This glossary collects the key concepts, people, and ideas you will meet in this lesson. Use it as a reference while you work through the material, or as a review before assessments. Type in the search box to filter entries.

Pushing beyond a single level

This lesson marks a turn in the course. Lessons 3, 4, and 5 reviewed theories that each work mainly at one ecological level: the individual, the social group, the organization, the culture, the political economy. The first half of this lesson asks what happens when we admit that those levels act on each other, and it introduces theories built to describe that interaction. The second half then asks the practical question: once you accept that influences are layered, where do you start?

The Einstein problem

Start with Albert Einstein, who wrote that a theory is more impressive the simpler its premises, the more different kinds of things it relates, and the wider its area of application. Einstein's general theory of relativity revolutionized physics, but it dealt primarily with gravity, one of the fundamental forces understood at the time alongside electromagnetism. He spent years trying to build a unified theory that explained both forces together and did not succeed. The parallel is plain. Each of the theories reviewed so far focuses on behavioural drivers at a single ecological level. Is that how the world works? The honest answer is: not likely.

The levels are a filing system. We divide explanatory theories by level because it helps us see what a theory refers to. At the same time, it is generally understood that factors at one level interact with those at other levels, even though most theory does not account for that. Suppose a theory connects the social dynamics of a family to the attitudes and behaviours of children. What accounts for those family dynamics? Socioeconomic constraints, low community efficacy, racism, gender roles, and more. The question for this lesson is how to account for the way all of those factors work together. Theorists often shy away from the attempt, and there is no wonder in that: it is a very difficult challenge. But if the ecological model is the field's basic understanding of the context within which behaviour occurs, then theory has to take the challenge on. We cannot stay pigeonholed in specific levels.

Equation versus path

What would a multilevel theory look like? Single-level theory is typically framed in somewhat linear terms, almost like an equation: think of the Health Belief Model, where perceptions add up to a likelihood of action, or the Transtheoretical Model, or diffusion of innovations. Multilevel theory is more likely to be framed as a path, or multiple paths, through the ecological layers. Those paths may be complex, like trees or networks of dendrites, and they may run in more than one direction: a behaviour change at the individual level may reverberate back up through the broader levels and change the character of the influence pathway itself. Admittedly, this can get abstract. The diagram below holds the contrast in mind.

The granddaddy: Urie Bronfenbrenner

This has to be the first multilevel theory examined, because it contributed so much to the ecological model in public health. Urie Bronfenbrenner (1917 to 2005) was a Russian-born American psychologist who focused on child development and was influenced by theorists who regarded human development as an interactive social process (Bronfenbrenner). He also cofounded the Head Start program in the United States, one of the most notable efforts ever made to address the effect of poverty on children. Canada's Aboriginal Head Start programs, launched federally in the 1990s in urban, northern, and on-reserve communities, carry the same idea into an Indigenous context with components spanning culture and language, education, health promotion, nutrition, social support, and parental involvement: a multilevel program in practice.

Most important for this course, Bronfenbrenner developed Ecological Systems Theory, a multilevel systems approach for understanding child outcomes, including behaviour (Bronfenbrenner, 1977). The basic idea is that child development and behaviour can be understood as outcomes of the interaction between an individual and several levels or layers within the surrounding system. The levels run outward from the child. Click each card for the definition and examples.

Adding biology and time

In the early 1990s and beyond, Bronfenbrenner began to amend the original theory to account for the interaction of the multiple systems with the individual's own biological processes, including psychological, cognitive, and genetic elements (Bronfenbrenner and Ceci, 1994). The systemic process came to be understood as one that involves the interaction of individuals, with all the internal biopsychological processes occurring at a particular developmental stage, and the multilayered environment. It was at this point that the temporal element was added, represented by the chronosystem, to reflect the evolving pattern of interaction as individuals develop and engage in different ways over time with the surrounding layers. The amended version is called Bioecological Systems Theory.

Adding the individual's own processes led to a distinction you will meet again in Flay's theory in the next section. Proximal processes occur between an individual and their immediate environment. Distal processes occur within that environment and eventually affect the immediate, proximal interactions. A conversation between a teenager and a parent about vaping is proximal. A provincial decision to restrict flavoured vaping products is distal, and it reaches the teenager through what stores stock, what friends can buy, and what the parent has read in the news.

Rodney in Ourlandia: the process in motion

A story illustrates the theory in motion, with one warning: the behavioural outcome does not arrive until the end. Open the stages in order and watch which systems are acting at each step.

Notice what the story does. No single fact explains Rodney. Gender beliefs (macrosystem), a changing economy (exosystem over time), family and school practices (microsystem), and adolescent biology all feed into the same outcome, and the outcome that arrives depends on an intervention made at a level Rodney never touches directly.

Using the theory: the Coordinated School Health Program

Many health promotion interventions are ecological in nature and draw from Bronfenbrenner, which is no surprise, because public health is grounded in the ecological model. A good example is the Coordinated School Health Program (CSHP), an approach to school health with eight components, adopted by the Division of Adolescent and School Health at the Centers for Disease Control and Prevention (CDC) as its general framework and then by the World Health Organization (WHO) in its global promotion of school health. The eight components have traditionally been drawn arrayed around a graphic of children, with no distinctions between components or ecological levels. It is reasonable to argue that this misses the ecological nature and complexity of implementation. Lohrmann proposed a complementary ecological model in which parents (microsystem), schools, and community institutions (mesosystem, exosystem) are organized into levels of influence working outward from the students. The table sorts the eight components in that spirit; note that Lohrmann's model is an attempt to organize the program into levels rather than an exact replica of Bronfenbrenner's.

Canada's version is Comprehensive School Health, promoted since 2005 by the Pan-Canadian Joint Consortium for School Health, a partnership of provincial and territorial health and education ministries. It organizes school health into four interrelated components: the social and physical environment; teaching and learning; healthy school policy; and partnerships and services. The four map readily onto Bronfenbrenner's layers, from classroom teaching in the microsystem to board policy and community partnerships in the exosystem, and the framework's insistence that the components be planned together is the ecological point in a sentence.

Things to think about: complexity

One critique stands out, and it is the biggest critique of all similar theories: complexity. With so many interacting levels and processes, how can you tell where the causal chain leading to a health behaviour lies? Almost as in chaos theory, a small change in one component within one level is likely to cause reverberations through multiple levels, and those reverberations may in turn change the environment at every level with which the individual was interacting in the first place. Two questions follow and shape the rest of the lesson. What kind of research could identify such causal chains? And how could you set up an intervention to create a causal chain that leads to better health? Flay's theory, next, is one attempt to make the chains traceable.

One waterfall, three streams

Bronfenbrenner gave us layers. Brian R. Flay and his colleagues set out to give us a map of how influence flows through them. The Theory of Triadic Influence (TTI) is the second general multilevel theory in this lesson and, because it was built by researchers working in applied settings, it comes with a clearer instruction for practitioners: find the stream, follow it upstream, and intervene where you can.

Three friends and a broken car

The TTI came out of a view that the most commonly used theoretical approaches, and the health promotion interventions built on them, address only pieces of the puzzle. Because of that, interventions are often narrowly targeted, and it is difficult to confirm the validity of any one theory. For some health issues the theories in use come from several disciplines, each with its own terminology and assumptions. This fragmentation is a real problem: how can we discuss or address a problem with any consistency when the people working on it are not speaking the same language?

A car that will not start makes the point. You enlist three friends. One insists it is all about temperature, because cars are made of metal and everything has to be at a medium temperature before you can fix anything. The second says it is always the electrical system. The third tells you to sit by the car and feel it, because nothing will be fixed until you share a vibe with it. The chances of getting the car fixed are not good. The TTI is an attempt to schematize the full spectrum of factors and processes involved in health-related behaviour so that there is a common point of reference, even when a given intervention, as most must, addresses only some of them. Like Bronfenbrenner's bioecological model, it holds that factors at multiple levels are linked in continuous, interactive pathways, and that some act in immediate, proximal ways while others act as long-term, distal or underlying causes.

Two dimensions

The model is organized around two basic dimensions (Flay, Snyder, and Petraitis, 2009).

Levels of causation. Contributing factors are sorted by how close they sit to the behaviour: (1) causally proximal or immediate factors, such as intentions and skills; (2) causally distal or predisposing factors, such as family relationships, social capital, and community laws or policies; and (3) underlying or ultimate causes, such as structural poverty, gender roles, and cultural goals and values.

Streams of influence. The factors operate across those causal levels via three streams. Each stream ends in a different proximal predictor of behaviour. Use the tabs to follow each one.

The factors at the various levels flow within each stream and act through the proximal predictors. Flay and colleagues draw the streams as arrows running from the social context, the cultural environment, and biology and personality, through social normative beliefs, attitudes, and self-efficacy, into decisions and intentions and then behaviour. A later version of the diagram, developed with specific reference to adolescent smoking, adds the causal levels as horizontal bands that the streams cross.

The waterfall

If that is hard to hold in mind, a picture helps. Behaviour at the individual level is like what happens at the bottom of a waterfall: the culmination of a cascade of influences that interact and mingle on their way down through the levels of causation to the point where the water hits the rocks or the pool below. You can look at the immediate determinants of a decision, but these can be traced back up the waterfall to their source. To influence behaviour, your task is to trace the factors back upstream and focus an intervention at a point where it is feasible and where you think it can make a difference in the chain. The grid below lets you practise the tracing for youth vaping.

Other elements of the theory

The TTI does more than sort factors. It acknowledges and integrates other social and health behaviour theories by causal level and stream, which is what makes it a common point of reference rather than a rival. The table collects the placements the theory's developers list.

There is more. Each stream has two substreams: an informative, cognitive, thoughts substream and a control, affective, feelings substream, both flowing from the most distant causes to the causes closest to the behaviour. The interactions are not easy to pin down; interactions between streams can be negative or positive influences. Once a behaviour occurs there is a feedback loop, based on the external and internal reaction to it, that causes new behaviours and new loops. Related behaviours have similar causes: smoking and substance abuse share distal causes almost entirely and differ somewhat in proximal causes, while smoking and skiing, a pair of less related behaviours, have fewer causes in common.

All of these levels, paths, and directions pose challenges for anyone collecting and analyzing data, whether to guide the choice of an intervention or to evaluate one said to be based on the TTI. Complex statistical methods that can trace the effect of multiple variables on each other and on behaviour are very likely to be necessary; path analysis and structural equation modelling are the usual candidates. The developers suggest thinking in terms of causal bundles related to specific behaviours, an idea that resembles the syndemic health conditions from Lesson 5, in which issues such as HIV/AIDS, substance abuse, and violence are related by similar causal circumstances.

Using the theory: adolescent smoking in Washington State

Because the TTI was developed by researchers working in an applied context, it has been used in intervention studies, in research on predictors of health risk behaviour, and in program planning. One example is a predictor study of adolescents' transition from initiating smoking to regular smoking, using a large sample from Washington State (Bricker and colleagues, 2009). The team assessed three kinds of factor: (1) risk factors based on an individual's psychological makeup, such as a thrill-seeking tendency, at the ultimate level in the individual stream; (2) psychological risk factors related to the social world, such as a tendency to do what friends do, at the distal level in the individual stream; and (3) the behaviours of others, specifically parent and friend smoking, at the distal level in the social stream. According to the TTI, the more proximal second and third types should influence smoking decisions more than the individual characteristics.

The results were mixed but gave some support. Adolescents with individual psychological risk factors were more likely to move through the transition stages, there was variation in how much the distal factors outweighed them, and the ultimate-level characteristics interacted with the distal factors in predicting progression. That interaction is what the theory expects, and it is also exactly what makes the theory hard to test cleanly.

Things to think about: two critiques

Critique 1: complexity. As with Ecological Systems Theory, the biggest issue is complexity. Tracing and tracking down the multiple influence pathways and interactions is not easy, and for that reason the theory is not easy to demonstrate. There are so many possible influence chains that an intervention or a research effort faces significant difficulties in trying to investigate them all. Despite the difficulties, approaches like this are important.

Critique 2: where things are placed. It is not always clear what the rationale is for positioning some factors at particular levels, with the causal implications that entails. One example: social competence is positioned at a causal level above social skills, though skills could be argued to be a necessary precursor to competence. This is partly an issue of presentation, and Flay and colleagues acknowledge that their diagrams are simplified and do not always capture the scope of interactions.

Other multilevel models

Beyond the two general theories, there are multilevel frameworks developed with reference to a particular health problem or population. Three are described here, and they belong in the same family as the risk and protective factors model you will meet in Section 4. Open each to see how it organizes the levels.

Notice what these three share with the general theories and with each other. Each sorts contributing factors into levels that run from the person outward to society, each assumes that the levels interact, and each concludes that action at one level alone is unlikely to hold. That conclusion is the bridge to the second half of this lesson, where the question becomes which level, and which factor, to act on first.

Where do you start?

This section turns from explaining behaviour to changing it. Now the fun begins. Or rather, not so fast. The transition from theory to practice is one of the most difficult things to accomplish, and how to do it well must be one of the most discussed topics in public health journals, meetings, and policymaking agencies. This section covers the first half of that transition: why it is hard, how to choose what to work on, and the assessment phases of the best-known planning framework, PRECEDE-PROCEED.

Why the transition is hard

You know about the complexity of identifying the causes of health behaviour. You have learned individual, social and group, organizational, political-economic, and cultural theories, and, in the previous two sections, theories that try to connect them. Most importantly, you have learned that each theoretical approach is a piece of an interrelated puzzle. The transition to practice means grappling with that complexity and arriving at a specific intervention, program, policy, communication strategy, message, or coordinated approach that could have an effect on the behaviours relevant to a health problem.

The wide range of factors that affect what people do can be pictured as the layers of an onion, and the ecological perspective remains the most useful framework for understanding problems as they exist in the real world. Then comes the constraint. Unless you have unlimited resources and unlimited time, you cannot address every factor you identify with a program component, so you have to pick and choose. For the young women in the sex trade in Thailand, would you tackle the political economy, urbanization, class, and poverty, or individual knowledge and skills? For blues musicians and smoking, social support and networks, or individual awareness? If you must pick, do it with a planning framework in mind. Most frameworks combine three things: an assessment of the problem and the factors behind the risk behaviours; a means of selecting from those factors and using theory to develop program components; and an assessment of the resources available: funds, people, materials, and time.

Picking your battles: importance and changeability

Green and Kreuter, the authors of PRECEDE-PROCEED, give planners a simple discipline for the selecting step. Each candidate factor is rated on two things. Importance: how strongly is it linked to the health problem, and how widespread is it? Changeability: is there evidence that a program can move it within the time and resources available? Plotting the ratings on a two-by-two grid sorts the candidates into four groups: high on both, the program's focus; important but hard to change, which calls for an innovative program with evaluation built in; changeable but unimportant, a low priority except where a visible early win matters politically; and low on both, no program. Try it on the falls case below.

The matrix is a discipline for thinking rather than a formula. Two things can override it. The community's own priorities, gathered in the first PRECEDE phase, may put an item the epidemiology rates as minor at the top of the list. And a factor that scores low on changeability for a health unit acting alone may score high once a municipal council, a pharmacy chain, or a provincial ministry joins in, which is the point the multilevel theories in the first half of this lesson were making.

Three components of any planning process

PRECEDE-PROCEED is one of the most well-known approaches for planning, implementing, and evaluating health promotion programs (Green and Kreuter, 1999). Its title says a lot about what it entails: a process of assessment and planning before putting a program in place (PRECEDE), followed by implementation and evaluation of the program (PROCEED). This sequence is more or less characteristic of all planning approaches, which involve three steps.

First, assessment, or diagnosis, of the health problem and its causal and supporting factors. Second, development of an intervention based on the assessment, targeting selected factors and using appropriate theory to guide the components. Third, evaluation, to determine whether the intervention was implemented as planned and whether it actually affected the factors it was meant to affect. This is common sense, but one point deserves stress: the three parts work as a continuous feedback loop. What you assess directs which theories apply, what intervention to undertake, and what variables to evaluate, while the way the program works or fails in the real world feeds back into the assessment and modifies the program in turn.

Why PRECEDE-PROCEED

Among the many planning frameworks, PRECEDE-PROCEED is useful in part because it is so comprehensive. It is designed to work within an ecological framework and to help guide the selection of factors to target, and it is oriented toward the community or local level, where the factors behind a health problem are easiest to identify and where, on a reasonable assumption, resources are best organized. Green and Kreuter put the orientation in a sentence: health behaviour "is seen increasingly not as isolated acts under the autonomous control of the individual, but rather as socially conditioned, culturally embedded, economically constrained patterns of living."

The PRECEDE phases: five assessments

The framework runs from the broadest assessment of the problem to the more targeted and specific, then to the administrative and community resources for the program, then to implementation and evaluation. We will follow a hypothetical rust-belt community, North Hill County, through the phases. Each item below gives a description of the phase, the North Hill County illustration, and a Canadian parallel: a health unit in a small British Columbia interior city, call it Cedar Valley, working on falls among adults over 75, the leading cause of injury-related hospitalization among older adults in Canada.

Five terms for the epidemiological assessment

Phase 2 relies on a small vocabulary that every planner needs. Click each card.

The five assessments at a glance

The table below adds two things the walk-through above does not carry for each phase: where the information comes from, and what it tells you.

By the end of phase 5 the planner has moved from a health problem to a set of named factors, a rationale for choosing among them, the theories that speak to each, and an inventory of what the community can bring. The next section picks up at the moment the program goes live, and asks how you would ever know whether it worked.

Proceed: implement, evaluate, and a second way to plan

This section finishes PRECEDE-PROCEED with the phases that follow the assessments, and then introduces a second planning model built for a narrower purpose: youth risk behaviour. The two models look different on the page, and the closing point of the section is that they share the same spine.

Phase 6: Implementation

This phase does not need a long explanation. It means putting your program in place after using the assessment process to design it; that is what is meant by proceed. One reminder matters. In today's program environment there is often a requirement to use evidence-based or model programs that are responsive to the community situation. Once you have made an assessment, there may be programs that already show consistent evidence of effectiveness with a similar population and situation, and it is sometimes advisable to use one instead of developing your own; many agencies maintain registries to help. The assessment still comes first: a model program adopted without one is the cookie-cutter mistake from Lesson 1.

Phases 7, 8, and 9: Evaluation

The last three phases involve evaluation: the data you collect about whether your program is working, in the sense of changing the health problem it is meant to affect. There are two families of reasons to evaluate. The first is external accountability. Dollars are scarce, and programs must show evidence of success. If your program works but you have no data to show it, funding may not continue; other communities will lack the data to argue for adopting it; and an innovative approach may not get the attention it deserves. The second is internal: evaluation tells you whether the program is working, in what ways, and why, so that you can adjust it. The three types correspond to phases 7, 8, and 9. The tabs define each and apply it to a Canadian program, British Columbia's Take Home Naloxone program, run through the BC Centre for Disease Control since 2012, which trains people likely to witness an opioid overdose and gives them a kit to reverse it.

Two other general types of evaluation are formative evaluation, used to develop a program, and summative evaluation, used to draw broad or in-depth conclusions from the evaluation data. Some evaluations also try to analyze the costs and benefits of a program. Lesson 11 returns to evaluation in depth.

The risk and protective factors planning model

The risk and protective factors model is the second example of a planning approach in this lesson. It is widely used by United States federal agencies, including SAMHSA, the CDC, and the Office of Juvenile Justice and Delinquency Prevention, to plan and fund programs on youth risk behaviour: HIV/AIDS risk, early sexual activity, adolescent pregnancy, tobacco, alcohol, and drug use, and violence. The same three processes remain, but the key difference is that it is a theory-based planning model focused on certain health risk behaviours as the targets of change, built from a synthesis of research about factors correlated with those behaviours. Examples of those risk factors: child development problems, parent-child interaction, and family conflict and abuse; community poverty; previous victimization by violence or regular witnessing of it; the absence of a positive adult role model; and differences in personality traits, affective states, and behavioural skills. Most research points to multiple causes, cumulative exposure, or interactions among individual, family, and community risk factors over development: a web of risk factors.

Hawkins, Catalano, and colleagues synthesized this research into a widely used approach that functions as a planning tool (Hawkins, Catalano, and Miller, 1992). The model lays out an algorithm of factors, or forces, that over the youth development process increase or decrease the likelihood that a given youth will engage in problem behaviours: violence, delinquency, substance abuse, school dropout, HIV/AIDS risk behaviour, and others. Risk factors increase the likelihood; protective factors reduce it. It helps to think of the process as a box with inputs and outputs.

The domains

Under the Hawkins and Catalano model, risk factors are organized into domains, and protective factors, which are not as well specified, into a smaller set of similar domains. The table lists both sets with a description of what each domain covers.

Not every youth exposed to risk factors becomes involved in risk behaviour. Many do not, and they are said to have protective factors that counter or buffer the risks. Programs and research have focused on these under the general term resilience, and the more recent trend has been toward programs that support resilience or build protective factors. The complex, fluid interaction between an individual and the risk and protective factors in one or more domains has been described as a web of influence, an idea that traces back to Bronfenbrenner's ecological perspective: the thread that ties the planning half of this lesson back to the theory half.

Using the model: three phases

How is the approach used to plan, implement, and evaluate programs for adolescents and young adults? Thinking about it as we did for PRECEDE-PROCEED, there are three phases. We will follow a hypothetical community that wants to address adolescent violence and alcohol and drug use.

Two models, one spine

These two models are not the only planning models for health promotion programs. They were chosen to give a sense of what planning such programs involves and how it opens the door to selecting and applying theory. The key is the assessment, implementation, evaluation process, which helps ensure that a program is tailored to the real-life situation of a community. The comparison below answers the last question of the lesson: how are the two models similar, and how do they differ?

Read the table one more time with the multilevel theories in mind. Both models ask you to look at more than one level, both admit that you cannot act on every level at once, and both make you write down why you chose the factors you chose. That written rationale is what turns a multilevel theory from an abstraction into a plan.

Bringing It All Together

This lesson joined two halves that belong together. The first half argued that the theories reviewed so far, each working at one ecological level, cannot account for the way factors at different levels act on each other, and it offered theories built for that purpose. Bronfenbrenner's Ecological Systems Theory nests the developing person inside a microsystem of direct contact, a mesosystem that connects those settings, an exosystem the person never enters but is shaped by, a macrosystem of culture and law, and a chronosystem of time; its bioecological amendment added the person's own biology and the distinction between proximal and distal processes. Flay's Theory of Triadic Influence turned the layers into a map, sorting every factor by level of causation (ultimate, distal, proximal) and stream of influence (intrapersonal, social, cultural-environmental), and gave practitioners the waterfall: trace the influences upstream and intervene where it is feasible, remembering that proximal causes predict behaviour while distal and ultimate causes explain it. Problem-specific models from the CDC, WHO, and UNICEF reached the same conclusion from different directions, and every one of them carried the same critique: complexity.

The second half accepted the complexity and asked the practical question. You cannot address every factor, so you pick your battles, and it is best to do that with a planning framework. Every framework shares three components, assessment, intervention, and evaluation, linked in a feedback loop. PRECEDE-PROCEED works through them in nine phases: five assessments that narrow from the community's quality of life to the predisposing, enabling, and reinforcing factors behind behavioural and environmental risks and then to the resources and politics of implementation; implementation itself; and three evaluations, process, outcome, and impact. The risk and protective factors model, synthesized by Hawkins and Catalano for youth risk behaviour, runs the same three processes with a theory built in, sorting factors into risk domains and protective domains and treating problem behaviour as the output of a box whose inputs can be changed.

The thread that runs through both halves is the one the course has followed since Lesson 1. Behaviour is produced by interacting levels; a program that acts at one level alone is working against a current; and the discipline of assessment, whether through Bronfenbrenner's rings, Flay's grid, a PRECEDE phase, or a risk domain, is what turns a theory of that current into a plan for changing it.

Final Knowledge Assessment

This assessment covers all material from this lesson. You must score 100% to complete the lesson. Review the feedback for any incorrect answers and try again.