The Scientific Method: AB-PA

In my 2008-2010 analysis of the pathology,  I saw how the Gardnerian PAS diagnostic model was attacked as “junk science.”  I knew the pathogen would attack an attachment-based model with the same attack, so an attachment-based model needed to be solidly grounded in the scientific method, and it is.

The good thing about being a scientist is that we can have conversations with science.  She’s really quite helpful and supportive once you get to know her.

I’m a psychologist.  I’m a social scientist.  I’ve had extensive conversations with science over the years.   The key thing with science, is she always wants you to make a prediction that can be verified… or disproven. 

Here is my conversation with science regarding the AB-PA diagnostic model:


Dr. Childress: Science, I need your help.

Science:  What is it?

Dr. Childress: I have information that needs to be grounded in the scientific method, so I want to walk the information through you.

Science: That’s what I’m here for.  What’s the information?

Dr. Childress:  It’s a diagnostic model of pathology.

Science: What is the prediction that the diagnostic model makes?

Dr. Childress: It makes the prediction that a specific constellation of three highly unusual symptoms will be displayed in the child’s symptoms.

Science:  Okay. The null hypothesis (H0) is that there will not be a specific pattern of symptoms in the child’s symptom display, and the research hypothesis (H1) is that there will be a specific pattern of symptoms in the child’s symptom display.  The dependent variable are the three symptoms predicted by the diagnostic model.  How are you operationalizing the dependent variable, the three symptoms?

Dr. Childress: The criteria for the diagnosis is the presence of three specific symptoms in the child’s symptom display, 1) attachment system suppression toward a normal-range parent, 2) narcissistic personality traits in the child’s symptom display, and 3) an encapsulated persecutory delusion in the child’s symptom display.

Dr. Childress: For the first diagnostic indicator, the operational definition for “attachment system suppression” is a child evidencing a desire to sever the parent-child attachment bond.  The specific evidence are verbal statements by the child indicating a desire to sever the attachment bond to the parent, child behavior indicating a desire to sever the parent-child attachment bond (hostile-rejection or anxious-avoidance), and refusal of opportunities to spend time with the rejected parent.

Dr. Childress: Also for the first diagnostic indicator, the operational definition for “normal-range parent” is a category rating of Level 3 or Level 4 on the Parenting Practices Rating Scale.  This category rating is a clinical judgement by the mental health professional based on the clinical data from the family.

Science: Those are fine.

Dr. Childress: For the second diagnostic indicator, the narcissistic personality traits, the diagnostic model predicts five specific narcissistic personality disorder traits in the child’s symptom display, 1) grandiosity, 2) absence of empathy, 3) entitlement, 4) haughty arrogance, and 5) splitting.  The operational definitions for these symptoms is provided by the American Psychiatric Association in their diagnostic criteria.

Science:  Those are fine.

Dr. Childress: The operational definition for the third criteria of an encapsulated persecutory delusion is provided by the American Psychiatric Association in their diagnostic criteria.  On the Brief Psychiatric Rating Scale, the child’s delusion belief would be rated a 4 or a 5.

Science.  That’s fine.  Your structure is grounded.  Your operational definitions are grounded. 

Science: Describe the research hypothesis, (H1), why does the diagnostic model predict this set of symptoms, and why does this set of symptoms support this diagnostic model?

Dr. Childress:  The diagnostic model is grounded in the work of Bowlby (attachment), Minuchin (family systems), and Beck (personality disorder), as well as the entire professional literature from these domains of professional knowledge.  The diagnostic model is described in the book Foundations, which describes the pathology at three separate and integrated levels of analysis, 1) family systems level, 2) personality disorder level, 3) attachment trauma level.  Beyond describing these three independent levels of analysis for the pathology, the description of the pathology is integrated across all three levels, explaining all of the interactions of these three factors. The diagnostic model explains:

  • How the attachment pathology creates the personality disorder pathology;
  • How the personality disorder pathology then creates the family systems pathology;
  • How the attachment pathology creates the family systems pathology;
  • How the family systems pathology creates the attachment pathology;
  • How the personality pathology creates the attachment pathology;
  • How the family systems pathology triggers the personality pathology.

Dr. Childress: The three symptoms predicted from the diagnostic model are each highly unusual individually;

1.) Attachment Suppression: Suppression of a child’s attachment bonding motivations toward a parent is extremely rare (associated only with sexual and physical child abuse), and is an impossible symptom toward a normal-range parent.

2.) Personality Disorder Traits: The presence of five narcissistic personality traits in a child’s symptom display is rare. Children typically do not display personality disorder pathology. This symptom indicator is strengthened within the diagnostic model by requiring multiple narcissistic symptoms (5; the threshold for a DSM-5 diagnosis of a personality disorder) and by specifically identifying which five narcissistic symptoms will be evidenced.  There is only one plausible explanation for the presence of narcissistic personality traits in the child’s symptom display, i.e., the psychological influence and psychological control of the child by a narcissistic parent, that it is the parent’s narcissistic traits toward the other spouse-and-parent that the child is acquiring and expressing.

3.) Persecutory Delusion: This is an impossible symptom. It is impossible for a child to develop a persecutory delusion toward a normal-range parent.  Yet the diagnostic model predicts that this impossible symptom will be present in the child’s symptom display.  Children never spontaneously develop an encapsulated persecutory delusion toward a normal-range parent; it is impossible for a normal-range parent to create an encapsulated persecutory delusion in the child.

Science: The structure of the design is a three-way interaction of independent variables. The independent variables are correlated; statistical analysis considerations apply regarding correlated independent variables.  The levels of the independent variables are outside of experimental manipulation, so the design is a quasi-experimental, natural groups design.

Dr. Childress: Since this is a diagnostic model not a research design, do I need to measure the levels of the independent variables?

Science: No.  The presence of the three diagnostic indicators will confirm the diagnostic model that predicted the presence of these symptoms.

Dr. Childress: What is the scientific strength of this diagnostic protocol?

Science: Confirmation of predictions made by a three-way interaction of variables represent extremely strong confirmation of the diagnostic model making the prediction (H1). If the three a priori predicted symptoms from a three-way interaction are found, this data would confirm the causal attributions of the model.  A three-way interaction of variables means developing alternative explanatory models is extremely unlikely.

Dr. Childress: So, science… is everything good to go?

Science: Yes. The diagnostic model is structurally sound.  The scientific hypothesis (H1) is that a pattern of three symptoms will be found in the child’s symptom display.  The null hypothesis (H0) is that the three predicted symptoms in the child’s symptom display will not be present. The independent variables are three nominal data categories, family systems pathology, personality disorder pathology, and attachment pathology in the family. The dependent variables are the three symptoms in the child’s symptom display, attachment suppression, specified personality disorder traits, and an encapsulated persecutory delusion.

Science: Since you are not manipulating or measuring levels of the independent variables, this is not a research protocol.  It is a scientifically grounded diagnostic protocol.  The diagnostic model makes predictions that are confirmed or disconfirmed on an individual basis.

Science: Statistical cautions will emerge if the scientific basis of the diagnostic protocol is changed into a research protocol because the three independent variables of the diagnostic protocol are conceptually entangled and correlated with each other.  For a diagnostic protocol, the entanglement of the independent variables is not a relevant issue because the outcome dependent variable (the predicted symptom pattern) provides confirmation for the diagnostic model as a whole.  For a research protocol that examines the relative influence of each independent variable (IV) on the dependent variable (DV), disentangling the constructs for the IV and operationalizing their definition would be required.

Science: The scientific methodology for the diagnostic protocol of the three diagnostic indicators (H1) allows causal statements regarding the origin of the three symptoms when they are present in the child’s symptom display.  The presence of these three symptoms in the child’s symptom display would represent scientifically grounded proof for the causal associations described in the model that predicted these symptoms.

Science: Of note, there are simpler diagnostic models that would be grounded in the scientific method and allow causal diagnostic statements for the child’s symptoms. If attachment system suppression is impossible toward a normal-range parent except in cases of child abuse, then the presence of this single symptom alone, in the absence of child abuse, would represent confirmation for the diagnostic model that predicted this symptom.

Science: Alternatively, if an encapsulated persecutory delusion toward a normal-range parent is impossible, then the presence of this symptom alone in the child’s symptom display would represent confirmation for the diagnostic model that predicted this symptom.

Science:  It is notable that the diagnostic model makes a prediction of all three symptoms being present, and that two of the predicted symptoms are impossible.  Prediction of two impossible child symptoms and one improbable symptom that are then found in the child’s symptom display is solid confirmation for the diagnostic model that made the prediction.

Dr. Childress: What are the limitations to the diagnostic protocol?

Science: First, that the null hypothesis (H0) will be confirmed and the three child symptoms will not be evident in the child’s symptom display. Since two of the diagnostic indicators are impossible (attachment system suppression toward a normal-range parent and an encapsulated persecutory delusion toward a normal-range parent), the likelihood of the null hypothesis is 100%.   These three symptoms should never be found in a child’s symptom display.

Science: If, however, the three symptoms of the diagnostic model are found, then this represents complete confirmation for the diagnostic model that predicted these symptoms, and there exists one method for creating the previously impossible symptoms, the method described by the diagnostic model.

Science: The second limitation is that there may be alternative models that describe the presence of the three symptoms found in the child’s symptom display.  Since the predicted symptoms are the product of a three-way interaction of independent variables, the possibility of an alternative explanatory model are incredibly remote, with a likelihood next to non-existent.

Science: Within the scientific method, when a model has had its predictions confirmed by evidence, it is the responsibility of critics to propose alternative explanations.  Therefore, if the three diagnostic indicators are present in the child’s symptom display, this confirms the diagnostic model that made the prediction of these symptoms and it becomes the responsibility of critics to develop alternative explanatory models (if this is possible). That represents the advancement of knowledge through the scientific method. If alternate explanatory models are proposed, then examination of differing predictions made by the two models can differentiate which one represents an accurate description of the child’s symptoms.

Science: Third, this scientific model is diagnostic, meaning that it is applied one person at a time.  If it is true in one person, that doesn’t mean it exists anywhere else in the world, until the second case, and the third case are found, each having the same pattern of diagnostic indicators.  The prevalence of the pathology is unknown by the diagnostic protocol itself and would require further epidemiological research relative to public health concerns.  A diagnostic protocol only provides information about pathology on a case-by-case basis.

Dr. Childress:  Science, are there additional advantages to the diagnostic protocol.

Science: If the diagnostic model confirmed by the three diagnostic indicators remains undisputed by opposing evidence (an alternative explanation for the three symptoms being present in the child’s symptom display), then the diagnostic model and it’s three operationally defined diagnostic indicators represent an exceptionally good diagnostic independent variable (IV) to include in future research protocols.  Dependent variables (DV) of the investigator’s choice could be examined relative to two groups (2 levels of the IV), those with the three symptoms of the diagnostic model, and those without the three symptoms of the diagnostic model (a 1×2 natural groups design).  Alternatively, differing comparison groups could be selected from families in the foster care system, and families seeking services for school-related child behavior problems (a 1×3 design).  The diagnostic framework itself could be broken down into categories of independent variable (IV), such as families meeting all the criteria (high), families meeting some of the criteria (medium), and families meeting none of the criteria (low).  If a second factor was considered in the research design, such as the gender of the child, that would create a factorial design.  If two levels of the diagnostic category are used (presence-absence of the three symptoms) and two levels of child gender are used, that would create a 2×2 factorial natural groups design.  If three categories of the diagnosis IV were used, such as children in high-conflict divorce, children in the foster care system, and children with school behavior problems, and four levels of a second IV for parent-child relationship type (father-son, mother-son, father-daughter, mother-daughter) the design becomes a 3×4 factorial design.  The increasing number of cells in the factorial design would require larger numbers (N) of research participants for statistical purposes.  Because subjects are not being randomized to conditions, it would not be an experimental design, but would be a quasi-experimental design using naturally occurring groups. Since the diagnostic model is being used in the construction of the independent variable (IV), the nature of the dependent variable (DV) is entirely at the discretion of the investigators.  The DV might be a child characteristic, such as depression or anxiety as measured on various rating scales, or self-esteem as measured by a rating scale, or the child’s emotional and psychological functioning as measured by scores on a standardized projective test (like the Robert’s Apperception Test for Children).  Or the DV might be a quality of the parent’s pathology, such as the degree of splitting or a history of childhood trauma as measured by the Adult Attachment Interview.  Or the DV might be an aspect of the spousal-relationship, as measured by ratings on attachment compatibility and personality scales.  Multiple dependent variables can be included in any design structure, with the increasing number of independent and dependent variables impacting statistical analysis and increasing the number (N) of research participants required and recommended for statistical analysis. The diagnostic model of the three symptoms is a tool in structuring the independent variable (IV) of a research protocol.  How it is used to structure the independent variable (IV) and what dependent variables (DV) are examined are the choice of the investigator.

Science: The diagnostic model can also structure single-case research designs.  For example, treatment protocols using the single-case design could examine response to treatment or alternative treatments. The inclusion criteria into the single-case research would be the three symptom indicators of the diagnostic model.  ABA single case protocols are easily constructed for treatment (baseline data – intervention data – removal of intervention data).  When dealing with the treatment of pathology, the simple ABA protocol that leaves the child in a no-intervention condition is not recommended for ethical reasons.  In treatment related research using the single-case design, the ABAB reversal design is used (baseline-intervention-baseline-intervention; ABAB). 

Dr. Childress: Thank you. So am I good to go?

Science: You’re good to go.


Craig Childress, Psy.D.
Clinical Psychologist, PSY 18857

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s