Universe

The universe consists of all survey elements that qualify for inclusion in the research study. The precise definition of the universe for a particular study is set by the research question, which specifies who or what is of interest. The universe may be individuals, groups of people, organizations, or even objects. For example, research about voting in an upcoming election would have a universe comprising all voters. If the research was about political parties' sponsorship of candidates, the research would include all political parties. In research that involves taking a sample of things for testing or review, the universe may include inanimate objects. For example, a researcher may want to determine the extent to which toys made in a particular country contain a certain type of paint. In this case, the universe would include all toys made in that country. Some survey research has a very small universe (e.g., a survey of individuals who have won 1 million dollars or more in the lottery), and some research has a large universe (a survey of the television-viewing habits of all adults in the United States).

Survey practitioners often do not expect to be able to measure the entire universe, because it is neither practical nor necessary. A survey of all the elements in a universe would not be a sample but would instead be a census. Taking a census is rarely practical because it is costly and it rarely is needed; that is, a good survey can gather as accurate (and sometimes more accurate) data about the universe.

Defining the universe is a key element in the design of an appropriate sampling strategy. The researcher must carefully consider precisely who or what is of interest and then how best to contact or include these individuals or things in the research. The importance of appropriately defining the universe should not be underestimated. A universe that has been too narrowly defined will exclude important opinions and attitudes, and a universe too broadly defined will include extraneous information that may bias or affect the overall results.

Once defined, the universe is used by the researcher to structure an appropriate sampling methodology, in particular by defining the target population and choosing a sampling frame or frames. Research being conducted with a universe of all adults who watch television, for example, can contact a random sample of such individuals using random-digit dialing or by sampling clusters of geographies and then sampling households within those geographies using addressed-based sampling or area probability sampling. Sometimes multiple frames are needed to adequately cover the universe. In many surveys, the researcher does not know (nor does she or he need to know in advance) how many individuals or things are qualified to be part of the universe. One way to estimate this information is to retain information from any screener used to identify and select qualified respondents. For example, imagine a researcher wanted to do a survey of all individuals who own both a dog and a cat in Australia. It is unlikely that there is any reliable information on the precise [Page 930]number of Australians who own both types of pets, and it is further unlikely that a listing of such pet owners exists. When screening randomly selected households, the researcher can determine how many of the individuals contacted qualify and can use this information to estimate the total universe size. For example, if the researcher randomly contacts 1,000 households and finds that 45 of these own both a dog and a cat, the researcher can then estimate that roughly between 3% and 6% of all households in Australia own both types of pets.

...