Classic data-set from Chapter 2 of Joop Hox’s Multilevel Analysis (2010). The popular dataset included student from different class (i.e., class is the nesting variable). The outcome variable is a self-rated popularity scale. Individual-level (i.e., level 1) predictors are sex, extroversion. Class level (i.e., level 2) predictor is teacher experience.

popular

Format

A data frame with 2000 rows and 6 variables:

pupil

Subject ID

popular

Self-rated popularity scale ranging from 1 to 10

class

the class that students belong to (nesting variable)

extrav

extraversion scale (individual-level)

sex

gender of the student (individual-level)

texp

teacher experience (class-level)