Behavioral Research

Table of Contents
1 General Definition
2 Use of the Construct in Health Behavior Theories

Measures and Measurements


Similar Constructs


Measurement and Methodological Issues




Measures Appendix

8 Published Examples

Download Full Text (PDF)

Other Constructs



Dispositional Optimism




Illness Representations

  Implementation Intentions
  Intention, Expectation, and Willingness
  Normative Beliefs
  Optimistic Bias
  Perceived Benefits
  Perceived Control
  Perceived Severity
  Perceived Vulnerability
  Self-Reported Behavior
  Social Influence
  Social Support

Social Influence
Thomas A. Wills, Michael G. Ainette, and Carmella Walker

<< Previous


Measures and Measurements

In this section we describe measures that have been utilized to index different aspects of social influence. Because many of these are single-item measures we do not detailed data about reliability. Validity of these measures has been demonstrated in longitudinal studies through demonstrating their ability to predict initiation and escalation of tobacco/alcohol use and other problem behaviors.

Social Modeling Measures. Many studies have used items that index smoking, alcohol use, and other behaviors by social network members. Measures of peer substance use typically ask: "How many of your friends smoke cigarettes / drink alcohol (beer, wine, liquor, or wine coolers) / use marijuana?" Answers are on ordinal scales having response points from None of My Friends to Four or More of My Friends (Wills & Cleary, 1999). Measures of tobacco and alcohol use by parents may have a dichotomous structure, for example "Does your mother/father smoke cigarettes?" with response Yes/No. Alternatives ask about parental substance use with numerical scales to index frequency of use (e.g., "During the last month how often did your father drink beer?" with responses Never to Three or More Times a Week. Measures of sibling use typically ask about sib's tobacco, alcohol, or other drug use during the past year on a frequency scale with response points such as Never to Five or More Times (Pomery, Gibbons, Gerrard, Cleveland, Brody, & Wills, 2005). These are presumed to represent modeling influences. Prospective studies consistently show items on peer or parental substance use to predict initiation and escalation of use among adolescents (e.g., Gibbons, Gerrard, Cleveland, Wills, & Brody, 2004; Gibbons, Gerrard, Vande Lune, Wills, Brody, & Conger, 2004; Wills & Cleary, 1999). There is less evidence on reliability of the single items used to measure modeling. Studies that have obtained measures of smoking from peers themselves have found that adolescents' smoking is correlated more highly with perceptions of peer smoking than with direct peer reports of their own smoking, though significant relations are found in both cases (Ennett & Bauman, 1994; Urberg, Shyu, & Liang, 1990). Thus it is desirable to obtain independent reports of parent and peer substance use when the research context makes this feasible.

Social Pressure Measures. While measures indicating that friends smoke have been presumed to be reflected in peer pressure for use, there is actually little direct evidence on this point. Urberg et al. (1990) developed a measure of explicit pressure with items that asked "Have you felt pressure to smoke cigarettes?" and "Have you felt pressure not to smoke?" with 4-point response scales (Never to Often). Variant items ask about whether a person has recently been offered cigarettes or other substances, for example "How many times have you been offered a drink in the past month?" with a write-in response (Graham, Marks, & Hansen, 1991). Data from explicit pressure items indicate that participants tend to report not having experienced much social pressure to smoke (Sussman, Hahn, Dent, & Stacy, 1993; Urberg et al., 1990), and in multivariate analyses these measures usually do not have significant relations to adolescent smoking (Urberg et al., 1990). Thus reviewers have concluded that perceived normative pressure from observed peer use may be more influential than explicit social pressure (Kobus, 2003). Still, it may be desirable to include these kinds of items in research for descriptive purposes to characterize the type of level of social pressure occurring in a given population.

Social Norm and Consensus Measures. Measures of normative influence assess the perceived prevalence of use and perceived approval for use in a respondent's social environment. A typical measure of perceived prevalence (consensus perception) provides the respondent with a ratio scale (0-10 scale or 100-point line) and asks: "What would you say is the percentage of kids in the XXth grade at your school who smoke cigarettes regularly (at least a few a week)?" The respondent then checks a point on the line to indicate his/her perception of the percentage of schoolmates who are smokers, drinkers, etc. Prospective studies show that such measures predict onset of smoking and alcohol use, controlling for other variables (Graham, Marks, & Hansen, 1992; Sussman et al., 1988). Measures of social norms ask about perceived reactions of network members to smoking or alcohol use (Appendix, Section 1) or ask the participant (plus parents, peers, etc.) how acceptable it would be for a youth to use tobacco or alcohol in various situations (Appendix, Section 2). Studies have indicated that measures of perceived norms show significant relations to adolescents' smoking and alcohol use controlling for actual peer use (Brody, et. al., 1998; Urberg et al., 1990). Findings on normative influence and consensus perceptions have led to the suggestion that peer use operates as a "silent influence" on youth smoking, because adolescents may be influenced by their perception that smoking is common and/or approach among age-mates (Kobus, 2003). Ironically these perceptions tend to be mistaken, as adolescents typically overestimate the prevalence of substance use among teens and underestimate how common negative attitudes about smoking and alcohol use actually are in the school population (Prentice & Miller, 1993; Sussman et al., 1988). Correcting erroneous perceptions about prevalence and normative acceptability has in fact proved to be useful in prevention programs (Gerrard et al., 2002, 2003, 2006; Hansen & Graham, 1991; Schroeder & Prentice, 1998).

Social Perception Measures: Adolescent "Crowds". Measures of recognized adolescent subgroups or "crowds" have been used in several studies. In these measures the adolescent is provided with a list of recognized crowds in the school and is asked which one he/she identifies with. For example Sussman et al. (1994) used a measure that stated, "People often hang out in different groups at school. Please circle the one group below which most closely matches the group you belong to." Studies have found reasonable replication across geographic areas in identifying groups labeled "freaks/stoners/dirtballs," "jocks/athletes," "hotshots" (brains, socials, or populars), and "regulars" or ordinary students, with an additional group termed "skaters/surfers" found in West Coast samples (Mosbach & Leventhal, 1988; Sussman, Dent, Stacy, Burciaga, Raynor, Turner, Charlin, Craig, Hansen, Burton, & Flay, 1990; Sussman, Simon, Stacy, Dent, Ritt, Kipke, Montgomery, Burton, & Flay). Measures of group identification show significant associations with current substance use, with the most use among freaks and skaters, and longitudinal studies show that group self-identifications predict onset and escalation of substance use (Sussman et al. (1994)).

Social Perception Measures: Prototypes of Users. Measures on prototypes of substance users assess the respondent's perception of the qualities of the typical person his/her age who smokes cigarettes, drinks alcohol, or engages in other problem behaviors. The measure gives an initial instruction to imagine a typical (same age/same sex) user and then rate the characteristics of the typical user on a set of adjective descriptors (Appendix, Section 3). A number of studies with adolescents have shown that while perceptions of users tend to be somewhat negative, persons with relatively more favorable perceptions of users are more likely to smoke or drink (e.g., Blanton et al., 1997; Gibbons et al., 1995). Prototype measures for various aspects of sexual behavior are in Blanton et al. (2001), Gibbons, Gerrard, and Boney-McCoy (1995), and Wills et al. (2003). It is important to note that prototypes for substance or sex abstainers, asking about perceptions of the typical teen who doesn't smoke, drink, etc., show a significant protective effect (Gerrard et al., 2002; Wills et al., 2003; Wills, Murry, Brody, Gibbons, & Gerrard, 2007).

Social Communication Measures. Communication measures tap the frequency and quality of communication between parents and youth about tobacco/alcohol use and other problem behaviors. Measures on the frequency of communication across a broad spectrum of behaviors are in the Appendix, Section 4A; this can be modified as necessary by selecting items that focus on a specific behavior. Frequency items may be combined when appropriate with items on quality of communication (Appendix, Section 4B). Measures of communication have been shown to correlate with children's norms about substance use and their level of risk behavior (Brody, Flor, Hollett-Wright, McCoy, & Donovan, 1999; Whitaker & Miller, 2000; Wills et al., 2003).

Media Exposure Measures. Measures on advertising exposure typically are single items asking about how often, in a recent time frame, the respondent has seen advertising about a given topic such as cigarette smoking (see Wakefield et al., 2003); an additional dimension is tapped through asking the respondent about his/her reactions to such items. Examples are included in the Appendix, Section 5. Measures of exposure to smoking and alcohol use in movies are more complicated. These involve first determining which movies (from a list of 50 or more) a respondent has seen, coding the amount of smoking/alcohol use in each movie through records made by trained coders, and then determining the total amount of movie smoking or alcohol use a given respondent has been exposed to (see Sargent et al., 2001). A variant measure involves asking the respondent about his/her favorite movie star and then determining exposure to smoking by the star in the movies the respondent has seen (see Distefan, Gilpin Sargent, & Pierce, 1999, 2004). Measures of exposure to advertising and movie smoking/alcohol use have been linked in several studies to onset of smoking and drinking among adolescents (Dalton, Sargent, Beach, Titus-Ernstoff, Gibson, Ahrens, Tickle, & Heatherton, 2003; Henriksen, Feighery, Wang, & Fortmann, 2004; Pierce et al., 1994; Sargent , Beach, Adachi-Meija, Gibson, Titus-Ernstoff, Carusi, Swain, Heatherton, & Dalton, 2005; Sargent, et al., 2006; Tickle, Sargent, Dalton, Beach, & Heatherton, 2001).

Neighborhood Context Measures. Measures of neighborhood context as utilized in previous research reflect several dimensions. Some neighborhood measures are obtained from census data through determining parameters such as average income, residential instability, or percent of families on welfare at the block-group level (Brody, Ge, Conger, Gibbons, Murry, Gerrard, & Simons, 2001). Measures derived from Sampson's theory of collective efficacy (Sampson, Raudenbush, & Earls, 1997) reflect dimensions of collective socialization, interpersonal trust and cohesion, and neighborhood characteristics. These are presented in the Appendix, Sections 6A-6C. Specific measures for school environments ask about attributes such as organization and discipline (Appendix, Section 6D). Measures of neighborhood and school environments have been found to be related to adolescent substance use, and/or to act as a moderator of other social influence measures, in several studies (Gibbons et al., 2004; Novak & Clayton, 2001).

<< Previous

Search | Help | Contact Us | Accessibility | Privacy Policy

DCCPSNational Cancer Institute Department of Health and Human Services National Institutes of Health

Health Behavior Constructs: Theory, Measurement, & Research  You can Quit smoking Now!