Close this search box.

Roadmap of Collective Research Projects

This page outlines a roadmap of coordinated research projects on family emotional process. It serves as a vision for how any number of compartmentalized, data-driven research efforts might combine into a broader body of knowledge. It is a working document that will be refined as needed.

This roadmap is an attempt to crowd source research on family emotional process. “Crowd sourcing” does not imply centralized control over any project or maintenance of confidential data. Instead, it is a vision of how research could proceed under a united vision. It is also an academic argument for certain gaps in the data driven scientific knowledge of family emotional process.

This vision was driven by the observation that most folks interested in family emotional process do not possess the technical training required to define falsifiable research questions. This lack of training can result in months and even years just getting to the start line of a formal research project, no matter how small. This roadmap defines small units of work that will contribute to a larger effort without getting discouraged from months of toil getting to square one.

It also serves as an outline to position more sophisticated research efforts into a broader body of knowledge. Dr. Stinson of Alaska Family Systems will provide free research supervision for anyone committed to being productive with their chosen piece of work.

Every project defined here is aimed at systematic data collection and inductive reasoning. There is no exception to this first rule of formal science and the standard for medical practice. Inductive reasoning has been covered in articles elsewhere on this site.

NOTE ON LEGAL AND ETHICAL CONSIDERATIONS FOR RESEARCH: The ideas contained in this page are provided for educational purposes only. Any person outside of Alaska Family Systems that performs research based on these ideas is responsible for managing any legal or ethical considerations for their own research on human subjects.

Please reach out with interest or questions using this form.

Table of Contents

  • Project 1: Collection of case formulations
  • Project 2: A definition of anxiety suitable for data collection
  • Project 3: Prerequisite Hypotheses for Research in Bowen Theory
  • Project 4: Identify the mechanism(s) of anxiety impacting a symptom
  • Project 5: One-page, jargon-free summary of behavioral health for medical specialists

Project 1: Collection of case formulations

The lack of a large sample of concise case formulations is a problem for research on family emotional process. The Family Diagram app solves the age-old problem of managing the complex kind of data required to argue such a case in concise form. What is needed now is multiple people accumulating a set of case formulations built in the app to demonstrate this problem and its solution.

At this stage, what formulations are presented are not as important as simply developing a set of formulations. For example, one case may argue that a child is focused upon. Another may argue that changes in anxiety were correlated with changes in a symptom. What matters now is more people developing competency in the app and demonstrating what a concise, data-driven argument looks like.

Alaska Family Systems will consult on the use of the app and organize such a sample of cases produced in the app. This project qualifies for free professional app licenses, and Dr. Stinson will donate consultation time to anyone interested in pursuing this. All cases can either remain confidential or consultation can be provided to modify the diagram to protect confidentiality.

The goals for each participant are outlined below:

Step 1: Enter a large number of cases into the app

Practice entering as many cases into the app as possible. Many professionals have access to many cases. These can be previous cases or current cases.

The focus is on quantity not quality. The primary goal should be getting comfortable entering cases into the app, moving from one case to another without going too deep. This task is like fly fishing, make only 5 casts, then move on to a new spot.

Competency will automatically come just by entering lots of cases. So will quality of the data. Keeping the analysis light will accomplish this.

Quality of data means a more concise list of events and people on the diagram. This data is used to argue the hypothesis that a particular emotional process has occurred in this family. The formulation must be clearly identified and the data must be presented for or against this hypothesis.

Step 2: Record short case formulation videos

After a number of cases have been collected, some cases will automatically emerge as a bit cleaner or clearer than others. Effort is then put into cleaning up these cases for a short 3-5 minute video case formulation. Getting the presentation to be this short will force a person to get clear on the case’s formulation.

A case does not have to prove a formulation or hypothesis. It can even be a null finding, i.e. there not being enough data to argue any theoretical formulation. Recording null findings is an important contribution to the theory at this stage.


There are more steps but only the first two are listed here so has to keep the eye on the immediate prize in the near term. For example, this simple set of case formulations is one step toward a research database on family emotional process. But, the Bowen Theory lexicon contains many broad terms and glorious language (e.g. “a science of human behavior”, “a theory based on facts alone”, “predict the outcome of therapy” etc.) that make it easy for researchers to get ahead of themselves with what is possible today.

A collection of concise video case formulations in a standardized format is a necessary contribution to family research. It is not common knowledge that collection exists for public consumption. Providing even a small set of formulations will help demonstrate this problem so that collective efforts can shift to address it. The app provides the standardized format and also a way to put the many dimensions of data into a single unified view for efficient presentation.

Project 2: A definition of anxiety suitable for data collection

The term anxiety is one of the most commonly used terms in Bowen Theory. It is one of the two main “variables” of the theory along with differentiation of self. But despite its pervasive use, there is no definition of the term that is precise enough for researchers to reliably collect volumes of data on it.

A data oriented definition would at least achieve demonstrated inter-rater reliability, if not mechanical precision. Inter-rater reliability would mean that multiple people would identify the same markers of anxiety in the same material such as interview transcripts, video recordings, timelines in the Family Diagram app, etc. Examples of mechanical measurement are a biometric device indicating when anxiety is going up and down or an absolute level, or a natural language model automatically coding an interview transcript or video recording for the same relative shifts or absolute levels of anxiety.

The de facto definition of anxiety in western culture more or less comes from psychiatry. That definition is not formally laid out, is quite subjective, and typically has something to do with arousal accompanied by an undesirable feeling of distress, overwhelm, feeling out of control of one’s own destiny, etc.

Aside from being subjective, the problem with the de facto definition for anxiety is that only applies to the human species. Bowen theory assumes a definition of anxiety that pertains to all species.

A person interested in this project would ask; What is a workable, data driven definition of anxiety? They would proceed collecting real world examples of phrases or behaviors that are “supposed to” indicate anxiety, proposing a definition for anxiety that matches those examples, presenting those examples and the definition to fellow researchers for critique, and then iterating from there so as to land on some workable definition.

If a single definition cannot be found, then a reliable set of examples can serve as a preliminary definition so long as a blind, independent rater can reliably determine if the single examples matches any one of the examples in the definition set.

Step 1: Collect supposed examples of anxiety

The first step is to collect examples of anxiety. This can be either markers of anxiety or indicators of shifts up or down in anxiety. These can be statements or behaviors from real world cases. Collection can happen on a word doc or in the Family Diagram app. The more examples the better.

Step 2: Play with a definition to summarize the examples

Come up with a short definition that captures the examples. Keep it no more than one sentence, ideally one phrase or even one word. If the definition doesn’t fit some examples, change it to fit. If multiple categories of anxiety emerge, note them.

Strive for a definition that is objective enough for any person on the street to use to identify anxiety in case examples. This person should be able to use this definition like a mindless robot without critical thinking. Eventually reach for a definition that a computer could use.

Step 3: Present the examples and definition for critique

This project requires many different perspectives on both the examples and working definition to refine it. Present both with the desire to poke holes in it. The goal is mechanical precision and so any hole should be smoked out like a disease infested rat.

Step 4: Iterate

This project may not be a one-and-done effort. It will likely take a lot of time, experimentation, and reflection to make progress. Plan for many repetitions of the experiment before insights are gained.


Though the term anxiety is so central to Bowen theory, it is too vague to generate large volumes of data for precise research. It should always be assumed that a project picking at a subjective definition like this will initially turn up more null results that positive or negative results. But keep in mind that even null results represent progress because they point to dead ends that can be left side in future efforts.

Food for thought on the problem of a definition of anxiety has been written here.

Project 3: Prerequisite Hypotheses for Research in Bowen Theory

A formal, data-driven case has not been made for funding research in Bowen Theory. Such a case would support the proposal that family relationships are involved in individual symptomology of some kind. Such a case would initially be focused on the individual so as to accommodate the way that the commercial world of medicine already thinks about medical problems.

A data-driven argument would be organized around evidence for the basic hypothesis of Bowen Theory, which is that the absolute level of individual anxiety partially modulates a presenting symptom, and that anxiety is typically modulated by close relationships, namely family (Havstad & Sheffield, 2018). “Anxiety” is focused as the entry point for this research. That is because this project is designed to occur in a medical context, where the paradigm and language is organized exclusively around A) pathology, and B) the individual.

This definition does not include level of “functioning” as mentioned in Havstad & Sheffield (2018) so as to keep the research hypothesis simple and the language mainstream. Anxiety may be the easiest part of Bowen Theory to communicate to those not familiar with the theory. That is, this project tests how often changes in the presenting symptom co-occur with changes in level of anxiety. The idea of “functioning” can hypothetically be considered at a later time as a moderator of anxiety, but it introduces too much complexity at this stage for a solid, straightforward research design.

This project is divided into three hypotheses to be tested on data from clinical assessment interviews. The three hypotheses are covered in detail here: Prerequisite Hypotheses for Research in Bowen Theory. They are to be tested in the order they are listed. The phrase “how often” was chosen carefully so as to emphasize the null hypothesis; that “anxiety” is not always involved in symptom modulation and that the relationship system is not always implicated in every fluctuation of every symptom.

These hypotheses were created under the pressure of producing income for the author. The need to produce income forces one to be very precise about goals and measurement of value provided to stakeholders. The goal was to avoid conclusions that rely on opinion and focus on evidence that not even the harshest sceptic could deny.

Step 1: Identify medical specialists interested in research

The researcher keeps their eyes open for medical specialists who are curious enough to welcome a researcher into their medical practice. Medical school has a way of forcing an MD to know the difference between fact and opinion within their own area of expertise. Nursing or PA training does not require the same level of scientific rigor, though there are always exceptions. Typically folks working the front desk of a medical practice do not possess an active interest in science or research. MD’s who run their own practices with many employees tend to have more intellectual capacity and curiosity available than those who are merely employed in a group or hospital.

Access to a medical practice provides access to patients in a harder scientific context without a referral to psychotherapy which is a much less scientific context. The researcher must hold themselves to this standard as opposed to the lower standards of conventional social science research. A medical practice also provides an opportunity to demonstrate a measurable value-add to a revenue stream should the hypothesis prove correct in enough cases.

Step 2: Approach the specialist with a research proposal

The researcher gets clear enough about the three hypotheses conceptually so as to provide an elevator pitch to any person and from any angle in conversation. They then write up a one-page description of the research design. This one page must start with a three-sentence comprehensive summary of the project, assuming that the provider has a very limited attention span.

Medical providers will usually ask for such materials. It is helpful to have them ready and be able to move on the project if the opportunity presents itself. A researcher should expect to volunteer their time initially so as to allow the most freedom in their work.

NOTE: Researchers working within professional practices must be familiar with patient confidentiality under HIPAA regulations, and the ethical code enforced by whatever professional licensing board they operate under. Possessing an academic credential certainly comes with increased credibility in this regard. Researchers without academic credentials are not legally restricted from conducting research on human subjects, but should expect due skepticism from any medical provider considering that provider would be legally and professionally responsible for any consequences of the research project.

Step 3: Conduct assessment interviews

The researcher conducts any number of assessment interviews about the context of the symptom. See the description of the hypotheses for what constitutes the context. Focus goes primarily to the first hypothesis of “context”. Once a connection to anxiety is observed, attention goes to the second hypothesis of “anxiety”. Then the hypothesis of “relationship”.

The researcher must maintain this frame of mind so as to pursue the null hypothesis throughout. Remember, the goal is to prove that each hypothesis level is not implicated in the symptom. The data must speak plainly for itself to those other than the researcher or the project is compromised.

At this stage, training in Bowen Theory will help greatly with the interview process. This includes having at least an intuitive grasp on what constitutes “anxiety” and how to lean into questions about context that may seem unrelated. This is required so long as there is no well-validated definition of anxiety for recording data (see above project on the definition of anxiety).

Record all interviews and use an automatic transcriber to produce text files from the audio recording. It is far easier to quickly scan text and pull out snippets compared to listening to interviews one minute at a time.

Remember, the hypotheses merely state that each level of context, anxiety, and relationship, are implicated in symptom expression. No assumption is made about how they are implicated or how much they are implicated. All biological research begins in this way, first that a connection exists must be established before one can wander into the infinite complexity of the measurement or exploring the mechanism at work. This goal here is merely to test whether or not research in the role of the family in a symptom is warranted, not to get into the family research itself.

Step 4: Track the process of interviewing

In this stage of exploratory research, there are always two projects at work; the content of the research question itself and the process of the research. In this case, whether or not context, anxiety, relationship, and eventually family are implicated in a particular symptom is the content of the research question. The process of the research has to do with refining the methodology as well as tracking any effect of the project itself.

In this case, the “effect” of the project would be whether or not simply conducting the interview changed the treatment plan for the medical symptom. The basic hypothesis of psychotherapy as clinical intervention is that the intervention produces change through insight. That is, that the patient will learn something about the nature of their symptom so as to adjust priorities more in line with reality.

At best, the hypothesis predicts that insight into the role of anxiety and relationships will help realign priorities so as to avoid wasting time and money on things that will not help. The goal is not to produce some magic curing intervention but to avoid wasting resources. If this hypothesis proves true, then a goal is to measure the resources saved, not the problems solved.

Step 5: Evaluate value-adds for stakeholders

The hypothesis about the process of the research is that the working model of the symptom will change through the process of the interview. This is an A-B, pre-post test of the effect of the interview itself as a clinical intervention. To measure this effect size, the researcher must identify the working model of the symptom before the interview, and then identify the working model of the symptom after the interview. This is a highly intellectual task related to the implicit model concept and will likely require consultation to complete.

The researcher must then attempt to identify:

  • The projected cost of a treatment plan based on working model A
  • The projected cost of a treatment plan based own working model B
  • The difference in the cost of treatment plans A and B

The third measure is a dollar amount is represents the value-add of the interview process. It may be that model B is more costly than model A. If that is the case, then there is no value-add. But if model B is less expensive than model A, a value-add has been found. The goal is to count the number of times that there was a value add, provide an average value add per case, and a total value-add across cases.

For example, a single mother sought medical treatment for her 8 year-old daughter’s painful nail biting. Her working model of the symptom before the assessment interview (i.e. model A) was, “It’s always been this way, it will never change, fix the pain.” Her working model after the assessment interview (i.e. model B) was, “It started when we fled the state to avoid my abusive boyfriend, she was attached to his parents, my own life is in chaos, and I am interested in exploring my own anxiety.” The working model had changed as an effect of the assessment interview itself.

The treatment plan based on working model A was: obtain prescription drugs for nail biting, schedule 2-3 follow-up appointments, prognosis is that nail biting will continue but will not be as painful. The cost of that treatment plan would be the overhead of this free clinic to provide the follow up appointments, the cost to the patient of the prescription drug. It was projected that the effect would be marginal considering the effect of the pain medication was partial and had undesirable side effects, and the original symptom would have persisted.

The treatment plan based on working model B was, mother attends 2-3 sessions of coaching, prognosis is that the symptom will decrease or abate. In this example, the daughter’s nail biting abated after the second coaching session with the mother and she did not follow up for a third. The cost of treatment plan B was zero because the assessment interview was conducted by a volunteer trainee in Bowen Theory and the intervention was “curative.” The effect of the interview “intervention” was compounded by there being no intervention other than further interviewing.

The effect of the interview intervention is observable once model A and model B are identified. Value-add is measured in the difference of treatment cost, where the cost of treatment A is easy to calculate and the associated prognosis is almost always not that great.

Stakeholders of this value-add must be identified. For example, who would have been responsible for the added cost of not conducting the interview? The provider? The insurance company? The patient? The patient’s pet parakeet? Once stakeholders are identified then the value-add can be presented to them for further iterations.

Step 6: Iterate

The goal of this project is to demonstrate value to stakeholders so as to produce a positive feedback loop of increased interest in the project from those stakeholders. If value cannot be demonstrated in measurable fashion then has the research truly produced a meaningful result? Enough increased interest from providers will automatically result in compensation for services.

Once value has been measured and a researcher is compensated for that value, the plan can scale to multiple providers and perhaps a plan funded by larger medical or professional entities. This allows the project to escape the traditional grant-based model that depends on large academic institutions or non-profits but does not create an incentive loop to achieve measurable outcomes. Such a for-profit model is self-sustained and attracts more talent and more resources for the research itself.


Real-world data is required to make the case for funding Bowen Theory research. Researchers must be clear on the basic clinical hypothesis of Bowen Theory, that enough medical cases are impacted by anxiety from functional position in the family system to warrant funded research on that variable. Such data must be free of jargon so as to be accessible by people not familiar with the theory itself.

Stakeholders must be identified for a measurable value-add of applying the theory. The theory offers no magic solutions but can point to the way avoid wasting time and money on solutions that yield questionable results. Providers in the medical field exist in a world of measurable, billable problems.

Legitimate data on the basics of Bowen Theory would speak loudly for itself in terms of dollars and cents. It would escape notions of research in Bowen Theory being suck behind non-Bowen theory aficionados not being able to “hear theory” or “see” anxiety or maturity. As in courts of law, the burden of proof lies on the prosecution, not the defense.

Project 4: Identify the mechanism(s) of anxiety impacting a symptom

That the absolute level of anxiety impacting symptoms is a fundamental claim of Bowen Theory. The idea of anxiety and its impact on health is an area of great interest to people interested in Bowen Theory. However, a formal argument has not been gathered in concise form.

This project is a review of published academic literature for or against the idea of anxiety impacting physiological problems. It would be a detailed look into the mechanisms that explain the relationship. For example, this would include literature on the impact of inflammation, arousal, threat response, and more on the ability to manage medical problems.

Knowledge about the biological mechanisms of anxiety would be combined with a definition for anxiety determined from the related project here that seeks a definition of anxiety suitable for data collection.

Project 5: One page, jargon-free summary of behavioral health for medical specialists

It is a fairly common assumption that “mental health” has an impact on a patient’s ability to manage many physical problems. However, how and when this is so is not well understood. A medical specialist can benefit from a simplified way of understanding how factors from this complicated world relate – or do not relate – to the problem at hand.

This project is about simplifying terminology. The terminology in Bowen Theory is far too vague and complicated for professionals outside the Bowen network to understand. This project assumes that this is due to a lack of precision in the language and theory, not a lack of interest or “emotional objectivity” in the audience.

Encapsulating the complexity of emotional process into a single page should do it. This page must only use measurable terms already known to the field. For example, it cannot contain the terms; triangle, emotion, anxiety, differentiation of self, reciprocity, child projection, Bowen, therapy, etc.

The line between when behavioral factors impact a patient’s ability to manage physical problems must be defined. It should rely on the state of research on biological mechanisms that relate behavioral and physical problems, but without academic. It would likely not assume the existence of mental or behavioral diagnoses, only that mental or behavioral factors impact medical problems.

Medical specialists, including PA’s, nurses, occupational therapists, etc. have narrowly defined scopes of practice and do not have the time for complicated psychological ideas. They need simple steps to follow in a clear framework. As of today, the ideas of Bowen Theory require a long, uncoordinated conversation to relay. Progress on this problem would improve both the clarity of what the theory can say and should say, and contact outside of the small network of people interested in it.

One consideration is that the Family Diagram app is perhaps in an ideal position to deploy such assessment steps. Such a tangible tool is likely required to force a simple evaluation routine. Perhaps the app needs to be modified for this purpose. All ideas are on the table.


Havstad, L., Sheffield, K., (2018). Study of weight loss as a model for clinical research: Shifts in the family system, the subjects functioning, and the course of clinical symptoms. Family Systems, 13(2), 9-31.

Share This

New From Alaska Family Systems

Get The Latest Updates

Subscribe For Notifications on New Posts

No spam, notifications only about new products, updates.

Sign Up For News

Stay current.