peers

An important question is who are your peers? Close friends are the most proximal and influential peer group

(Halliday & Kwak, 2012; Prentice & Miller, 1993). Most of the previous research used “typical/average student” as a ref-psychological reality
erence group that lacks  for the respondents because they have in mind a group of real people, their

close friends (Prentice & Miller, 1993), which are more predictive of individual behavior (Perkins, 2014). In this paper,
we use a large nationally representative sample of U.S. middle and high school adolescents and include information on
the actual peers (nominated friends) who were directly asked about their substance use. Our data are from the school
social networks of friends in the National Longitudinal Survey of Adolescent to Adult Health (Add Health). From it, we
estimate the effect of misperception about the rates of friends’ substance use on own use by adolescents in Grades 7–12.

We utilize “any use” as the measure of perceived and actual substance use. This helps alleviate possible social desir-
ability bias due to stigma and recall biases in the self‐reports of substance use, which are less likely to be a concern when

focusing on the extensive rather than the intensive margin of substance use (Johnson, 2014). For example, for the

intensive margin, memory impairment that was found in heavy and frequent drug users was implicated in the inaccu-
rate and inconsistent reporting of substance use in surveys. Also, given the illicit and socially undesirable nature of

substance use behaviors, they tend to be stigmatized and the perceived stigma was found to be associated with
underreporting of such behaviors (for the review of literature, see Johnson, 2014).
For the extensive margin decision, we include all three of the high use substances by adolescents—alcohol, smoking,
and marijuana—(Johnston et al., 2018) which are also the most amenable to peer influence (Clark & Loheac, 2007).
Despite the fact that substance use rises sharply during teen years and is heavily influenced by school peers, little
research has utilized data on school‐aged adolescents (D’Amico & McCarthy, 2006; Juvonen, Martino, Ellickson, &
Longshore, 2007) with the previous research almost exclusively focusing on college students (Borsari & Carey, 2003;
Perkins, 2014).