Phase 17 Reliability Pilot — Eight-Source Coding Test and Codebook Revisions

Design limitation: This is a single-researcher stability and ambiguity pilot, not a true inter-rater reliability study. No Cohen’s kappa or agreement percentage is reported because an independent second coder has not yet applied the codebook.

Pilot sample

1. Lydia Maria Child, The Mother’s Book (1831) — domestic/moral manual.


2. L. Emmett Holt, The Care and Feeding of Children (1901) — pediatric manual.


3. John B. Watson, Psychological Care of Infant and Child (1928) — behaviorist manual.


4. Benjamin Spock, The Common Sense Book of Baby and Child Care (1946) — pediatric/psychoanalytic popular manual.


5. Thomas Gordon, Parent Effectiveness Training (1970) — communication skills.


6. James Dobson, Dare to Discipline (1970) — evangelical/psychological authority.


7. 
   Ebony special issue on Black children (August 1974) — magazine/editorial corpus.


8. James Comer and Alvin Poussaint, Raising Black Children (1992) — Black child development and social context.

First-pass coding results

Ambiguities exposed

1. 
   Causal versus relational responsibility: Watson makes parents causally responsible for outcomes while teaching little reciprocal relationship work. These must be separate fields.


2. 
   Mutual but asymmetrical: nineteenth-century mutuality based on parent affection and child obedience differs from modern mutual influence plus greater adult responsibility. One code obscures this difference.


3. 
   Warmth versus connection: affection can be a tool of moral influence rather than recognition of the child’s independent account. Warmth alone cannot stand in for relational accountability.


4. 
   Expert supplementation versus validation: a source can both train parents and validate their authority. Multi-coding is required.


5. 
   Social-system responsibility: the original relationship-responsibility field inadequately captures racism, poverty, school systems, and coercive state action.


6. 
   Audience versus burden: a book addressed to “parents” may operationally assign nearly all work to mothers. Directed audience and practical responsibility must be coded separately.


7. 
   Silence: absence of apology language cannot be coded as opposition to apology. “Not discussed” is mandatory unless resistance is explicit.


8. 
   Magazine issue versus article: an issue’s table of contents establishes editorial agenda but not each article’s argument. Evidence-unit level must be recorded.

Codebook revisions — Version 1.1 supplement

New field A: Type of parental responsibility

• Material/provision.


• Moral/legal authority.


• Causal-developmental.


• Technical/managerial.


• Emotional caregiving.


• Relational maintenance.


• Repair after conflict.


• Institutional advocacy.

New field B: Form of asymmetry

• Role-based asymmetry: parent’s rank creates child duties.


• Capacity-based asymmetry: parent’s maturity and power create greater parent duties.


• Mixed asymmetry: both logics present.

New field C: Function of warmth

• Bond/comfort.


• Moral influence and compliance.


• Developmental security.


• Communication access.


• Repair/reconnection.


• Unspecified affection.

New field D: Level of analysis

• Child.


• Parent.


• Mother–child dyad.


• Father–child dyad.


• Marital subsystem.


• Whole family.


• School/clinic/institution.


• Social-economic-cultural system.

New field E: Evidence unit and access depth

• Full text; substantial excerpt; abstract/review; table of contents/index; bibliographic record.


• Manual-level, chapter-level, article-level, letter-level, issue-level, institutional-policy level, remembered narrative.

Stability assessment

The codebook successfully distinguishes major historical models, but first-pass ambiguity was highest for relationship responsibility and expert role. The new fields reduce forced either/or judgments and preserve coexistence of hierarchy, affection, developmental causation, and institutional constraint.

A later independent-coder exercise should use at least twelve sources and calculate agreement for categorical fields. Until then, all synthesis claims should rely on transparent excerpts and confidence labels rather than quantitative reliability.

Pilot verdict

The codebook is sufficiently stable for the comparative argument and case-study phases, provided the Version 1.1 supplement is used and no numerical reliability claim is made.