In China, food consumption has shifted from a situation of national poverty and hunger to one of overweight and obesity caused by excessive macronutrient intakes [24, 25]. According to the China Nutrition and Chronic Disease Report in 2012, the overweight and obesity rates among adults were 22.8% and 7.1%, respectively, while the latest survey results in 2020 show that the overweight and obesity rates have reached 34.3% and 16.4%, respectively [26]. Extensive studies have revealed that the increasing prevalence of overweight and obesity is a major cause of noncommunicable diseases, such as diabetes, cardiovascular diseases and hypertension; it also increases the risk of kidney and colon cancers [27, 28, 29]. This is a significant risk to individual and public health. Thus, understanding the impact of Internet use and its underlying mechanisms on nutritional outcomes has far-reaching implications for policymakers looking to intervene in the prevalence of overweight and obesity in developing countries like China.

More importantly, the effects of Internet use on nutritional outcomes are inconsistent across genders, since males are more likely to be affected by Internet addiction [30, 31, 32] and experience more changes in their nutritional outcomes. According to the available data, the proportion of male Internet users is 2.4% higher than that of females [33]; regarding nutritional outcomes, it is estimated that males are more likely to be overweight and obese in recent years. In 2004, the average BMI and obesity rate of males were lower than those of females, but by 2018, there were 48 million obese adult males and 37 million obese females aged 18 to 69, meaning there were 11 million more obese males than females. The average BMI of males in China increased from 22.5 in 2004 to 24.7 in 2018, and the rate of overweight people increased from 20.9% in 2004 to 45% in 2018 [34]. As a large country with a population of over 1.4 billion, the issue of overweight and obese males in China deserves to be more fully studied and discussed. Changes in people’s income levels and lifestyles are a major cause of this phenomenon. However, the Internet plays a huge role in changing people’s lifestyles. Therefore, this study intends to demonstrate the gender-specific effect of the Internet use on males.

The main aim of this study is to investigate the effect of Internet use on BMI and overweight. The hypothesis of this study is that the Internet, as a medium of information dissemination, enables people who are online to receive more information. Information about healthy eating and health-related information can influence numerous behaviours that can lead to a reduction in people’s BMI and a lower likelihood of being overweight. This is reflected in changes in energy intake, macronutrient intake, improved dietary knowledge and increased time spent exercising. The main research method used in this study is an econometric model, the endogenous switching regression (ESR) model. The main function of this model is that it reduces estimation bias due to unobservable factors and allows for average treatment effects to be derived from counterfactual analysis.

The remainder of the paper will be organised in the following manner. The second section addresses the methods, data, the econometric methods and the sample used in this study. The third section contains the results and discussion, the determinants of Internet use, the impact of Internet use on BMI and overweight, the heterogeneity analysis, and the mechanism analysis. The final section concludes the study.

2. Methods and data

2.1 Method

2.1.1 Endogenous switching regression model

This study focuses on analysing the effect of Internet use on males’ nutritional outcomes. Still, Internet use as a self-determined behaviour does not occur randomly. It might be influenced by subjective and objective factors, some of which cannot be observed and controlled for. These unobservable factors may be related to the outcome variable, leading to the endogeneity problem of Internet use. Thus, ESR model is applied to control for both observable and unobservable characteristics and it allows counterfactual analysis to calculate the average treatment effects. An instrumental variable should be selected when using the ESR model. The availability of Internet cafes in the community is a potential instrumental variable for Internet use. A valid instrumental variable has to satisfy both relevance and exogeneity. First, the number of internet cafes in the community is highly correlated with whether males use the Internet. Second, the number of Internet cafes in the community is not necessarily associated with males who have elevated BMI or who are overweight.

The ESR model generally consists of two stages. In the first stage, a probit or logit model is used to estimate a decision equation for Internet use among males. In the second stage, outcome equations for BMI, overweight and dietary knowledge are developed separately for males to estimate the determinants of the outcome variables. Thus, the ESR model involves the following three equations:

The decision equation regarding whether to use the Internet or not:

Di*=γZi+πIi+μi,Di={1ifDi*>00otherwise

The outcome equation for males using the Internet:

Yni=βnXni+εni if Di=1

The outcome equation for males who do not use the Internet:

Yui=βuXui+εui if Di=0

In Eqn. 1, Di represents the binary choice variable of whether males use the Internet, Zi is the set of vectors that affect whether the Internet is used, Ii represents the vector set of the instrumental variables, and μi is the error term. Eqn. 2 represents the outcome variable equation for males who use the Internet. Yni represents BMI or overweight or dietary knowledge. Xni represents the variable of interest affecting the outcome variable. ϵni represents the error term of the equation. The variables in Eqn. 3 have the same meanings as those in Eqn. 2, with the difference that Eqn. 3 represents the decision equation for the outcome variables for males who do not use the Internet.

2.1.2 Estimating treatment effects

The average treatment effect can be further calculated with the help of ESR model. The observed and unobserved counterfactual outcomes of using the Internet are presented below [35, 36]:

The expectations of the outcome variables for males using the Internet (observed):

E[Yni∣Di=1]=βnXni+σμnλin

The expectations of the outcome variables for males who use the Internet in the absence of Internet use (counterfactual):

E[Yiu∣Di=1]=βuXni+σμuλin

The average treatment effect on the treated is shown below:

ATTi=E[Yin∣Di=1]−E[Yiu∣Di=1]

2.2 Data

2.2.1 The sample

The sample used in this study is from the China Health and Nutrition Survey (CHNS), a collaborative project between the University of North Carolina and the Chinese Centre for Disease Control and Prevention. The data set includes 12 provinces and 3 municipalities: Liaoning, Heilongjiang, Jiangsu, Shandong, Henan, Hubei, Hunan, Guangxi, Guizhou, Yunnan, Zhejiang, Shaanxi, Beijing, Shanghai and Chongqing. 10 periods of data collection are conducted in the survey of 7200 families with more than 30,000 individuals: 1989, 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011 and 2015.

The data used in this study are longitudinal data from the years 2004 to 2015. The sample for this study focuses on a group of males aged 25–60 years old. The final sample size, after excluding missing values and outliers, is 12,846. As the CHNS information on energy intake and macronutrient intake is only updated in 2011, a sample size of 10,256 is used in the analysis of the channels of the role of the Internet on BMI and overweight. All of the following data analyses are carried out using STATA version 17 (StataCorp, College Station, TX, USA).

2.2.2 Variables

All the relevant variables are defined in Table 1. The treatment variable is whether one uses the Internet. It is defined in the questionnaire as “Can you access the Internet?”. Based on all the samples, the proportion of Internet use is 28.3%, and according to the most recent data from 2015, the proportion of Internet use reaches 45.7%. The ratio of Internet use shows an increasing trend over the years. The nutritional outcomes comprise two variables: body mass index (BMI, kg/m2) and overweight. BMI is calculated using body weight divided by height squared and is a commonly used measure of physical health. According to the Asian population criteria proposed by the World Health Organization and the Chinese criteria proposed by the National Health and Family Planning Commission of the People’s Republic of China [37, 38], a person with a BMI equal to or greater than 24 is defined as overweight.

Table 1.Descriptive statistics of the variables.
VariablesDefinitionsObs.AllInternet usersNon-Internet usersDifference
Treatment variables
Internet use1 if use Internet, 0 otherwise12,8460.283 (0.450)
Nutritional outcomes
BMIBody mass index (kg/m2)12,84623.839 (3.300)24.396 (3.307)23.619 (3.272)0.777***
Overweight1 if BMI ≥24, 0 otherwise12,8460.453 (0.498)0.528 (0.499)0.423 (0.494)0.105***
Channel variables
Energy intakeNatural logarithm of 3-day ave: Energy (kcal)10,2567.727 (0.314)7.674 (0.339)7.744 (0.304)−0.070***
CarbohydrateNatural logarithm of 3-day ave: Carbohydrate (g)10,2565.721 (0.375)5.573 (0.378)5.767 (0.362)−0.194***
ProteinNatural logarithm of 3-day ave: Protein (g)10,2564.264 (0.333)4.305 (0.331)4.251 (0.332)0.055***
FatNatural logarithm of 3-day ave: Fat (g)10,2564.269 (0.547)4.360 (0.525)4.239 (0.551)0.121***
Dietary knowledgeDietary knowledge indexa12,8467.180 (4.146)8.934 (3.887)6.489 (4.040)2.445***
ExerciseExercise time per day (hours)12,8460.255 (1.022)0.520 (1.366)0.151 (0.826)0.369***
Control variables
AgeYears12,84645.262 (9.509)41.321 (9.660)46.814 (8.987)−5.493***
Age squaredNatural logarithm of age squared12,8467.577 (0.452)7.387 (0.483)7.652 (0.415)−0.265***
EducationYears of education12,84610.059 (4.268)13.286 (4.215)8.788 (3.560)4.498***
Marital status1 if the head of the household is married, 0 otherwise12,8460.927 (0.260)0.898 (0.302)0.938 (0.240)−0.040***
Work1 if currently has a job, 0 otherwise12,8460.885 (0.319)0.936 (0.244)0.865 (0.342)0.071***
Urban1 if living in an urban area, 0 otherwise12,8460.340 (0.474)0.605 (0.489)0.236 (0.425)0.369***
Household sizeHousehold size12,8463.185 (1.342)3.294 (1.297)3.142 (1.357)0.152***
IncomeNatural logarithm of the per capita household income (yuan) inflated to 2015 value12,8469.560 (1.347)10.194 (1.006)9.311 (1.382)0.883***
Instrumental variable
Internet caféNumber of internet cafés in the community.12,8461.867 (4.873)2.582 (5.125)1.586 (4.740)0.996***
Source: Author’s calculations using the CHNS data (2004–2015). Standard errors are presented in parentheses. ***p < 0.010. a Calculating by scoring 12 dietary knowledge questions. Using a 5-point scale to distinguish the level of accuracy in responding to each question. The higher the score, the higher the level of dietary knowledge. BMI: body mass index.

Table 1 shows a significant difference in BMI and overweight between males who use the Internet and those who do not at the 1% statistical level. This indicates that there may be significant effects of Internet use on the BMI and overweight of males, and these differences may be caused by a combination of observable and unobservable factors, so further work needs to be done to estimate the effects of Internet use on the level of the nutritional health of males. Looking at the distribution of the BMI sample values, as shown in Fig. 1, the kernel density plot for all the years in the sample indicates that the proportion of overweight individuals has reached 45% and the ratio of obese people has exceeded 10.8%. Not only is the overall percentage of overweight individuals alarming, but the rate of BMI growth among Chinese males is also very noteworthy. As shown in Fig. 2, the BMI of urban males is higher than the overall average, while the average BMI of rural males is lower than the overall average and much lower than that of urban males. The mean BMI shows an upward trend overall, whether observing the entire sample or the urban and rural samples separately.

Distribution of body mass index (BMI). Note: A body mass index equal to 24 or above is defined as overweight. A body mass index equivalent to 28 or above is defined as obesity. Source: CHNS 2004, 2006, 2009, 2011 and 2015.

Fig. 1.Distribution of body mass index (BMI). Note: A body mass index equal to 24 or above is defined as overweight. A body mass index equivalent to 28 or above is defined as obesity. Source: CHNS 2004, 2006, 2009, 2011 and 2015.

Average body mass index (BMI) changes in urban and rural areas in 2004–2015. Source: CHNS 2004, 2006, 2009, 2011 and 2015.

Fig. 2.Average body mass index (BMI) changes in urban and rural areas in 2004–2015. Source: CHNS 2004, 2006, 2009, 2011 and 2015.

The channel variable consists of four components, the first of which is energy intake. This is the average of the three-day energy intake of the respondents. In terms of total energy intake, internet users intake less energy. The second is macronutrient intake, including carbohydrates, proteins and fats. In CHNS, food consumption records of the participants are taken with the food consumption record form for 3 days and nutrient intakes are evaluated as a result of the average of these 3 days. High-quality protein intake has an effect on appetite suppression, as does simple carbohydrate-rich (such as sugar-sweetened beverages) increases the likelihood of obesity, and a high intake of low-quality fat may also lead to obesity [39]. Carbohydrate and fat are considered to be two important macronutrients related to obesity, and the balance of intake of these two macronutrients is an important element of obesity in the male population [40]. Moreover, these three macronutrients have been shown to be important as part of people’s daily intake, which is why we chose them as one of the channels [41]. Table 1 shows that the intakes of these three nutrients differ significantly between Internet users and non-users, so Internet use has an effect on macronutrient intake. The other two channel variables are the time spent exercising and dietary knowledge. Exercise time is defined as the amount of time spent exercising each day, including ball games, running and swimming. Dietary knowledge index is a synthetic indicator that is calculated by scoring respondents’ answers to 12 questions related to dietary knowledge using a 5-point scale to distinguish the level of accuracy in responding to each question, ranging from correct to incorrect on a scale of 2 to −2. The final scores for all the questions are added up to obtain the variable of dietary knowledge. Thus, higher dietary knowledge scores indicate higher levels of dietary knowledge. The mean values of exercise time and dietary knowledge are higher for Internet users than those who do not use the Internet. The mean value of exercise time per day for Internet users is 0.520 hours, and the mean level of dietary knowledge is 8.934 points. This compares with an average of 0.151 hours of exercise and a dietary knowledge level of 6.489 points for males who do not use the internet.

Drawing on existing studies, the control variables are selected to include individual-level variables of interest (age, age squared, education, marital status and current employment) and household-level variables (residence location, household size and annual per capita household income) [41, 42]. The average age of the sample is approximately 45 years old, and the average number of years of education is about 10 years. More than 92% of the respondents are married, about 88% are currently working, 34% of the sample live in the city, and the average household size is about 3 persons—the covariates for males who use the Internet and males who do not are statistically significantly different. Specifically, males who use the Internet are younger, more educated, somewhat less prone to be married, more likely to be currently working and more inclined to be living in the city in a slightly larger household and with a somewhat higher income. The characteristics of males who use the Internet are not consistent and are significantly different from males who do not use the Internet, and such selectivity bias can lead to biased estimation results.

3. Results

This section begins with a correlation analysis of the key variables, confirming the existence of a correlation between internet use and both the outcome variable and the channel variable. In the second section, an analysis of the factors influencing internet use is conducted, which forms the basis of the ESR model. In the third section, the effects of BMI and overweight are estimated using the ERS model. The fourth section analyses urban-rural and educational attainment, two variables that can make a significant difference, to see the impact of Internet use on different samples. The fifth section conducts a mechanistic analysis in an attempt to find the channels through which the Internet acts on BMI and overweight.

3.1 Correlation analysis

Table 2 provides a Pearson correlation analysis of Internet use with the outcome variables as well as the channel variables. As shown in Table 2, Internet use is significant at the 1% level with both outcome variables and with all three sets of channel variables. Specific results are reported below. First, Internet use is positively correlated with BMI and overweight. Second, Internet use is not consistently correlated with the three nutrients in the same direction. It is negatively correlated with carbohydrate intake and is positively correlated with protein intake and fat intake. Third, Internet use is positively associated with dietary knowledge and is positively correlated with time spent exercising. This is only the correlation between the treatment, outcome, and channel variables. After adding relevant control variables and considering the endogenous problems, the relationship between Internet use and nutritional outcomes needs further analysis.

Table 2.Correlation matrix for study variables.
InternetBMIOverweightEnergy intakeCarbohydrateProteinFatDietary knowledgeExercise
Internet1
BMI0.106***1
Overweight0.095***0.802***1
Energy intake−0.095***−0.0030.0071
Carbohydrate−0.221***−0.075***−0.065***0.751***1
Protein0.070***0.069***0.073***0.728***0.507***1
Fat0.095***0.082***0.088***0.628***0.078***0.455***1
Dietary knowledge0.266***0.117***0.108***−0.077***−0.183***0.028***0.090***1
Exercise0.162***0.053***0.060***−0.028***−0.078***0.053***0.043***0.070***1
Note: Author’s calculations using the CHNS data (2004–2015). ***p < 0.010. BMI: body mass index.

3.2 Determinants of Internet use

The estimation results using ESR model are presented in Table 3. In the analysis of this paper, the instrumental variable corresponding to Internet use is used to better estimate the determinants of such use based on the presence of Internet cafes in the community. rho0 and rho1 indicate that there is a significant correlation between the determinants of the Internet use equations for BMI and overweight and the determinants of BMI and that there is endogeneity, which can be further calculated with the help of ESR model. The Wald test shows that Prob > chi2 = 0, a result that suggests that Internet use is not an exogenous variable. All the above descriptions suggest that ESR model used in this study can estimate males’ health. Furthermore, the counterfactual analysis can calculate the mean treatment effect.

The second and fifth columns of Table 3 report the determinants of Internet use, which are calculated using Eqn. 1. The results show that the coefficient of the age-squared term is negative and significant, indicating an inverted u-shaped distribution of Internet use among males. The positive and significant coefficient of years of education shows that males with higher education are more likely to use the Internet, which is consistent with existing studies [43, 44]. The negative and statistically significant coefficient of marital status indicates that single people are more likely to use the Internet than married people, which may be because single people have a greater willingness to socialise. The Internet provides a broader platform for making friends. The coefficient of current work status is positive and significant, indicating that work increases the likelihood of using the Internet compared to those who do not work. In this information age, the Internet is an essential means of communication, and all kinds of work are becoming more and more closely related to the Internet. The coefficient of urban is positive and significant, indicating that males living in urban areas are more likely to use the Internet compared to those living in rural areas.

The third and fourth columns of Table 3 report the determinants of BMI. The sixth and seventh columns report the determinants of overweight. The results show that the coefficients of age, age squared, education, urban and income are statistically significant for both Internet users and non-users. All the factors with significant coefficients show a positive effect on BMI, except for age, which has a u-shaped impact.

Table 3.The Determinants of internet use based on ESR model.
BMIOverweight
VariablesInternetInternet usersNon-usersInternetInternet usersNon-users
Age0.0013 (0.0140)−0.2028*** (0.0537)−0.1340*** (0.0337)−0.0021 (0.0141)−0.0511* (0.0203)−0.0420** (0.0139)
Age squared−1.0088*** (0.2897)3.6226*** (1.0978)2.7198*** (0.7345)−0.9369*** (0.2921)1.1848** (0.4078)0.7931** (0.3096)
Education0.1151*** (0.0054)0.1348** (0.0505)0.0931*** (0.0100)0.1194*** (0.0046)0.0132 (0.0132)0.0503*** (0.0052)
Marital status−0.1089* (0.0571)0.1518 (0.2228)0.3459** (0.1343)−0.1196* (0.0571)0.1061 (0.0807)0.0758 (0.0585)
Work0.1422** (0.0544)0.3457 (0.2513)−0.1339 (0.1020)0.1335* (0.0548)0.0657 (0.0956)−0.0558 (0.0430)
Urban0.5994*** (0.0339)1.1056*** (0.2971)0.7013*** (0.0877)0.5993*** (0.0339)0.2157** (0.0814)0.3400*** (0.0382)
Household size−0.0084 (0.0121)−0.0005 (0.0505)−0.0321 (0.0250)−0.0085 (0.0122)−0.0094 (0.0197)0.0008 (0.0108)
Income0.1583 (0.0172)0.2292* (0.1018)0.1512*** (0.0251)0.1588*** (0.0171)0.0252 (0.0315)0.0653*** (0.0117)
Internet café0.0089** (0.0029)0.0088** (0.0030)
Constant3.3048* (1.5683)−1.5988 (5.7484)7.2321* (4.0498)2.8924* (1.5796)−7.3468*** (2.1548)−5.4214** (1.7122)
YearYesYesYesYesYesYes
ProvinceYesYesYesYesYesYes
rho10.6597** (0.2417)0.1510 (0.1899)
rho00.1514*** (0.0291)0.4501*** (0.0896)
Log pseudolikelihood−37,591.32−13,070.13
Wald test of indep. Eqnchi2 (2) = 35.57 Prob > chi2 = 0chi2 (2) = 19.24 Prob > chi2 = 0
Obs.12,84612,846
Note: Author’s calculations using the CHNS data (2004–2015). Standard errors are presented in parentheses. *p < 0.10, **p < 0.05, ***p < 0.010. BMI: body mass index.

3.3 The effect of Internet use on BMI and overweight

The results in the second column of Table 4 are calculated using Eqn. 4 for the BMI and overweight expectations for males who use the Internet. The values in the third column of Table 4 are determined using Eqn. 5 to obtain estimates of the outcome variables for males who do not use the Internet. The results in the fourth column of Table 4 are calculated using Eqn. 6 for the outcome, the average treatment effect on the treated (ATT). The last column estimates the degree of variation in the outcome variable. The results in Table 3 show that the expectation of BMI for males using the Internet is 24.399, while the expectation of BMI for males not using the Internet is 25.316, with an ATT value of −0.917 and is statistically significant. Therefore, the expectation of BMI is reduced by 3.626% due to Internet use. For males using the Internet the expectation of being overweight is 0.528, while for males not using the Internet the expectation of being overweight is 0.800 and the ATT value is −0.272 and is statistically significant. Thus, the expectation of being overweight is reduced by 33.963% due to the use of the Internet.

Table 4.The overall effects of Internet use on nutritional outcomes based on the ESR model.
Internet usersNon-usersATTChange (%)
BMI24.399 (0.017)25.316 (0.018)−0.917*** (0.025)−3.626
Overweight0.528 (0.002)0.800 (0.002)−0.272*** (0.001)−33.963
Note: Author’s calculations using the CHNS data (2004–2015). Standard errors are presented in parentheses. ***p < 0.010. ESR: endogenous switching regression model. ATT: average treatment effect on the treated. BMI: body mass index.

3.4 Heterogeneous analysis

The results are estimated separately for the urban and rural samples because of the large differences in Internet infrastructure conditions and economic levels between urban and rural areas of China, which may have different effects on Internet use. The results in Table 5 show the ATT for the outcome variables. Urban males experience a 3.609% reduction in their BMI and a 28.109% decrease in the likelihood of being overweight after using the Internet. Rural males experience a 0.358% reduction in their BMI and a 15.400% decrease in the likelihood of being overweight after using the Internet. It is clear that the impact of Internet use is higher for urban males. Table 5 also reports the extent to which males who have completed nine years of compulsory education or not responded differently with regard to Internet use. Males who have completed 9 years of compulsory education experience a 3.401% reduction in their BMI and a 31.500% decrease in the likelihood of being overweight. In comparison, males who have not completed 9 years of compulsory education experience a 1.020% reduction in their BMI and a 16.410% decrease in the likelihood of being overweight after using the Internet. Internet use more greatly impacts males who have completed nine years of compulsory schooling.

Table 5.The effect of heterogeneity on BMI and overweight.
CategoryBMIOverweight
Internet usersNon-usersATTChange (%)Internet usersNon-usersATTChange (%)Obs.
Urban24.618 (0.015)25.540 (0.020)−0.922*** (0.025)−3.6090.562 (0.002)0.782 (0.002)−0.220* (0.001)−28.1092195
Rural24.056 (0.035)24.143 (0.031)−0.086* (0.047)−0.3580.478 (0.004)0.564 (0.004)−0.087*** (0.003)−15.4001435
Compulsory education24.618 (0.015)25.540 (0.020)−0.869*** (0.026)−3.4010.562 (0.002)0.782 (0.002)−0.246* (0.002)−31.5002587
Non-compulsory education24.056 (0.035)24.143 (0.031)−0.246* (0.002)−1.0200.478 (0.004)0.564 (0.004)−0.093*** (0.003)−16.4101043
Note: Author’s calculations using the CHNS data (2004–2015). Standard errors are presented in parentheses. *p < 0.10, ***p < 0.010. BMI: body mass index. ATT: average treatment effect on the treated.

3.5 Channel analysis

Table 6 reports the effects of Internet use on dietary knowledge, energy intake, macronutrient intake and the time spent exercising. These are the four channels used in this study to analyse the effects of Internet use on the nutritional health of males. The estimation results are still calculated using the ESR model and counterfactual analysis. First, the ATT value for dietary knowledge means a 3.068% increase in dietary knowledge for non-Internet users after using the Internet, as calculated by counterfactual analysis. Second, the ATT of energy intake for men indicates that non-Internet users have a 1.211% lower energy intake after using the Internet. Third, the results for macronutrient intake are statistically significant, with a 2.067% decrease in carbohydrate intake, a 2.685% decrease in protein intake and a 12.603% decrease in fat intake. Finally, the ATT of Internet use on exercise time in males is 0.198 and statistically significant, revealing a 61.649% increase in exercise time.

Table 6.The impact of Internet use on macronutrient intakes, exercise time and dietary diversity.
VariablesInternet usersNon-usersATTChange (%)
Dietary knowledge8.927 (0.027)8.661 (0.025)0.266*** (0.037)3.068
Energy intake7.671 (0.002)7.764 (0.003)−0.094*** (0.004)−1.211
Macronutrient intake
Carbohydrate5.573 (0.004)5.691 (0.004)−0.118*** −(0.005)−2.067
Protein4.305 (0.002)4.424 (0.002)−0.119*** (0.003)−2.685
Fat4.361 (0.002)4.989 (0.003)−0.629* (0.004)−12.603
Exercise0.520 (0.005)0.322 (0.003)0.198*** (0.006)61.649
Note: Author’s calculations using the CHNS data (2004–2015). Standard errors are presented in parentheses. *p < 0.10, ***p < 0.010. ATT: average treatment effect on the treated.

4. Discussion

This contribution of Internet use to dietary knowledge in males is consistent with the findings of existing studies. It has been shown that Internet use has a significant positive impact on the health of males in rural China, improving the quality of their diet [45, 46]. Some college students actively access health-related information on the Internet [47]. When Internet users have a higher level of dietary knowledge, they will, at the same time, pay more attention to the healthiness of their dietary practices. This healthy food intake is reflected in a healthier intake of macronutrients and less overconsumption. Existing studies focus on the fact that longer leisure-time Internet use leads to longer sedentary activities, which can lead to “sedentary disease”, resulting in a higher BMI and the increased likelihood of being overweight [15, 48]. These studies have only focused on developed countries (e.g., the United States and Australia) and not on populations in developing countries, especially not China. According to the available studies, leisure time and leisure time physical activity (LTPA) among the Chinese have not always been on the rise, with Nan noting that the average leisure time in some major Chinese cities fell to 2 hours in 1990, almost half the 1986 level [49]. Zou pointed out that the age-standardised LTPA prevalence among Chinese adults aged 18 and beyond increased from 7.13% in 2000 to 11.79% in 2011 and then declined to 7.33% in 2015 [50]. So, there is also no evidence that Internet use leads to an overall increase in leisure sedentary time among adults. In contrast, the results of this study show that adult men who use the Internet spend more time exercising than males who do not use the Internet, suggesting that increased exercise time is an important channel through which Internet use reduces BMI and overweight in males.

The main contributions are as follows. First, ESR model is applied to the topic of the impact of Internet use on nutritional outcomes to address the possible issues of the samples not being randomly distributed and of unobservable factors leading to biased estimates. Second, counterfactual analyses estimate that Internet use reduces BMI levels by approximately 3.6% and the likelihood of being overweight by about 34% in Chinese males. Third, for the first time, the impact of Internet use on nutritional outcomes is analysed in urban and rural areas and for males with different levels of education. The results reveal that urban areas are more affected than rural areas and that males with higher levels of education are more affected than males with lower levels of education. Fourth, the mechanisms through which Internet use reduces BMI and overweight are analysed for the first time.

The limitations of this study lie in two areas. On the one hand, it is due to the limitation of data timeliness, the sample only includes information up to 2015. From 2015 to the present, China’s internet infrastructure has been greatly improved and the gradual rise of short-form video software has gradually lowered the threshold of internet use, allowing the internet to increase its level of influence on people. With recent research data, it would be possible to study the current impact of the internet on people in more detail and with greater accuracy. On the other hand, there are a large number of missing observations in the surveys of various types of internet use activities and the corresponding hours of use. Therefore, this study is unable to explore the effect of different uses and duration of the Internet use on BMI and overweight. Further exploration is needed in future studies.

5. Conclusions

This study estimates the effect of Internet use on males’ health and dietary knowledge using data from the CHNS for the five periods of 2004, 2006, 2009, 2011 and 2015 and based on a total valid sample size of 12,846. This study applied the ESR model to address bias due to observable and unobservable factors while also grappling with the endogeneity of Internet use as a behaviour that causes the inconformity of the sample to a random distribution.

The results show that Internet use leads to a significant reduction in BMI and a significant decrease in the likelihood of being overweight in males. Regarding rural and urban differences, Internet use significantly reduces BMI levels for urban and rural males, with a somewhat larger effect for urban males. In terms of the completion of compulsory education, Internet use results in a significant reduction in BMI levels and the likelihood of being overweight in males, regardless of whether they completed nine years of compulsory education or not, but it has a greater impact on males who have completed nine years of compulsory schooling. The analysis of the mechanisms of Internet use on BMI and overweight in males through four channels shows that Internet use significantly increases dietary knowledge, reduces energy intake and macronutrient intake and increases the time spent exercising.

Overweight and obesity in males are increasingly becoming a significant part of public health expenditure, and the question of reducing BMI and promoting healthy eating in males has become a topical issue. This article can provide effective ways to reduce the expenditure by analysing the impact of Internet use on BMI and overweight in males. Studies have used the Internet to promote weight or waist circumference loss with good results by sending emails or targeting information about weight loss programmes through software [18, 51, 52]. The significance of this study is the confirmation of the role of this behaviour of Internet use on the increased dietary knowledge of adult males and the resulting reduction in energy intake and macronutrient intake, increase in exercise time. Efforts should be made to continue the positive and active role of the Internet in knowledge diffusion and information dissemination to disseminate correct dietary knowledge and healthy eating practices to a wider population through more effective methods.

Availability of Data and Materials

The data presented in this study are available on reasonable request from the corresponding author.

More information, see https://www.cpc.unc.edu/projects/china.

Author contributions

YJR—Study Design. WGL—Data Analysis, Writing Original Draft. JL, YJR, JPL and WGL—Manuscript Review and Editing. All authors have read and agreed to the published version of the manuscript.

Ethics approval and consent to participate

Not applicable.

Acknowledgment

Weigang Liu and Jian Liu acknowledge the China Scholarship Council for their support.

Funding

This work was supported by the Chinese Universities Scientific Fund (grant number: Z1010422003), the Qinchuangyuan Project of Shaanxi Province (grant number: QCYRCXM-2022-145), and the major projects of the humanities and Social Sciences base of the Ministry of Education (22JJD790052).

Conflict of interest

The authors declare no conflict of interest.

References

World Bank. World development indicators. 2022. Available at: https://data.worldbank.org/indicator (Accessed: 08 September 2022).

[Google Scholar]

Hanson W. A global internet: the next four billion users. New Space. 2015; 3: 204–207.

[Google Scholar]

Linders D. From e-government to we-government: defining a typology for citizen coproduction in the age of social media. Government Information Quarterly. 2012; 29: 446–454.

[Google Scholar]

Hampton KN, Livio O, Sessions Goulet L. The social life of wireless urban spaces: Internet use, social networks, and the public realm. Journal of Communication. 2010; 60: 701–722.

[Google Scholar]

Miyatake K, Nemoto T, Nakaharai S, Hayashi K. Reduction in consumers’ purchasing cost by online shopping. Transportation Research Procedia. 2016; 12: 656–666.

[Google Scholar]

Monsuwe TPY, Dellaert BGC, De Ruyter K. What drives consumers to shop online? A literature review. International Journal of Service Industry Management. 2004; 15: 102–121.

[Google Scholar]

Lissitsa S, Kol O. Generation X vs. Generation Y—a decade of online shopping. Journal of Retailing and Consumer Services. 2016; 31: 304–312.

[Google Scholar]

Dimmick J, Chen Y, Li Z. Competition between the internet and traditional news media: the gratification-opportunities niche dimension. Journal of Media Economics. 2004; 17: 19–33.

[Google Scholar]

Cameron AF, Webster J. Unintended consequences of emerging communication technologies: instant messaging in the workplace. Computers in Human Behavior. 2005; 21: 85–103.

[Google Scholar]

Jayawardhena C, Foley P. Changes in the banking sector—the case of Internet banking in the UK. Internet Research. 2000; 10: 19–31.

[Google Scholar]

Li J, Zhou X. Internet use and Chinese older adults’ subjective well-being (SWB): the role of parent-child contact and relationship. Computers in Human Behavior. 2021; 119: 106725.

[Google Scholar]

Hong Y, Chang H. Does digitalization affect the objective and subjective wellbeing of forestry farm households? Empirical evidence in Fujian Province of China. Forest Policy and Economics. 2020; 118: 102236.

[Google Scholar]

Zhou X, Cui Y, Zhang S. Internet use and rural residents’ income growth. China Agricultural Economic Review. 2020; 12: 315–327.

[Google Scholar]

Ma W, Nie P, Zhang P, Renwick A. Impact of Internet use on economic well-being of rural households: evidence from China. Review of Development Economics. 2020; 24: 503–523.

[Google Scholar]

Vandelanotte C, Sugiyama T, Gardiner P, Owen N. Associations of leisure-time internet and computer use with overweight and obesity, physical activity and sedentary behaviors: cross-sectional study. Journal of Medical Internet Research. 2009; 11: e28.

[Google Scholar]

Baek S-I, So W-Y. Association between times spent on the internet and weight status in Korean adolescents. Iranian Journal of Public Health. 2011; 40: 37–43.

[Google Scholar]

Melchior M, Chollet A, Fombonne E, Surkan PJ, Dray-Spira R. Internet and video game use in relation to overweight in young adults. American Journal of Health Promotion. 2014; 28: 321–324.

[Google Scholar]

Beleigoli AM, Andrade AQ, Cancado AG, Paulo MNL, Diniz MDFH, Ribeiro AL. Web-based digital health interventions for weight loss and lifestyle habit changes in overweight and obese adults: systematic review and meta-analysis. Journal of Medical Internet Research. 2019; 21: e298.

[Google Scholar]

Souza BB, Cembranel F, Hallal ALC, d’Orsi E. Consumption of fruits and vegetables and association with life habits and nutritional status: a prospective study in a cohort of the elderly. Ciência & Saúde Coletiva. 2019; 24: 1463–1472. (In Portuguese)

[Google Scholar]

Aghasi M, Matinfar A, Golzarand M, Salari-Moghaddam A, Ebrahimpour-Koujan S. Internet use in relation to overweight and obesity: a systematic review and meta-analysis of cross-sectional studies. Advances in Nutrition. 2020; 11: 349–356.

[Google Scholar]

Anderson JG. Consumers of e-Health: patterns of use and barriers. Social Science Computer Review. 2004; 22: 242–248.

[Google Scholar]

Reinfeld-Kirkman N, Kalucy E, Roeger L. The relationship between self-reported health status and the increasing likelihood of South Australians seeking Internet health information. Australian and New Zealand Journal of Public Health. 2010; 34: 422–426.

[Google Scholar]

Belojevic G, Sokolova Djokic L, Gligorova B, Banjari I, Stojanovic M, Stojanovic D. Body mass index and the literacy on obesity in relation to media following. Iranian Journal of Public Health. 2018; 47: 1166–1171.

[Google Scholar]

Popkin BM. Global nutrition dynamics: the world is shifting rapidly toward a diet linked with noncommunicable diseases. The American Journal of Clinical Nutrition. 2006; 84: 289–298.

[Google Scholar]

Popkin BM, Gordon-Larsen P. The nutrition transition: worldwide obesity dynamics and their determinants. International Journal of Obesity. 2004; 28: S2–S9.

[Google Scholar]

National Health Commission of the People’s Republic of China. 2022. Available at: http://en.nhc.gov.cn/ (Accessed: 17 July 2022).

[Google Scholar]

Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. The New England Journal of Medicine. 2003; 348: 1625–1638.

[Google Scholar]

Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. The Lancet. 2014; 384: 766–781.

[Google Scholar]

Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012. CA: A Cancer Journal for Clinicians. 2015; 65: 87–108.

[Google Scholar]

Chou C, Condron L, Belland JC. A review of the research on internet addiction. Educational Psychology Review. 2005; 17: 363–388.

[Google Scholar]

Shaw M, Black DW. Internet addiction: definition, assessment, epidemiology and clinical management. CNS Drugs. 2008; 22: 353–365.

[Google Scholar]

Wang L, Luo J, Bai Y, Kong J, Luo J, Gao W, et al. Internet addiction of adolescents in China: prevalence, predictors, and association with well-being. Addiction Research & Theory. 2013; 21: 62–69.

[Google Scholar]

China Internet Network Information Center. The 48th Statistical Report on China’s Internet Development. 2021. Available at: https://www.cnnic.com.cn/IDR/ReportDownloads/ (Accessed: 19 November 2021).

[Google Scholar]

Wang L, Zhou B, Zhao Z, Yang L, Zhang M, Jiang Y, et al. Body-mass index and obesity in urban and rural China: findings from consecutive nationally representative surveys during 2004–18. The Lancet. 2021; 398: 53–63.

[Google Scholar]

Liu M, Min S, Ma W, Liu T. The adoption and impact of E-commerce in rural China: application of an endogenous switching regression model. Journal of Rural Studies. 2021; 83: 106–116.

[Google Scholar]

Ma W, Renwick A, Nie P, Tang J, Cai R. Off-farm work, smartphone use and household income: evidence from rural China. China Economic Review. 2018; 52: 80–94.

[Google Scholar]

Akram DS, Astrup AV, Atinmo T, Boissin JL, Bray GA, Carroll KK, et al. Obesity: preventing and managing the global epidemic. World Health Organization-Technical Report Series. 2000; I–XII+1–253.

[Google Scholar]

WS/T 428-2013. Criteria of weight for adults. National Health and Family Planning Commission of the People’s Republic of China: Beijing. 2013.

[Google Scholar]

San-Cristobal R, Navas-Carretero S, Martínez-González MÁ, Ordovas JM, Martínez JA. Contribution of macronutrients to obesity: implications for precision nutrition. Contribution of macronutrients to obesity: implications for precision nutrition. Nature Reviews Endocrinology. 2020; 16: 305–320.

[Google Scholar]

Hossain S, Munni US, Khan T, Fatema K, Ali L. Association of macronutrient intake with overweight and obesity among bangladeshi population: a hospital based study. Annals of Nutrition and Metabolism. 2013; 63: 1382–1382.

[Google Scholar]

Ren Y, Li H, Wang X. Family income and nutrition-related health: evidence from food consumption in China. Social Science & Medicine. 2019; 232: 58–76.

[Google Scholar]

Wang R, Feng Z, Liu Y, Qiu Y. Is lifestyle a bridge between urbanization and overweight in China? Cities. 2020; 99: 102616.

[Google Scholar]

Lera-López F, Billon M, Gil M. Determinants of Internet use in Spain. Economics of Innovation and New Technology. 2011; 20: 127–152.

[Google Scholar]

Močnik D, Širec K. The determinants of Internet use controlling for income level: cross-country empirical evidence. Information Economics and Policy. 2010; 22: 243–256.

[Google Scholar]

Sun Y, Dong D, Ding Y. The impact of dietary knowledge on health: evidence from the china health and nutrition survey. International Journal of Environmental Research and Public Health. 2021; 18: 3736.

[Google Scholar]

Ma B, Jin X. Does internet use connect us to a healthy diet? Evidence from Rural China. Nutrients. 2022; 14: 2630.

[Google Scholar]

Tariq A, Khan SR, Basharat A. Internet use, eHealth literacy, and dietary supplement use among young adults in Pakistan: cross-sectional study. Journal of Medical Internet Research. 2020; 22: e17014.

[Google Scholar]

Matusitz J, McCormick J. Sedentarism: the effects of internet use on human obesity in the United States. Social Work in Public Health. 2012; 27: 250–269.

[Google Scholar]

Nan C, Tsai CTL. The relationships between leisure time, perceived stress and subjective well-being in young and middle-aged working adults in China. Leisure Sciences. 2022; 1–19.

[Google Scholar]

Zou Q, Wang H, Du W, Su C, Ouyang Y, Wang Z, et al. Trends in leisure-time physical activity among Chinese adults — China, 2000–2015. China CDC Weekly. 2020; 2: 135–139.

[Google Scholar]

Seo DC, Niu J. Evaluation of internet-based interventions on waist circumference reduction: a meta-analysis. Journal of Medical Internet Research. 2015; 17: e181.

[Google Scholar]

McCully SN, Don BP, Updegraff JA. Using the internet to help with diet, weight, and physical activity: results from the health information national trends survey (HINTS). Journal of Medical Internet Research. 2013; 15: e148.

[Google Scholar]