Data
This report is based on the data from three telephone surveys conducted by the Pew Research Center’s Internet & American Life Project.
The first data set comes from telephone interviews conducted between December 28, 2009 and January 19, 2010, among a sample of 2,259 adults, 18 and older. For results based on the total sample, one can say with 95% confidence that the error attributable to sampling and other random effects is plus or minus 2.3 percentage points. For results based on internet users (n=1,675) or “online news users” (n= 1,582), the margin of sampling error is plus or minus 2.7 percentage points. In addition to sampling error, question wording and practical difficulties in conducting telephone surveys may introduce some error or bias into the findings of opinion polls. This survey was conducted on landline telephones (N=1,697) and cell phones (N=562) and is meant to be representative of all adults in the continental United States.1
The second data set comes from telephone interviews conducted between April 29 and May 30, 2010, among a sample of 2,252 adults, age 18 and older. Interviews were conducted in English. For results based on the total sample, one can say with 95% confidence that the sample margin of error is plus or minus 2.4 percentage points. For results based Internet users (n=1,756), the margin of sampling error is plus or minus 2.7 percentage points.2
The most recent data come from telephone interviews conducted by Princeton Survey Research International between August 9 and September 13, 2010. The survey was administered to a sample of 3,001 adults, ages 18 and older, using a combination of landline and cellular. Interviews were conducted in English or Spanish. The sample margin of error is plus or minus 2.5 percentage points and plus or minus 2.9 percentage points for just Internet users (n=2,065).3
Sources
1. Pew Internet & American Life Project Survey, Health, December 28, 2009 and January 19, 2010. Available at https://legacy.pewresearch.org/internet/Shared-Content/Data-Sets/2010/January-2010–Online-News.aspx [link]
2. Pew Internet & American Life Project Survey, Cell Phones, April 29 and May 30, 2010. Available at https://legacy.pewresearch.org/internet/Shared-Content/Data-Sets/2010/May-2010–Cell-Phones.aspx [link]
3. Pew Internet & American Life Project Survey, Health, August 9 and September 13, 2010.
4. Income, Poverty, and Health Insurance Coverage in the United States: 2009, September 2010. Available at http://www.census.gov/prod/2010pubs/p60-238.pdf
5. Pew Internet & American Life Project Survey, Reputation Management, April 29 and May 30, 2 August 18 – September 14, 2009. Available at https://legacy.pewresearch.org/internet/Shared-Content/Data-Sets/2009/September-2009–Reputation-Management.aspx [link]
Appendix
For investigating the control variables, we used the crosstabs procedure, which offers tests of independence and measures of association and agreement for nominal and ordinal data, and the chi square test to measure statistical significance among groups.
The chi-square test measures the discrepancy between the observed cell counts and what one would expect if the rows and columns in the cross tab were unrelated.
We investigated the controlling factors of community type (rural, suburban, urban), education (some high school, high school, some college, college graduate), race (White, African-American, Hispanic, Other), gender, and age (divided into generational groups of Generation Y (ages 18 -33), Generation X (ages 34-45), Trailing Boomers (ages 46-55), Leading Boomers (ages 56-64), Matures (ages 65-73), and After Work (age 74+) as layering effects in the cross tab analysis and chi-square analysis.
The use of layering effects allowed us to use the chi-square test for determining whether there is a relationship among income group, email or internet usage, and the specific laying factor.
For determining the strength of the relationship, we used the symmetric measures of Phi, Cramer’s V, and the Contingency Coefficient. In addition for testing of significant, by convention, we look for a value of 0.3 or higher to indicate strong, practical effect if the relationship was statistically significant.
