Date: Fri, Aug 14, 1998  5:03 PM

Survey Sampling Methodology
for Cross-County MetroLink Extension
Public Opinion Survey of July, 1998

Since the survey results were announced and posted on the Council's web page, several people have asked for additional information regarding the ways individuals were selected to participate in the survey.  To provide answers, this brief paper on sampling techniques was prepared.

Explanation of Sampling

Sampling is the process by which individuals are selected from a population for participation in a survey or poll. Through the process of selecting participants, pollsters seek to collect information from a representative group of individuals, and use this information to better understand the characteristics of a larger population

Defining the Population & Polling Area

For the CrossCounty survey, we were trying to get a sense of the perceptions and opinions of people living relatively close to the alignment of the proposed extension. The first step in the process was to decide on an appropriate number of people to contact. For the CrossCounty survey,

500 people were surveyed, including 100 respondents each from the City of St. Louis, University City, and Clayton, 101 from Maplewood, 49 from Richmond Heights, and 50 from Brentwood.  These selections by community were made using Council population estimates for residential areas in close proximity to the extension. The number of participants resulted in an overall margin of error for the survey of +/ 4.4%.

Random Calling

Once the boundaries of the phoning area were defined and the number of calls selected, a random sample of phone numbers from the area was collected. Phone numbers were drawn at random from a directory that lists phone numbers by street address. Only addresses within the predefined geographic area were collected. These numbers were then organized into groups of fives.

The phoning then proceeded as follows: A phoner is given a sheet with about ten groups of five numbers. They go to the first group, start at the top, and dial. If they get an answer, they conduct the interview, and then cross the rest of the five numbers off the list. (Only one person in each group of five is interviewed.) If there's no answer, they make a note of the time they called and move to the next number.  After four calls to each person in the group of five, the group of five is discarded. No household receives more than four calls.)  If a person is called and a busy signal is received, the next person is called. If an interview is conducted with this person, the number with the busy signal is not called again.

Sampling Goals

The purpose of the random sampling process is to ensure that an appropriate geographic spread is achieved. The selection process is designed to preclude the possibility of contacting individuals in only a small portion of the survey area. Aside from geographic considerations, information regarding respondents is not known by the pollsters until the calls are placed.

Steven W. Schukraft, Engagement Team Manager
c/o CrossCounty Conceptual Design Study
701 N 15th Street, Suite 1001E, Tenth Floor
St. Louis, MO 63103
email crosscounty@marymeans.com
hotline >(314) 621-4499