RESNA 26th International Annual Confence
A stark digital divide remains between persons with and without disabilities. This paper utilizes linear regression modeling to identify the causes of differences in Internet usage between the two groups. Regression models indicate that computer ownership, family income, educational status, metropolitan status and race affect Internet usage in both persons with and without disability. The Chow test (F = 6.90, p = .01, alpha = .05) suggests that these models of Internet usage are statistically different and thus suggesting further need to study the impact of factors such as use of assistive technology and human support in Internet usage for persons with disabilities.
Digital technology has assisted in generally achieving greater prosperity in the U.S. even though the growth of digital technology has been uneven. Several groups of individuals still use digital technology significantly less than the national average (1). Persons with disabilities comprise one such group of individuals whose use of digital technology lags behind the national average. In the year 1998, Americans with disabilities were less than half as likely to own a computer and about one-quarter as likely to use the Internet as compared to persons without a disability (1). In the year 2000, this digital divide seemed to have expanded even though the percent of individuals who accessed the Internet and who owned a computer have risen for both the persons with and without a disability (2). Thus disability seems to be one major barrier towards full participation in the new digital economy. To reduce the disparity in the use of digital technology between persons with and without disabilities, we need to investigate the reasons behind this inequality so that full participation in this digital economy can be further promoted. Thus, there is a need to identify the factors that make the use of digital technology different in persons with and without disabilities.
The use of the Internet can positively impact the lives of persons with disabilities as it can be increasingly used to conduct day-to-day activities (2). Persons with disabilities can use the Internet to pay bills, find directions, shop on-line, all from the convenience of their home. Access to Internet appears to be affected by multiple factors like every computer owned is not used to access the Internet. This seems to hold true for all populations (1). To reduce this difference between the availability of computers and their use to connect to the Internet, it becomes essential to identify the factors that might affect Internet use. It is also important to identify the relationship between these variables and the strength of each of these factors in predicting the use of Internet when all the other variables are kept constant. This paper identifies the statistical models of use of the Internet in persons with and without disabilities. These linear regression models may assist in future policy development to reduce the digital divide between persons with and without disabilities.
This paper hypothesizes that computer ownership, family income, individual's educational status, race of the person, age of the person, and whether a person lives in the metropolitan area, all affect the use of the Internet at home. This paper also hypothesizes that the linear regression models of Internet usage are statistically different for persons with and without disabilities.
This paper used the data set collected by using the August 2000 Internet and Computer use Supplement of the Current Population Survey (CPS) (3). The CPS nationally surveys approximately 50,000 U.S. households. The sample represents of all fifty states and the District of Columbia (2). The Internet and Computer Use Supplement includes questions on household computer ownership and Internet usage in the form of an Internet connection at home. The data on computer ownership and Internet usage along with demographic questions and disability status questions provided the variables to perform the linear regression analysis. The disability status question identified 2668 individuals who answered that they had some form of work disability. Out of the remaining 132318 cases without an identified disability, a random sample of 2668 records was used to analyze the modeling for persons without a disability. The multivariate regression model was used to find a relationship between the independent variables (computer ownership, family income, educational status, race, age, and whether a person lives in the metropolitan area) and the dependent variable (Internet usage) for both the population with and without a disability. The Chow test of homogeneity between models was used to find the whether the relationships between the independent and the dependent variables are statistically different for individuals with and without a disability thus suggesting the statistical significance of affect of disability on Internet use.
Table 1 summarizes the Internet usage regression models for the persons with and without a disability. Models I and II describe the behavior of the dependent variable, Internet use, for individuals with and without a disability respectively. The combined model portrays the Internet use pattern discounting the disability status. Based on the R2 values, Model I predicts 57.9% of Internet use behavior for individuals with a disability, Model II predicts 59.8% and the combined model predicts 62% of Internet use for individuals in the three groups. The Chow test calculated an F = 6.90 ( p = .01, alpha = .05).
The model of Internet usage is statistically different for persons with disabilities and without disabilities. Regression models indicate that computer ownership, family income, educational status, metropolitan status and race affect Internet usage in both persons with and without disability. The model indicates that providing a person with a disability with a computer at home significantly increases their chances of using the Internet. Further research is needed to identify the factors that influence the patterns of Internet use in individuals with a disability so that better strategies can be implemented to increase digital inclusion. To meet this goal, some variables need to be better defined. The disability variable needs to be defined more specifically so that one gets a better description of the kind disability. Age and Internet use pattern need to be defined more as a continuum rather than as dichotomous variables. As the models 1 and 2 are statistically different, there is high likelihood that the use of assistive technology and human support impact Internet usage in persons with disabilities. Thus, there is also a need to include questions that identify the factors, like Assistive Technology and human support, which enhance the use of Internet at home.
This paper could not have been completed without the support and guidance of Drs. Margo Anderson and Roger O. Smith at the University of Wisconsin-Milwaukee.
Bhagwant S.
Sindhu, MS, OTR
Department of Occupational Therapy - University of Florida
P.O. Box 100164, Gainesville, FL
32610
Voice (352) 846-1018, Fax (352)
846-1021
bsindhu@hp.ufl.edu
Table 1:Summary of Regression Models for Internet Use in the year 2000
a :Indicates significance at a 5 percent confidence level.
b :Indicates significance at a 10 percent confidence level. Non a and b values signify a 1% confidence level
x :t ratios are in the values in the parentheses
: The first values for the constant is the unstandardized coefficient whereas the first value for rest of the variables are the standardized coefficients.