# Methodology, Pew Research Center

## Methodology

### Summary

The September 2005 Tracking Survey (Online Dating Extension), sponsored by the Pew Internet and American Life Project, obtained telephone interviews with a nationally representative sample of Trio,215 adults living te continental United States telephone households. The survey wasgoed conducted by Princeton Survey Research International. The interviews were conducted ter English by Princeton Gegevens Source, LLC from September 14 to December 8, 2005. This report is based on the findings of a daily tracking survey of Americans’ use of the Internet. Statistical results are weighted to keurig known demographic discrepancies. The margin of sampling error for the accomplish set of weighted gegevens is ±,1.9%. For results based Internet users (n=Two,252), the margin of sampling error is plus or minus Two.Trio percentage points. Online daters (n=204) have a margin of sampling error of plus or minus 7.Five percentage points.

Details on the vormgeving, execution, and analysis of the survey are discussed below.

### Vormgeving and Gegevens Collection Procedures

The sample wasgoed designed to represent all continental U.S. telephone households. The telephone sample wasgoed provided by Survey Sampling International, LLC (SSI) according to PSRAI specifications. The sample wasgoed drawn using standard *list-assisted random digit dialing* (RDD) methodology. *Active blocks* of telephone numbers (area code + exchange + two-digit block number) that contained three or more residential directory listings were selected with probabilities ter proportion to their share of listed telephone households, after selection two more digits were added randomly to finish the number. This method ensures coverage of every assigned phone number regardless of whether that number is directory listed, purposely unlisted, or too fresh to be listed. After selection, the numbers were compared against business directories and matching numbers purged.

### Voeling Procedures

Interviews were conducted from September 14 to December 8, 2005. Spil many spil Ten attempts were made to voeling every sampled telephone number. Sample wasgoed released for interviewing ter replicates, which are representative subsamples of the larger sample. Using replicates to control the release of sample ensures that accomplish call procedures are followed for the entire sample.

Calls were staggered overheen times of day and days of the week to maximize the chance of making voeling with potential respondents. Each household received at least one daytime call ter an attempt to find someone at huis. Te each contacted household, interviewers asked to speak with the youngest adult masculine presently at huis. If no masculine wasgoed available, interviewers asked to speak with the oldest female at huis. This systematic respondent selection technology has bot shown to produce samples that closely mirror the population te terms of age and gender.

### Weighting and Analysis

Weighting is generally used ter survey analysis to compensate for patterns of nonresponse that might bias results 17 . The weight variable balances the interviewed sample of all adults to match national parameters for hook-up, age, education, wedloop, Hispanic origin, region (U.S. Census definitions), and population density. The White, non-Hispanic subgroup wasgoed also balanced on age, education and region. Thesis parameters came from a special analysis of the Census Kantoor&lsquo,s 2004 Annual Social and Economic Supplement (ASEC) that included all households ter the continental United States that had a telephone.

Weighting wasgoed accomplished using Sample Balancing, a special iterative sample weighting program that at the same time balances the distributions of all variables using a statistical technology called the *Deming Algorithm*. Weights were trimmed to prevent individual interviews from having too much influence on the final results. The use of thesis weights te statistical analysis ensures that the demographic characteristics of the sample closely approximate the demographic characteristics of the national population. Table 1 compares weighted and unweighted sample distributions to population parameters.

### Effects of Sample Vormgeving on Statistical Inference

Post-data collection statistical adjustments require analysis procedures that reflect departures from elementary random sampling. PSRAI calculates the effects of thesis vormgeving features so that an suitable adjustment can be incorporated into tests of statistical significance when using thesis gegevens. The so-called “design effect” or deff represents the loss ter statistical efficiency that results from systematic non-response. The total sample vormgeving effect for this survey is 1.22.

PSRAI calculates the composite vormgeving effect for a sample of size n, with each case having a weight, wi, spil:

Ter a broad range of situations, the adjusted standard error of a statistic should be calculated by multiplying the usual formula by the square root of the vormgeving effect (&radic,deff). Thus, the formula for computing the 95% confidence interval around a percentage is:

where p&circ, is the sample estimate and n is the unweighted number of sample cases ter the group being considered.

The survey&rsquo,s margin of error is the largest 95% confidence interval for any estimated proportion based on the total sample&mdash, the one around 50%. For example, the margin of error for the entire sample is ±,1.9%. This means that te 95 out every 100 samples drawn using the same methodology, estimated proportions based on the entire sample will be no more than 1.9 percentage points away from their true values te the population. It is significant to reminisce that sampling fluctuations are only one possible source of error te a survey estimate. Other sources, such spil respondent selection bias, questionnaire wording and reporting inaccuracy, may contribute extra error of greater or lesser magnitude.

### Response Rate

Following is the utter disposition of all sampled telephone numbers:

Table Two reports the disposition of all sampled telephone numbers everzwijn dialed from the flamante telephone number sample. The response rate estimates the fraction of all eligible respondents te the sample that were ultimately interviewed. At PSRAI it is calculated by taking the product of three component rates: Legitimate

- Voeling rate &ndash, the proportion of working numbers where a request for vraaggesprek wasgoed made &ndash, of 72 procent Nineteen
- Cooperation rate &ndash, the proportion of contacted numbers where a consent for vraaggesprek wasgoed at least originally obtained, contra those refused &ndash, of 44 procent
- Completion rate &ndash, the proportion of primarily cooperating and eligible interviews that were ended &ndash, of 90 procent

Thus the response rate for this survey wasgoed 28 procent.