Surveys are conducted to elicit information about attitudes or prevalence of some characteristic of some defined population. The population may be residents of some geographic area, owners of a particular make of an automobile, units of some product coming off an assembly line, employees of some specific company, or any other defined set of people or things. The survey sponsor may be interested in the reaction of citizens to some proposed legislation, the proportion of recent vehicle purchasers who are satisfied with their choice of a car, the percentage of sales of a consumer product to teen-agers, or any of countless other examples.

We are always mindful that a survey must be valid in order that the data derived from it will be useful and represent the phenomenon being measured. A valid survey has several characteristics:

• The "sampling frame" must represent the population of interest
• Probability sampling must be used
• The size of the sample must be sufficiently large
• Questions must be worded to avoid bias
• Non-response must be kept to manageable limits

To illustrate the variety of the many surveys we have designed and conducted, a few representative examples are described below:

• A large, multistage sample of absentee voters and election officials was drawn for a survey to assess the success of the Absentee Voter Assistance Act. The sample included members of the five armed services as well as expatriate Americans registered with American embassies abroad.  The survey included many common questions, but a content analysis of open-ended questions on respondent experience was most informative.

• A unique design was applied to obtain a sample of key, early-reporting precincts whose vote results could be used to help project the winner of presidential, gubernatorial and senatorial elections on Election Night for one of the major TV networks.

• A selection of courier routes was selected to determine the interstate character of the materials carried by the couriers. The object of the survey was to determine whether the employees of the courier companies were subject to federal or state labor regulations as part of a case in Federal District Court.

• A determination of the time spent by meat packer employees in donning protective clothing and obtaining their tools before clocking in and the time spent in cleaning up after the work shift. Distinctions were drawn among job classifications and shifts.

• To help determine whether a minimum utilization rate (MUR) should be a condition for Medicare certification, an on-site survey of a sample of kidney dialysis centers was made to measure characteristics relating to quality and cost.