I. Introduction

The Internet has become a marker in time for our historical moment known as the Information Age. One could compare this period to the Space Age during the 60's. The Internet as a symbolic representation is the metaphorical manifestation of the information era. Presently there are an estimated forty-million host computers connected to the Internet. By the year 2000 there will be an estimated two-hundred million people who use the Internet on a regular basis. The demographics of Internet users at present is heavily weighted with white males with a college education and incomes over $50,000. Over 90% of all users of the Internet are white and most reside in the United States. The fastest growing group of new users are women. Among sites on the World Wide Web over sixty percent are commercial in nature.

Concerns over access to the Internet should be noted and discussed if the Internet is in fact a tool necessary for economic success in the 21^st Century. The present disparity among users of the Internet may show us trends which could persist in the future. This paper suggests an analytical model for understanding Internet usage demographics which may help to illuminate the present and future scenarios of Internet demography.

II. Hypothesis

The purpose of this sociological analytic project will be to discover the existence of a relationship between the use of the Internet across race categories among women in the United States. Specifically the analysis will be concerned with Internet usage among women of the similar educational and income categories but different (self-declared) race categorizations.

H1. Internet usage among women of similar education and income categories is higher for that of white women compared to that of women categorized in American minority groups. Usage disparities of the Internet among these groups will echo that of white men of similar educational and income distinctions compared to men in American minority groups of similar income and college educational attainment.

III. Expected Relationships

The expected relationship will be the effect the variable of race has as it affects Internet usage among women. I am particularly interested in how race affects Internet usage among women of similar educational and income categories. Racial disparities of Internet usage among women, who have been a historically marginalized group in the United States, while controlling for income and education may further illuminate the use of the Internet as a race privileged or culturally specific activity of particular groups at this historical moment. Additionally by looking at categories based upon the sex of a group, IE females, who have been lagging in Internet usage compared to overall population percentages of women in the United States may provide insights into the Internet as a gendered and racialized sphere in social-cyber space.

IV. Control Variables.

1. Income Level Per Year.

Income determines one's ability to have access to consumer goods IE. Computer technology and as well is a marker which places those within certain income categories in professional occupations in which the use of computer technology is generally incumbent. Using income as a control variable remove and as well adds to our analysis economic considerations concerning access to the Internet and provides us with more clarity around factors affecting Internet usage as it relates to the issue of race among women. Put another way by comparing women of diverse race or ethnic backgrounds and of the same income categories we can theorize about race or ethnicity as a marker of Internet usage in controlling for income.

2. Educational Attainment.

The attainment of a college education is seen as a great equalizer in terms of social class and social mobility in the United States. Individuals who have received a college degree tend to inhabit the more mainstream areas of society in the United States. Measuring women as a group who have completed a college education will enable us to determine from this categorization the disparity of Internet usage among those with a college education. Educational attainment as a barrier or factor concerning Internet usage will further illuminate Internet usage as a race privileged or selective activity among those of diverse race categories by looking at those of similar educational backgrounds.

V. Operational Definitions of Variables.

For each of the following questions circle the number of the category which reflects characteristics about you. This survey is for female respondents only.

1. Dependent Variable.

A. Internet Access and Usage:

Do you personally have access to the Internet at either home, office or school and use it at least two or more hours per week?

1.YES

2.NO

2. Independent Variable.

A. Race or Ethnicity

What ethnic or race category would best describe you?

1. African-American

2. Hispanic-American

3. White

4. Asian-American

5. Other

3. Control Variables.

A. Income Level.

What was your individual income over the last year before taxes?

1. 0-$14,999

2. $15,000 -$30,999

3. $31,000-$49,999

4. $50,000 or over

B. College Educational Attainment.

Do you have a college degree?

1. YES

2. NO

VI. Statistical Level of Measurement

A. Appropriate Measure of Central Tendency.

The Variables:

1. Race or Ethnic Group (Nominal Level Variable)

2. Income Level. (As an Interval/Ratio variable the appropriate measure of Central Tendency is the Standard Deviation).

3. Educational Attainment. (For the purposes of this survey our aim was to delineate those with a college education therefore a yes/no or Nominal answer was suitable for our data).

4. Internet access and usage (Access to the Internet and usage thereof was our primary question and only required a yes/no answer or Nominal variable).

.

The variables number one, three and four are nominal level variables and thus the measure of a central tendency applicable is the mode. This variable is a descriptive variable which cannot be ranked or ordered. At the nominal level variables has neither rank, distance nor a true zero point therefore the only method of comparison available to us is to measure and compare the number of cases in each category and use the measure of a central tendency known as the mode. The mode tells us the category with the most number of cases which for the purposes of this statistical analysis is the information which we need in order to complete our analysis.

The Income Level Variable is an interval/ratio variable as it has a true zero point and a measurable distance between points on the scale. Therefor the appropriate measure of Central Tendency we will use is the Mean. At the Interval/Ratio level of measurement we have equal distance between the units and an absolute zero point. To interpret the Income Level variable we find the average of all the cases or the Mean of the income level of all the respondents. From this statistical tool we can determine how the Mean income relates to Internet usage among the respondents in diverse ethnic and race categories and as well use other statistical tools to interpret the mean.

B. Measures of Dispersion

The Variables:

1. Race or Ethnic Group. (Nominal/ Based on descriptive characteristic of race or ethnic background).

2. Income Level. (Interval/Ratio, Based upon categories of income levels with a true zero point and measurable distances between categories).

3. Educational Attainment. (Nominal/For the purposes of this survey our aim was to delineate those with a college education therefore a yes/no answer was suitable for our data).

4. Internet access and usage (Nominal/Access to the Internet and usage thereof was our primary question and only required a yes/no answer).

Indexes of Qualitative Variation (IQV) is the appropriate measure of dispersion for variables at the nominal level, of which all of our variables except Income Level, are at the nominal level. We could use range however due to the small number of categories there would be no meaningful findings which would enable us further in our statistical analysis. Therefore (IQV) will be the statistical tool of choice for informing our measure of dispersion.

Standard Deviation. Income Level as an interval/ratio variable signifie the appropriate Measure of Dispersion as the Standard Deviation. In using the Standard Deviation we can see how scores are spread around the mean. The mean distribution gives us the average of all scores and the Standard Deviation informs the distribution of scores around the Mean score. From this statistical analysis we can determine income levels as it relates to Internet usage and see where the highest concentration of Internet usage occurs related to Income Levels.

C. The Frequency Distribution Shape

The Variables:

1. Race or Ethnic Group. (Nominal/ Based on descriptive characteristic of race or ethnic background).

2. Income Level. (Interval/Ratio, Based upon categories of income levels with a true zero point and measurable distances between categories).

3. Educational Attainment. (Nominal/For the purposes of this survey our aim was to delineate those with a college education therefore a yes/no answer was suitable for our data).

4. Internet access and usage (Nominal/Access to the Internet and usage thereof was our primary question and only required a yes/no answer).

A. Race or Ethnic Classifications

The majority of respondents will fall in category three (White). Answer three (White) will be the mode for this survey question. The total of all other respondents not affirming category three will not equal the total of category three. This prediction is based upon normal distributions of American minority groups according to United States Census population data. We expect those survey respondents will reflect these numbers.

B. Income Level Variable.

Income Levels will have a mode of category three or the $31,000 to $49,999 range. Therefore the shape of the Frequency Distribution will have a peak just past the middle of the ranges of income with a quick drop off occurring when we reach the $50,000 or over a category. The Income Levels of the first two categories will have a number of respondents which reflected in a Distribution Curve will be a slow steady rise to the peak of the third category. This prediction is based upon Census Data which shows the Mean income in the United States to be around $30,000 with a large drop of in population amounts above the per capita income of $50,000 per year.

C. Internet Usage

This variable has only two possible responses. I predict the category of a negative response to this question will be the modal category. Predicting the usage rates of these sample populations will be reflective of other Internet User Surveys I have studied. In doing cross tabulations of Internet Usage those respondents who are white with incomes in the third category $31,000 to $50,000 and possessing a college degree will occupy the mode category for Internet Usage. Frequency Distribution shapes will be predicated upon the cross-tabulations we run using different variables which will naturally affect the shape of the frequency distribution charts. In a bar graph of Internet users all white respondents will occupy the highest level on a bar graph.

D. Educational Attainment

I am unwilling to predict the mode of this category for respondents as a whole. However as indicated in the previous discussion of Internet Usage I expect when cross tabulating results with income and race we will see a similar distribution of white female respondents with incomes of $31,000 to $49,999 occupying the modal category of educational attainment as an affirmative to obtainment of a college degree. As a group proportionally represented women in American minority groups will have fewer affirmative responses to attainment of a college education.

VII. Measure of Association

The hypothesis contains two variables, Internet usage and race at the nominal level of measurement. In examining this hypothesis a measure of association between the two variables under consideration is warranted. The independent variable, race and the dependent variable, Internet usages being both nominal level variables dictate the use of Lambda as a measure of association. Lambda indicates the strength and direction of the relationship between the two variables under consideration. In using Lambda we can find the number of errors predicted while disregarding the independent variable. Additionally we can predict the number of errors while taking into account the independent variable. PRE will also be used in order to inform us about the proportional reduction in errors which are achieved when shifting form one prediction rule to another. Predictions about the measure of association between the independent and dependent variables will show a strong realtionship between race and Internet usage.

Introduction of the control variable income level will affect the outcome of the relationship between the independent variable, Internet usage and the dependent variable race or ethnic background. The bivariate table will generate a strong relationship between non-Internet usage and race or ethnic background being that of an American minority group. Introduction of income level will decrease the strength of this relationship in the higher income brackets among women in American minority groups correlated with Internet usage. Or to put it more concisely women in American minority groups with higher incomes will have a stronger association with an affirmative answer to Internet usage. However I predict that white women with higher incomes will show a higher measure of association concerning Internet usage than the same association with women in American minority groups. Patterns of association are used to discover interpretation patterns about variables under discovery.

Our control variable of college educational attainment will potentially generate a positive association between college educational attainment and Internet usage among all categories of respondents. This control variable will weaken the association between the Independent and Dependent variables as it shows a strong association between college educational attainment and Internet usage. However I predict that the association between Internet usage and race will be stronger than that of the measurement association factor of college educational attainment and Internet usage. Or white women with a college education compared with women in American minority groups with a college education will show a stronger association with Internet usage. Although the association among women in American minority groups will be higher concerning Internet usage than those without a college education the associaton will be stonger among college educated white women.

VIII. Conclusion

The population of Internet users is rapidly expanding and changing. As with other high-tech consumer goods the price of computers continues to decrease. This change and the resultant main streaming affect this will have on Internet usage will potentially and significantly change the population parameters of Internet users. Therefor the results of this survey and analysis are only a snapshot of Internet users at this historical moment and further surveys should be conducted to monitor the changes among the cohort of Internet users.

The dominant use patterns of the Internet by the white college educated middle-class populations could have significant impact on future social inequalities if access to and use of the Internet is in actuality a function necessary for attainment and maintenance of a particular social class. Some have spoken of a rift between populations of the information haves and the have-nots. Society is moving to a more information-based society as opposed to a manufacturing society and the failure of certain groups to gain access to this resource base may further excaserbate social inequalities around race, gender and class. Therefore efforts should be made through public education and public Internet access to insure that all citizens of the United States who wish to gain access to the Internet are able to do so.

This survey shares problems typical of other surveys. Reliability and truthfulness of the respondents are a concern when discussing social class issues such as income levels and college educational attainment. Therefore, a large sample would need to be gathered to reduce the level of error due to misrepresentations. To check for survey accuracy comparisons could be made with United States Census data and disparities or similarities could be noted and discussed.

Further surveys about Internet usage may want to differentiate between Internet usage and access at the home as compared to Internet usage at one's place of employment or school. While benefits of access can be realized by having access to the Internet in general, home access signals a particular privilege which may show up as a race or class privilege. An analysis of the particular place and types of Internet usage will aid in our understanding of how the Internet is utilized and by whom. Further delineations for statistical analysis could be added in the variable of Educational Attainment such as specific questions about levels of educational attainment such as degree attainments. The variable of sex could be added to compare and cross tabulate across sex boundaries.

Questions and analysis around Internet usage will continue to aid in our understanding of who and how the Internet is utilized