Observation

From Displayr
Jump to: navigation, search

An observation is the entity that is analyzed in a study.

For example:

  • Most surveys of public opinion of adults, or adults that are likely or eligible to vote, have the adults as the observations.
  • A survey of schools might have the school as the observation, with one or more people in the school, providing information about the school.
  • A study investigating how mice react to changes in diet may have the mouse at different points in time as the observation. For example, if there are 30 mice in the experiment, and the mice is observed prior to the change in diet, 1 week after the change in diet, and 2 weeks after the change in diet, then there are 90 observations.

Observations and data sets

Typically, the observation corresponds to a row of the data set being analyzed. This is a requirement for most advanced data analysis techniques.

Changing the observation

A common transformation of a data set is to change the structure, so that the observations change. For example:

  • Aggregation can be used to change a data set of people into a data set of households or countries.
  • Stacking can be used to change the observations from people to people-in-situations.


Common types of observation

Studies of organizations Studies of people
Transaction

Employee
Decision-making unit
Company division
Company
Site
Industry grouping
Number of employee grouping
Turnover grouping (e..g, under $1 million, etc.)

Person
Household
Voter
Family
Decision maker
Transaction
User
Person-by-usage occasion
Job
User
Exchange
Shopping basket
Shopping trip
Income grouping

Also known as

  • Case.
  • Unit.
  • Unit of observation.
  • Analysis unit
  • Analysis subject
  • Element
  • Population unit
  • Unit of analysis

See also

Sampling Unit