Recognizing and preserving paired observations

In bivariate data, two variables are recorded for each observational unit. A paired observation keeps both values attached to the same unit, allowing questions about their joint pattern. The variables may be numerical, categorical, or one of each; the defining feature is the pairing.

Recognize the record

First identify the entity described by one complete row, such as a tree whose age and canopy width are recorded together. Age and width are numerical variables; a tree species recorded alongside age would instead be a categorical variable. Each row still represents one tree, regardless of whether its two values are numbers or labels.


A shared identifier can link measurements kept in separate files. Known common row order can also preserve the match when the lists refer to the same units in the same order. Equal length alone does not establish a record link; when both variables vary, multiple value pairings may fit the same lists.

Preserve what the pair tells you

The separate lists retain each variable's marginal distribution, such as which ages occur and how often. Separate lists generally do not determine the joint distribution: the counts or values for specific combinations, because multiple pairings may fit the same marginal values.


If one variable is constant, however, the value pairs can be uniquely determined even without unit links.


When multiple pairings fit, the lists alone do not justify a particular claim about an association between the two measurements. Even a genuine association does not establish that one variable causes the other.

Worked example

Suppose three trees have records A: age 5 years and width 1.8 meters; B: age 3 years and width 2.1 meters; C: age 7 years and width 1.2 meters. These are three paired observations, not six trees.


If the age column and width column are each sorted upward on their own, the first displayed pair becomes 3 years and 1.2 meters. That pair did not occur in the original records.


Sorting complete rows by age instead keeps each tree's age, width, and identifier together. The row count stays three under either sorting method, but only whole-row sorting preserves the observed matches.



A survey records each worker’s shift category and minutes of travel to work in the same row. What is the observational unit in this bivariate data set?

Choose an answer

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors

A clinic recorded ages of 20 and 60 years and blood pressures of 110 and 150 mm Hg for two patients. The ages and pressures survive as separate lists, but all patient identifiers and original row order were lost. Which claim is justified?

Choose an answer

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors

Researchers record a bird’s nest type and whether a parasite is present for each of 42 birds. Which description of the recorded structure is accurate?

Choose an answer

Create a free account to get started

Unlock full course notes by signing up

  • Video lessons taught by top educators
  • Practice questions with guided solutions
  • Written review notes for reference
  • Expert answers anytime from course tutors