Sociolinguistics 4, Exercise 2: Labov 1966

Findings as the source gives them: lower-class speakers rarely pronounced
postvocalic /r/ in casual speech, while middle-class speakers shifted toward
the higher-status form when paying attention.

Why this supports the attention-to-speech model: Labov proposed that styles
lie on one dimension measured by the amount of attention paid to speech.
Casual speech gets the least self-monitoring, and the shift toward the
higher-status /r/ appears when speakers pay more attention.

A criticism from the source: attention is hard to quantify, and the model
implies one style per formality level (Hindle, 1979).