Use of the Delphi method and nominal group technique in front-end market segmentation.

Abstract: The purpose of this study was to perform front-end segmentation to identify the most appropriate population subgroup for physical activity programming in Utah. A three-round Delphi survey was conducted among 48 physical activity professionals in the state. Analysis involved the use of mean score ranks and confidence intervals (95%). A nominal group procedure was also conducted to supplement results of the Delphi survey. Results indicated that children from middle-class income families be targeted in school settings. This study demonstrated that the use of the Delphi and nominal group were practical and serviceable procedures in front-end audience segmentation.

**********

Although the benefits of physical activity are well documented, approximately 78 percent of adults in the United States do not engage in regular and sustained physical activity (5 or more sessions per week, 30 minutes or more per session, regardless of intensity). Over one-fourth of adults are not active at all in their leisure time (DHHS, 2000a).

Data from the 1999 Youth Risk Behavior Survey indicate that only 64% of U.S. adolescents in grades 9-12 had participated in activities that made them sweat and breathe hard for at least twenty, minutes on at least three days per week (DHHS, 2000b). Data from the National Health and Nutrition Examination Survey (III) reveal that approximately 30% of boys and girls ages 8-10 do not engage in three or more sessions of weekly play or exercise that results in sweating or breathing hard (Andersen, Crespo, Bartlett, Cheskin, & Pratt, 1998).

The gap between recommended levels of physical activity and actual or reported levels of physical activity can be attributed to both personal and external influences, including environmental and social factors. These factors are documented in the Surgeon General's Report on Physical Activity and Health (DHHS, 1996).

From a health promotion delivery perspective, one factor that minimizes the effectiveness of interventions designed to increase physical activity is a lack of segmentation to identify population subgroups, who, for several reasons, may be the most ideal targets. Programs that target specific subgroups become especially critical when public budgets and resources for physical activity are limited and measurable outcomes are expected. In addition, targeted approaches based on the subgroup's wants, needs, and expectations, such as those associated with social marketing frameworks, have shown promise in closing gaps between recommended and actual health behaviors, including physical activity (Lefebvre, Harden, & Zompa, 1988; Owen, Bauman, Booth, Oldenburg, & Magnus, 1995; Neiger, Thackeray, Merrill, Miner, Larsen, & Chalkley, 2001).

Segmentation refers to the process of dividing a population into distinct subgroups based on characteristics that influence an audience's responsiveness to marketing interventions, such as product benefits they find most attractive or the spokespersons they trust the most (Forthofer & Bryant, 2000). While segmentation will not guarantee a successful intervention, poor segmentation or lack of segmentation will likely result in program failure (Slater, 1995).

Several benefits of market segmentation have been identified. They include providing direction for appropriate resource allocation (Wind, 1978; Swenson, 1990), ensuring programs are customer focused and serve customer needs (Swenson, 1990; Weinstein, 1994; Albrecht & Bryant, 1996), guiding the marketing strategy (Wind, 1978; Weinstein, 1994), ensuring better message design tailored to the audience, and creating a better overall campaign (Slater, 1995; Albrecht & Bryant, 1996). Segmentation requires decision making on three issues: 1) the data or data collection methods that will provide the most reliable foundation for segmentation; 2) the level of specificity required within an identified subgroup; and 3) who will make segmentation decisions.

The most general segmentation strategies use demographic and geographic variables (Slater, 1996). Although these data alone are not usually sufficient to identify an audience's potential response to marketing interventions, they can provide an adequate foundation for segmentation efforts. For example, segmentation decisions cannot be made until an appropriate health problem has been identified. Once identified, the health problem is generally presented in context of its prevalence or associated risk factors among a particular age, race, sex, or socioeconomic stratum. After preliminary decisions result in the selection of a subgroup, more complex segmentation decisions can be made using variables that provide further texture and detail about the identified subgroup. For example, preliminary decision-making may identify African American women 50 years of age or older as an appropriate subgroup for breast cancer screening services. Further analysis may suggest that working women within this subgroup be targeted on the basis of expressed willingness to participate and recruit other participants. The result of this systematic two-stage approach is a more narrowly defined subgroup based on factors that influence or determine consumers' response to the health promotion intervention. We propose the use of the terms front-end and back-end segmentation to help make this distinction. Searches on several standard databases have not revealed the use of these terms in other published works.

Front-end segmentation incorporates demographic or geographic data to provide initial, or general direction, while back-end segmentation may use data from focus groups, in-depth interviews or other survey research, commercial market data sets, etc., to make more specific segmentation decisions about the identified subgroup. Data collection in front-end segmentation usually involves quantitative methods since demographic and geographic data are generally reported in context of epidemiologic data. Back-end segmentation may involve either quantitative or qualitative data or both. The primary purpose of front-end segmentation is to identify a subgroup. The primary purposes of back-end segmentation are to refine and describe the subgroup.

Additionally, segmentation requires some decision making about how specific or narrow to make the selected subgroup. This decision, however, is best made during back-end segmentation after appropriate data have been collected and analyzed, a process commonly referred to as formative research in social marketing (Bryant, 1998). The key is to narrow the sub group sufficiently during front-end segmentation so further refinement and formative research are manageable during back-end segmentation.

Finally, another important aspect of segmentation in general relates to who should actually make decisions that result in identifying a preferred subgroup. It is favorable that a broad, yet manageable number of stakeholders (those who have a vested interest in the program) be involved. If a broad representation of stakeholders is involved early in decision making, including segmentation, their support can significantly aid and sustain ensuing interventions.

ORGANIZATIONAL SETTING OF THE PROGRAM

A challenge recently facing the Utah Department of Health was the need to maximize limited funds for physical activity by segmenting the state population into a more manageable, yet appropriate subgroup. Although physical activity data for Utah suggest relationships exist between a sedentary lifestyle and low income and low education levels, higher income and more highly educated populations respond more favorably to communications and other interventions that promote physical activity (Utah Department of Health, 1998).

There appear to be no significant relationships in Utah data between age, gender (though males are slightly more sedentary), and location of residence (rural vs. urban) and physical activity levels. All non-White races combined in Utah are more sedentary than Whites, though all combined non-White races comprise just over 10 percent of the total population and there are insufficient physical activity data to examine non-White races individually. In addition, there is nothing unique about physical activity data in Utah compared to national data that would make segmentation especially obvious.

In the absence of conclusive epidemiologic data related to common demographic variables, alternative strategies appeared to be warranted to make front-end segmentation decisions. Accordingly, the Delphi method and Nominal Group technique were identified as appropriate methods to obtain professional consensus from key stakeholders involved with physical activity in the state. These front-end segmentation procedures were considered a first step in a broader segmentation and social marketing planning process.

DESIGN AND IMPLEMENTATION DESIGN

A three-round Delphi survey was conducted to identify the most appropriate population sub group and setting for physical activity programs in Utah. The Delphi technique is an established method for obtaining consensus (Duffield, 1993), and has been used to identify problems, define needs, establish priorities, plan curriculum, and identify and evaluate related solutions (Jairath & Weinstein, 1994; Ziglio, 1996). It generally consists of a series of questionnaires which are developed or refined in sequential stages until consensus is achieved. The Delphi technique has a long history of use in health and medicine (Fink, Kosecoff, Chassin, & Brook, 1984).

A nominal group procedure was used to validate results from the Delphi survey and complete the front-end segmentation process. Gilmore and Campbell (1996) recommend the nominal group process incorporate five to seven people who are knowledgeable about issues under examination. Participants are asked to respond to questions without open discussion before responding in a round-robin fashion that continues until every response from every participant is recorded. Responses are clarified through discussion, then participants select and rank options and make final votes. Since its development in the late 1960s, the nominal group procedure has been used in a variety of professional settings (Gilmore & Campbell, 1996).

SAMPLE

A total of 61 professionals were initially identified by the authors as qualified to provide informed opinion on ideal population subgroups and settings for physical activity. Names of these individuals were verified by interviews with some of those identified, as well as other health professionals with broad knowledge of Utah's public health and physical activity infrastructure. Potential Delphi survey respondents were selected from community-based programs, health care settings, departments of exercise and sports science or physical education in universities throughout the state, the Utah Department of Health, the State Office of Education, local health departments, selected schools, and various worksites.

Criteria for inclusion in the Delphi survey included having a college degree in exercise science, health education, physical education, or a closely related field; a minimum of five years of professional experience related to the promotion of physical activity; a minimum of 25 percent of current job responsibility related to physical activity; a broad orientation in physical activity (i.e., not limited to sports events or fitness classes); and being identified as a leader by professional colleagues. A total of 48 respondents (79 percent) returned the first round of the Delphi survey, and subsequent rounds were limited to these 48 respondents. The response rate for Rounds 2 and 3 was 100 percent. Delphi survey participants were intended to represent the highest level of expertise from Utah's most knowledgeable physical activity professionals.

The nominal group process involved seven physical education/healthy lifestyle curriculum directors from the seven largest school districts in the state representing 61% of all students enrolled in public schools. The composition of nominal group participants was based on results form the Delphi survey. Participants were considered to be the most knowledgeable and best trained physical activity professionals currently working with both children and adolescents in the state. Smaller school districts throughout the state do not routinely employ this type of curriculum director and their faculty do not generally possess the level of expertise required for the nominal group procedure.

INSTRUMENTATION

The questionnaire used in Round 1 of the Delphi survey included seven demographic variables composed of 32 items. Use of demographics was selected as an appropriate framework for front-end segmentation. The study variables were: age (children = 5-12 years, adolescents = 13-18, young adults = 19-25, young middle-aged adults = 26-44, older middle-aged adults = 45-64, and senior adults = 65 and older); gender, race/ethnicity (White, Hispanic, African American, Asian and Pacific Islanders, and American Indian, Eskimo, and Aleut); income (low = <$20,000, middle = $20-50,000, high = >$50,000); setting (worksites, schools, health care facilities, churches, and community groups); education level (less than high school, high school diploma or equivalent, some college, and a college degree); and general criteria ranking (i.e., which of the demographic variables was most valuable for segmenting the population). Items were reviewed by four surveillance system experts to help organize demographic categories.

In Round 1, participants were asked to rank all 32 items using two criteria: justifiable and probable. Justifiable was defined as a combination of three factors: the least active population, the population most susceptible to health problems related to physical inactivity, and the total number of people affected. Respondents were asked to combine all three components of the justifiable criterion to produce a general score. This was done to establish a general risk estimate. Probable was defined as the likelihood the population would increase its physical activity level. Ranking involved assigning a value of 1 (most justifiable or most probable) to 3, 4, 5, or 6 (depending on the item, least justifiable or least probable) for each of the 32 items across the seven variables.

In subsequent rounds, respondents were asked to review their original responses and either retain them or change them based on the mean scores of all participants or written comments from previous rounds. This is a standard approach in Delphi survey methodology to achieve consensus for items where some degree of agreement has not yet been established.

In Round 3, the justifiable and probable criteria were combined for forced decision-making. Participants were required to simultaneously consider risk estimates and likelihood of engaging in physical activity.

Following Round 3 of the Delphi survey, a nominal group procedure was used to achieve consensus on age and clarify an additional finding from the Delphi survey related to race. The nominal group process used the same criteria as the Delphi survey (justifiable and probable criteria) and determined whether children or adolescents was the most appropriate age segment, and whether dropping race as a segmentation variable was warranted. The nominal group process was performed due to inconclusive results from the Delphi survey. Participants considered these issues independently for one week prior to the nominal group discussion.

ANALYSIS

Consensus was achieved in the Delphi survey when an item within each variable was ranked highest for both the justifiable and probable criteria, and the confidence intervals (95 percent) associated with the mean scores were non-overlapping, meaning that the difference was statistically significant. When consensus was achieved for a variable, it was deleted from analysis in subsequent rounds. Consensus was achieved in the nominal group process when all participants concurred fully in a final vote.

PROGRAM OUTCOME MEASURES

Consensus was achieved in Round 1 of the Delphi survey on one of the seven variables. Table 1 displays common agreement on lead ranks (mean scores) and significantly different confidence intervals related to gender only. While children and adolescents were identified as priority population segments for physical activity, an inverse order was observed between the justifiable and probable criteria. This means that although adolescents were perceived to be at higher risk than children for physical inactivity and more susceptible to problems related to physical inactivity, children were perceived to be more likely to increase their current levels of physical activity.

White and Hispanic populations received the highest ranks for race on both the justifiable and probable criteria, though consensus was not achieved. Based on conclusive written comments in Round 1 suggesting race should not be used as a segmentation criterion, it was dropped from the Delphi survey. While schools received the highest rank for setting on both the justifiable and probable criteria, confidence intervals were not significantly different than worksites as a setting. Inverse results were observed for income where low income populations were perceived to be at higher risk, and high income populations were viewed as more likely to increase physical activity levels. This paralleled results for education, where less educated individuals were perceived to be at higher risk and higher educated individuals were thought to be more likely to increase physical activity levels. The two variables that were reported as most significant for making segmentation decisions were age and setting. However, confidence intervals were not significantly different.

Consensus was achieved in Round 2 for setting and general criteria ranking, but not for age, income, or educational attainment. Table 2 indicates that schools were identified as the priority setting, and that age, followed by setting, was significantly more important than the other variables as segmentation variables. The same findings and patterns that were noted in Round 1 for variables not attaining consensus were observed in Round 2.

Round 3 provided consensus on income (medium or middle income), but not age. Table 3 displays that children ranked slightly higher than adolescents when combining the justifiable and probable criteria. However, the difference was not significant using confidence intervals. Level of education was dropped as a survey variable in Round 3 because of the preference to focus on children and adolescents as key population segments related to age (i.e., all would have less than a high school education). Since age and setting were identified as the two most important segmentation variables and both were related to children and adolescents, this decision appeared warranted.

The nominal group process resulted in a unanimous decision to target children in Utah for physical activity programming. Based on schools being selected as the preferred setting, unanimous agreement was obtained that either children or adolescents could be justified as a primary population segment. However, two arguments related to available resources and the perceived impact of physical activity among children versus adolescents ultimately resulted in the decision to target children. There was unanimous agreement in the nominal group discussion that a distinct difference exists in resource allocation for physical activity between children and adolescents. For example, there are a sufficient number of trained physical education teachers and specialists in junior (grades 7-9) and senior (grades 10-12) high schools, as well as physical education resources. Conversely, there are substantially fewer such resources in elementary schools. There was also a unanimous response that elementary school represents a period of time in life when children are developing attitudes and habits about physical activity and other health behaviors. There was common agreement that programs can have a greater influence and impact on attitudes and behaviors at younger, rather than older, ages.

Because race was eliminated from analysis in the Delphi survey on the basis of written comments, and not by consensus, a final discussion point focused on the importance of race as a segmentation variable. There was unanimous agreement that no racial or ethnic distinction be made in segmenting children for physical activity programming. Participants reported that this had less to do with the value of race as a segmentation variable and more to do with equity for children of all color.

DISCUSSION

Delphi survey respondents perceived no justification for targeting one gender at the exclusion of the other. However, this determination was made before consensus was achieved for age and setting. Although few studies have examined gender differences in physical activity levels among children, it has been reported that boys are more physically active than girls (Kohl & Hobbs, 1998; DHHS, 1999; Sallis, Simons-Morton, Stone, Corbin, Epstein, & Faucette et al., 1992). In general, differences between genders will require program developers to pay attention to formative research data in back-end segmentation, particularly as it pertains to the design of messages and interventions.

Likewise, race was considered an inappropriate segmentation criterion. Racial and ethnic minorities, as compared to Whites are often at higher risk for health related problems. But because approximately 90% of Utah's population is White, it is difficult to exclude Whites in the absence of convincing data or other evidence that clearly supports segmentation of a particular race. Conversely, it was suggested in Round 1 that future physical activity programs not exclude ethnic minorities. Individuals from middle income households were identified as a priority target for physical activity. This was likely more a result of combining the justifiable and probable criteria, since the first two rounds of the Delphi survey produced inverse rankings for income. Whereas individuals from high income households ranked higher on the probable criterion, individuals from low income households ranked higher on the justifiable criterion in both Rounds 1 and 2.

The most appropriate setting identified in the survey was schools. This clearly related to the interest and focus of interventions being directed at children and adolescents. Harsha (1995), stated schools are an ideal setting to reach children and adolescents with health promotion programs. Schools may be a natural setting to promote healthy lifestyles and physical activity because topics are already included in health education curricula, and mandated physical activity programs exist (Pigg, 1989). High percentages of children and adolescents are also enrolled in physical education classes at school (Ross & Gilbert, 1985).

Delphi respondents identified children and adolescents as priority targets for age, although consensus was not achieved. In Round 2, adolescents were identified as more justifiable, though the difference was not significant. In contrast, children were seen as significantly more probable to engage in physical activity. In Round 3, when asked to combine both justifiable and probable, children received a higher ranking, but the confidence intervals were not statistically different.

While Delphi survey participants suggested children are more likely than adolescents to engage in physical activity, nominal group participants suggested that childhood represents a period of life more conducive to developing strong attitudes, beliefs, and behaviors related to physical activity. This appears to be supported by available literature. For example, the most extensive and promising research on interventions for promoting physical activity among young people has been conducted with students in schools, primarily at the elementary school level (DHHS, 1996). One study reported that when schools are used as a primary setting for physical activity with appropriate training and support, physical activity levels among children can be raised substantially (Luepker, Perry, McKinlay, Nader, Parcel, & Stone, 1996).

Using the Delphi and Nominal Group procedures and involving a broad representation of stakeholders in front-end segmentation serves many useful purposes. First, it assures the Utah Department of Health that the preliminary sub group selected represented the preference of a broad cross section, and not just a few individuals. Second, it helps justify the decision to target children to an even wider array of stakeholders (i.e., legislators, school administrators, other public health agencies, etc.). Finally, it creates vested interest in programs and activities. For example, the 55 experts who participated in the Delphi and Nominal Group techniques are more likely to support physical activity programs pertaining to children in schools because of their participation in preliminary decision making.

Front-end segmentation identified children in elementary schools as a primary target audience for physical activity in the state. A central objective of back-end segmentation is to refine this preliminary target audience. Refinement may be based on factors such as readiness to change, capacity to support change, or support for physical activity in general, all of which may be discovered in focus groups or other survey research. Back-end segmentation may involve surveys of parents, teachers, PTA representatives, or school administrators to identify ideal locations, schools, grades, or even an ideal gender or race. Available market research data, including psychographics, may reveal characteristics of parents in certain locations that relate to key characteristics about their children's physical activity patterns. In summary, back-end segmentation will be used to understand wants, needs, values, benefits, and barriers, to further describe and refine the selected subgroup.

As illustrated in this case study, segmentation is a multi-step process. The use of the Delphi and Nominal Group procedures was the first phase of a comprehensive social marketing project. A systematic approach to front-end segmentation, as described here, enables practitioners to carry out the requisite back-end segmentation which further narrows the subgroup and provides direction for the development of interventions. The resulting strategies and communications which reflect consumer preferences are more likely to resonate with target audiences and be more effective at modifying health behavior.

Table 1. Delphi Results for Round One: Justifiable and Probable *

Variable                    Mean-J    C.I.-J      Mean-P    C.I.-P

Children                    3.04    2.42 - 3.66   1.70    1.33 - 2.06
Adolescents                 2.78    2.29 - 3.28   3.09    2.57 - 3.60
Young Adults                3.83    3.43 - 4.23   3.13    2.77 - 3.49
Young Middle Aged           3.19    2.74 - 3.64   3.65    3.26 - 4.04
Old Middle Aged             3.09    2.58 - 3.59   4.24    3.83 - 4.65
Seniors                     3.77    3.22 - 4.31   4.57    4.07 - 5.06
Male                        2.73    2.49 - 2.97   2.47    2.20 - 2.74
Female                      2.23    1.97 - 2.49   2.29    2.00 - 2.58
Both                        1.26    1.10 - 1.41   1.41    1.25 - 1.57
White                       2.05    1.57 - 2.53   1.37    1.09 - 1.65
African American            3.10    2.59 - 3.60   3.20    2.83 - 3.56
Asian American              3.12    2.73 - 3.51   3.05    2.71 - 3.39
American Indian             3.17    2.70 - 3.64   4.20    3.79 - 4.60
Hispanic                    2.21    1.86 - 2.57   2.81    2.46 - 3.16
Low Income                  1.40    1.23 - 1.57   2.60    2.39 - 2.81
Medium Income               1.69    1.57 - 1.87   1.73    1.55 - 1.92
High Income                 2.71    2.50 - 2.92   1.69    1.43 - 1.94
Worksite Setting            2.56    2.14 - 2.97   3.00    2.62 - 3.38
School Setting              1.78    1.42 - 2.15   1.47    1.20 - 1.74
Health Care Setting         3.64    3.22 - 4.07   3.67    3.27 - 4.06
Church Setting              3.49    3.11 - 3.87   3.91    3.56 - 4.26
Community Setting           2.64    2.30 - 2.99   2.49    2.18 - 2.80
Some H.S. Education         1.63    1.33 - 1.93   3.26    2.91 - 3.61
High School Diploma         1.63    1.46 - 1.80   2.59    2.34 - 2.83
Some College                2.74    2.50 - 2.98   2.09    1.88 - 2.29
College Degree              3.28    2.95 - 3.62   1.80    1.46 - 2.15
Age                         1.83    1.45 - 2.21   2.21    1.75 - 2.67
Gender                      3.98    3.48 - 4.47   3.79    3.29 - 4.29
Ethnicity                   4.20    3.77 - 4.63   4.88    4.51 - 5.25
Income                      3.87    3.42 - 4.32   3.86    3.34 - 4.38
Setting                     2.67    2.16 - 3.18   2.66    2.23 - 3.08
Education                   3.62    3.18 - 4.07   3.00    2.55 - 3.45

* Lower scores reflect higher rankings
Mean-J = Mean Score for the Justifiable Criterion
C.I.-J = Confidence Intervals for Justifiable
Mean-P = Mean Score for the Probable Criterion
C.I.-P = Confidence Intervals for Probable
Table 2. Delphi Results for Round Two: Justifiable and Probable *

Variable              Mean-J     C.I.-J      Mean-P     C.I.-P

Children               2.92    2.36 - 3.47    1.38    1.09 - 1.66
Adolescents            2.40    1.96 - 2.85    2.91    2.46 - 3.37
Young Adults           3.61    3.22 - 4.00    3.29    2.99 - 3.59
Young Middle Aged      3.24    2.83 - 3.66    3.73    3.40 - 4.06
Old Middle Aged        3.29    2.78 - 3.80    4.27    3.90 - 4.64
Seniors                4.04    3.52 - 4.57    4.53    4.04 - 5.03
Male                   2.78    2.62 - 2.93    2.51    2.31 - 2.71
Female                 2.09    1.92 - 2.26    2.36    2.16 - 2.55
Both                   1.17    1.04 - 1.30    1.23    1.09 - 1.37
Low Income             1.33    1.17 - 1.50    2.70    2.52 - 2.88
Medium Income          1.72    1.56 - 1.88    1.77    1.61 - 1.92
High Income            2.81    2.66 - 2.95    1.51    1.30 - 1.72
Worksite Setting       2.29    1.99 - 2.60    2.94    2.63 - 3.24
School Setting         1.58    1.26 - 1.91    1.23    1.03 - 1.43
Health Care Setting    3.89    3.51 - 4.28    3.96    3.63 - 4.28
Church Setting         3.58    3.25 - 3.92    3.85    3.50 - 4.20
Community Setting      2.65    2.36 - 2.93    2.47    2.21 - 2.73
Some H.S. Education    1.53    1.26 - 1.80    3.48    3.18 - 3.78
High School Diploma    1.67    1.51 - 1.84    2.70    2.51 - 2.88
Some College           2.72    2.49 - 2.94    3.04    1.84 - 2.25
College Degree         3.46    3.16 - 3.75    1.52    1.23 - 1.81
Age                    1.60    1.29 - 1.91    1.79    1.44 - 2.15
Gender                 4.09    3.58 - 4.59    4.15    3.79 - 4.51
Ethnicity              4.45    4.06 - 4.84    5.13    4.82 - 5.44
Income                 3.87    3.47 - 4.27    3.98    3.52 - 4.44
Setting                2.66    2.21 - 3.11    2.57    2.20 - 2.95
Education              3.49    3.13 - 3.85    2.83    2.46 - 3.20

* Lower scores reflect higher rankings
Mean-J = Mean Score for the Justifiable Criterion
C.I.-J = Confidence Intervals for Justifiable
Mean-P = Mean Score for the Probable Criterion
C.I.-P = Confidence Intervals for Probable
Table 3. Delphi Results for Round Three:
Justifiable and Probable Combined

Variable        Mean    SD    Confidence Intervals

Children        1.39   0.49       1.24 - 1.54
Adolescents     1.56   0.50       1.42 - 1.71
Low Income      1.71   0.58       1.54 - 1.88
Medium Income   1.35   0.48       1.21 - 1.49
High Income     2.94   0.24       2.87 - 3.01

REFERENCES

Albrecht, T. L., & Bryant, C. (1996). Advances in segmentation modeling for health communication and social marketing campaigns. Journal of Health Communication, 1, 65-80.

Andersen, R.E., Crespo, C.J., Bartlett, S.J., Cheskin, L.J., & Pratt, M. (1998). Relationship of physical activity and television watching with body weight and level of fatness among children: Results from the Third National Health and Nutrition Examination Survey. Journal of the American Medical Association, 279, 938-942.

Bryant, C. (1998, June). Social marketing: A tool for excellence. Session presented at the eighth annual Social Marketing in Public Health Conference, Clearwater Beach, FL.

Department of Health and Human Services (1996). Physical Activity and Health: A Report of the Surgeon General. National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention: Atlanta, GA.

Department of Health and Human Services (1999). Guidelines for school and community programs to promote lifelong physical activity among young people. Morbidity Mortality Weekly Report, 46 (no. RR-6, pp 1-36). Division of Adolescent and School Health, Centers for Disease Control and Prevention. Atlanta, GA.

Department of Health and Human Services (2000a). 2000 BRFSS Summary Prevalence Report. National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention: Atlanta, GA.

Department of Health and Human Services (2000b). Youth Risk Behavior Surveillance--United States, 1999. Morbidty Mortality Weekly Report, Division of Adolescent and School Health, Centers for Disease Control and Prevention; 49 (SS-5): 1-94. Atlanta, GA.

Duffield, C. (1993). The Delphi technique: A comparison of results obtained using two expert panels. International Journal of Nursing Studies, 30,227-237.

Fink, A., Kosecoff, J., Chassin, M., & Brook, R. H. (1984). Consensus methods characteristics and guidelines for use. American Journal of Public Health, 979-983.

Forthofer, M. S., & Bryant, C. A. (2000). Using audience-segmentation techniques to tailor health behavior change strategies. American Journal of Health Behavior, 24, 36-43.

Gilmore, G. D., & Campbell, M. D. (1996). Needs assessment strategies for health education and health promotion (pp. 65-72). Dubuque, IA: Brown and Benchmark.

Harsha, D.W. (1995). The benefits of physical activity in childhood. American Journal of Medicine and Science, (Supplement 1); S109

Jairath, N., & Weinstein, J. (1994). The Delphi methodology: A useful administrative approach. Canadian Journal of Nursing Administration, 7, 29-42.

Kohl, H. W., & Hobbs, K. E. (1998). Development of physical activity behaviors among children and adolescents. Pediatrics, 101, 549-554.

Lefebvre, R.C., Harden, E. A., & Zompa, B. (1988). The Pawtucket Heart Health Program: Social marketing to promote community health. Rhode Island Medical Journal, 71, 27-30.

Luepker, R. V., Perry, C. L., McKinlay, S. M., Nader, P. R., Parcel, G. S., & Stone, E.J. (1996). Outcomes of a field trial to improve children's dietary patterns and physical activity: The child and adolescent trial for cardiovascular health (CATCH). Journal of the American Medical Association, 275,768-776.

Neiger, B. N., Thackeray; R., Merrill, R. M., Miner, K. Larsen, L., &

Chalkley, C. M. (2001). The impact of a social marketing on fruit and vegetable consumption and physical activity among public health employees at the Utah Department of Health. Social Marketing Quarterly, (7) 1, 10-28.

Owen, N., Bauman, A., Booth, M., Oldenburg, B., & Magnus, P. (1995). Serial mass-media campaigns to promote physical activity: Reinforcing or redundant? American Journal of Public Health, 85, 244-248.

Pigg, R. M. (1989). The contribution of school health programs to the broader goals of public health: The American experience. Journal of School Health, 59, 25-30.

Ross, J. G., & Gilbert, G. G. (1985). The national children and youth fitness study: A summary of findings. Journal of Physical Education, Recreation, and Dance, 56, 45-50.

Sallis, J. F., Simons-Morton, B. G., Stone, E. J., Corbin, C. B., Epstein, L. H., & Faucette N., et al. (1992). Determinants of physical activity and interventions in youth. Medicine and Science in Sports and Exercise, 24 (Supplement):S248-S257.

Slater, M. D. (1995). Choosing audience segmentation strategies and methods for communication. In E. Maibach & R.L. Parrott (Eds), Designing health messages: Approaches from communication theory and public health practice (pp. 817-846). Newbury Park, CA: Sage.

Slater, M. D. (1996). Theory and method in health audience segmentation. Journal of Health Communication, 1, 267-283.

Swenson, C.A. (1990). Selling to a segmented market: The lifestyle approach (pp. 12-16). New York: Quorum Books.

Utah Department of Health. (1998). Physical Activity in Utah--BRFSS (pp. 1-4). Bureau of Health Education, Division of Community and Family Health Services: Salt Lake City, UT.

Weinstein, A. (1994). Market segmentation: Using demographics, psychographics and other niche marketing techniques to predict and model customer behavior (pp. 6, 9). Chicago, IL: Probus.

Wind, T. (1978). Issues and advances in segmentation research. Journal of Marketing Research, 15, 317-337.

Ziglio, E. (1996). The Delphi method and its contribution to decision making. In M. Adler & E. Ziglio (Eds). Gazing into the oracle: The Delphi method and its application to social policy and public health (pp. 3-33). London: Jessica Kingsley.

Brad L. Neiger, PhD, CHES Michael D. Barnes, PhD, CHES Rosemary Thackeray, PhD, MPH Natalie Lindman, MS

Brad L. Neiger, Ph.D., CHES, Michael D. Barnes, Ph.D., CHES, and Rosemary Thackeray, Ph.D., MPH, are affiliated with the Department of Health Science, College of Health and Human Performance, Brigham Young University. Natalie Lindman, MS, is a community health specialist in the Physical Activity Program at the Utah Department of Health. Address all correspondence to Brad Neiger, PhD, CHES, Department of Health Science, College of Health and Human Performance, Brigham Young University, PO Box 22107, Provo, Utah 84602-2107, Phone: [phone]; FAX: [phone]; e-mail: [e-mail].