Your data is collected, so the literature review now does two jobs. It frames your findings against what the field already knows, and it protects you from the reviewer who asks whether you missed something. Both jobs reward a search you can reproduce and describe, not one you did by memory. This section walks through databases, systematic searching, citation chasing, knowing when to stop, reference management, and how to keep track of everything you read.

The databases that matter for infant emotion research

Each database indexes a different slice of the literature. Searching only one leaves gaps that a committee member will find.

• PsycINFO (APA) is your primary source. It has the deepest coverage of developmental and clinical psychology, and it uses a controlled vocabulary, the Thesaurus of Psychological Index Terms, that lets you search by concept rather than by whatever word an author happened to choose. This is where emotion regulation, temperament, and attachment work lives.
• PubMed / MEDLINE covers the biomedical and neuroscience side. Use it for physiological measures (vagal tone, RSA, cortisol, HPA axis), pediatric samples, and anything published in medical or infancy journals. Its controlled vocabulary is MeSH.
• Web of Science is less about subject coverage and more about citation structure. Its strength is forward citation chasing (the "Times Cited" and "Citing Articles" links) and mapping how a paper has been used since publication. Scopus does the same job if your library has it instead.
• ERIC is the education database. It is worth a pass for early childhood, caregiving, childcare settings, and parent intervention work that psychology databases index thinly.
• Google Scholar is broad and catches preprints, dissertations, book chapters, and gray literature the indexed databases miss. Treat it as a supplement, not a base. It has no controlled vocabulary, no reliable field limits, and its relevance ranking is opaque, so you cannot describe a Scholar search as systematic. Its real value is the "Cited by" link for forward chasing and finding a hard-to-locate full text.

Building a systematic search

A good search is built, not typed. Break your question into concept blocks, usually two to four. For a dissertation on infant emotion regulation, your blocks might be: the population (infants, neonates), the construct (emotion regulation, affect regulation, self-regulation), and possibly a paradigm or measure (still-face, cortisol, vagal tone).

Within each block, list every reasonable synonym and combine them with OR. Then combine the blocks with AND. That is the whole logic.

Controlled vocabulary versus keywords

Use both, deliberately. Keywords (free-text words in title and abstract) catch brand-new terms and idiosyncratic phrasing. Controlled vocabulary (PsycINFO subject headings, MeSH terms) catches papers regardless of the author's word choice and handles the fact that "affect regulation" and "emotion regulation" are often the same construct. A strong search combines a subject-heading search with a keyword search on the same concept using OR. Look up the exact heading in the database thesaurus rather than guessing.

Truncation and wildcards

infant* retrieves infant, infants, infancy. regulat* retrieves regulate, regulation, regulatory, dysregulation is not caught, so add it. Phrases go in quotation marks so "still-face" is treated as a unit. Note that operators and wildcard symbols differ slightly across platforms, so confirm each one in the database's help before trusting your yield.

Snowballing: chasing citations in both directions

Database searches miss things. Citation chasing catches them, and it is how experienced researchers find the papers everyone in a subfield actually cites.

• Backward snowballing means mining the reference list of a key paper or a recent review. This walks you back toward the foundational work. When you find a strong review or meta-analysis on infant emotion regulation, its reference list is a curated map of the field.
• Forward snowballing means finding everything that has cited a key paper since it was published. This walks you forward to current work and shows you how a construct or measure evolved. Use Web of Science "Citing Articles," Scopus "Cited by," or Google Scholar "Cited by" for this. Connected Papers and Litmaps are visual tools that build a citation network around a seed paper, which is a fast way to spot clusters and the obvious papers you have not read yet.

A practical loop: pick two or three anchor papers you know are central, chase backward from their reference lists and forward from their cited-by lists, add anything relevant to your library, then repeat from the newly added anchors until you stop finding new names.

Knowing when the review is complete enough

You are done when the search is saturated, meaning new searches and new citation chains keep returning papers you already have. The concrete signals:

1. The same author names, samples, and instruments recur across independent searches.
2. Backward and forward chasing from new anchors surfaces nothing you have not logged.
3. You can state the major theoretical positions and the main empirical disagreements without gaps.

Saturation is about the conceptual space being covered, not a paper count. A focused dissertation construct might saturate at 60 well-chosen sources. Document the point where returns flattened so your methods narrative has a defensible stopping rule.

Reference management

Zotero is the recommended tool and it is free and open source. It captures citations from your browser with one click, stores the PDF alongside the record, generates APA 7 references and in-text citations, and plugs into Word or Google Docs. EndNote (paid, often provided through university licenses) and Mendeley are the main alternatives and do the same core job, so if your library already supports one, use it. The tool matters less than using one consistently from the start.

Organizing the library

• Attach the PDF to every record and let Zotero rename files consistently. A folder of "download(3).pdf" files is unusable in a year.
• Use collections for structure (by chapter, by construct, by measure) and tags for cross-cutting attributes a single collection cannot capture, for example still-face, longitudinal, physiological-measure, to-read, key-paper.
• Fix metadata as you import. Zotero pulls author, year, and journal automatically but often mangles capitalization and page ranges, and a bad record produces a bad reference. Clean it once at import, not at 2 a.m. before submission.

Tracking what you read: the synthesis matrix

Reading without a structured record means rereading. Build a synthesis matrix (also called a literature grid), one row per study, with columns you can compare across. A workable set for infant emotion regulation:

• Citation (author, year)
• Sample (n, age, population, recruitment)
• Design (cross-sectional, longitudinal, experimental)
• Measures / paradigm (still-face, RSA, cortisol, observational coding, parent report)
• Key findings
• Limitations and gaps
• Relevance to my study

Read down a column instead of across a row and the review writes itself. The Measures column shows you which paradigms dominate and which are thin. The Gaps column becomes the justification for your own study. A spreadsheet is enough. You do not need special software.

Narrative review versus systematic review

These are different products and a dissertation can use both in different places.

A narrative review synthesizes and argues. It is selective, organized by theme or theory, and it builds toward your rationale. This is what most dissertation literature chapters are, and it is the right form when your goal is to situate your study and defend your hypotheses. Its rigor comes from fair coverage and honest treatment of contradictory findings, not from an exhaustive protocol.

A systematic review answers a tightly bounded question using a pre-specified, reproducible protocol: defined inclusion and exclusion criteria, documented searches across named databases, screening at title/abstract then full text, and transparent reporting (PRISMA is the standard guideline, and a meta-analysis adds quantitative pooling on top). It fits when a whole chapter or a standalone paper is the systematic review itself, for example "a systematic review of physiological measures of infant emotion regulation."

You have already run your study. What is left is thinking, structuring, and writing, and this is exactly where an AI assistant earns its place or ruins your credibility. The line is simple. AI is good at operating on text and ideas you supply and can verify. It is dangerous the moment it becomes a source of facts you take on faith.

What AI Is Genuinely Good For

Treat Claude Code as a fast, tireless collaborator who is well read but occasionally makes things up with total confidence. Within that frame, it is strong at:

• Brainstorming search terms and synonyms. Generating alternate keyword strings, MeSH-style term clusters, and spelling variants to run through PsycINFO, PubMed, Google Scholar, or Web of Science. You still run the searches yourself.
• Restructuring arguments you have already made. Taking a rambling paragraph and proposing three cleaner orderings, or showing you where your logic skips a step.
• Summarizing and reorganizing your own notes. Give it your coding memos or reading notes and ask for an outline. This is safe because the content originated with you.
• Explaining a statistical method. Walking through what a multilevel model, a growth curve, or an intraclass correlation is doing, in plain language, so you can decide whether it fits your design.
• Draft-then-heavily-edit. A rough first pass of a transition, a limitations paragraph, or a cover letter that you then rewrite in your own voice.
• Analysis code. R or Python for cleaning, reshaping, plotting, and modeling. You read every line and confirm the output makes sense against your data.
• Mechanical APA checks. Flagging formatting issues in a reference list, headings, or table structure. Verify anything it claims about the current APA edition against the manual.

The Hallucinated Citation Problem

This is the single failure mode most likely to damage a dissertation. Language models generate text that is statistically plausible, not text that is true. When you ask for a citation, the model can produce a reference that looks completely real, correct author style, believable journal, plausible year, formatted perfectly in APA, that does not exist. It can also attach a real author to a paper they never wrote, invent a DOI, or fabricate a direct quote and attribute it to a real researcher. These fabrications are convincing precisely because the model is good at imitating the surface form of scholarship.

How to verify, every time

1. Find the actual source. Search the DOI or exact title in your library databases or Google Scholar. If you cannot locate it, assume it is fabricated.
2. Open the real paper. Confirm the authors, year, journal, and volume match. AI often gets one field wrong.
3. Read the relevant passage yourself. Confirm the paper actually says what the AI claimed. Do not trust an AI summary of a paper as grounds for citing it.
4. Verify every quotation character for character against the original, including the page number.
5. Re-derive every statistic from the primary source or your own output. Never cite a number the AI produced.

A safer workflow: do not ask AI to give you citations at all. Do your own literature search, then paste the abstracts or full text of papers you have already found and ask the AI to help you compare, organize, or synthesize them. Now it is working from real sources you control.

Good Prompting Patterns for Research

• Give it context. Paste your actual notes, your data description, your draft section. Grounded prompts produce grounded, useful output. Vague prompts invite invention.
• Ask it to critique, not agree. Models default to being agreeable. Explicitly instruct it to find weaknesses, disagree, and push back.
• Ask for counterarguments and alternative explanations. Especially for causal claims about infant behavior, where confounds and alternative interpretations are the reviewers' first target.
• Iterate. Treat the first answer as a draft. Respond with "too generic, tie it to my design," or "you missed the nesting of observations within infant."
• Ask it to show its reasoning so you can check the logic rather than just the conclusion, particularly for anything statistical.

Copyable prompts

Academic Integrity and Disclosure

Before you rely on AI for anything that touches the dissertation, confirm what your program, your advisor, and your committee expect. Norms differ by institution and are changing, and CUNY, your department, and your chair may each have a position. Some committees want a disclosure statement describing how AI was used. Ask directly rather than assume.

The governing principle: AI is a thinking aid, never a ghostwriter. The intellectual work, the argument, the interpretation of your infants' behavior, the theoretical positioning, must be yours. If you cannot explain and defend a sentence in your own words, it does not belong in your document. The voice a committee reads in a defense should match the voice on the page. Heavy AI drafting produces a smooth, generic register that experienced readers notice, and that you will struggle to defend live.

Reproducibility and Logging

• Keep a log of substantive prompts and outputs that shaped your analysis or writing, dated. If a committee or a journal asks how AI was used, you can answer precisely. It also lets you retrace a decision months later.
• Version your analysis code and note where AI contributed, so any result can be reproduced from raw data through to output by a human reading the script.
• Never let AI touch conclusions about your raw data without human verification. It can help you write the code and interpret output you have checked, but the inferential judgment, whether an effect is real, what it means for infant regulation, whether an assumption holds, stays with you. Re-run the key analyses yourself and confirm the numbers in your text match your actual output.

Used this way, Claude Code makes you faster at the parts of the dissertation that are genuinely yours. Used carelessly, it inserts confident errors into a document that will be scrutinized line by line. The discipline of verifying every fact and keeping the voice your own is what separates the two.

This is a fast orientation to the scholarly terrain around infant emotional regulation. You know this ground. Use it to check that a construct you are writing about is defined the way the field defines it, to find the right seminal name to cite, or to locate where your own contribution sits relative to the live debates.

Core Constructs and How They Differ

Precision here matters because reviewers in this area are unforgiving about construct slippage. The most common critique of infant regulation papers is that reactivity and regulation get collapsed.

Reactivity versus regulation

Emotional reactivity is the threshold, intensity, latency, and rise time of an emotional response to a stimulus. It is the arousal that arrives before any modulation. Emotion regulation refers to the processes that shift the trajectory, intensity, or duration of that emotional response, whether up or down. In infancy the honest position is that these are partially confounded in observation. A baby who looks calm may be low in reactivity, high in regulation, or both. A strong analysis states which behaviors it is treating as reactivity indices and which as regulatory, and defends the split.

Self-regulation versus co-regulation

Self-regulation is the infant's own contribution: gaze aversion, self-soothing behaviors like thumb sucking, self-distraction, orienting away from a distressing stimulus. Co-regulation (dyadic regulation) is the caregiver and infant regulating the infant's state together. In early infancy regulation is overwhelmingly dyadic. The caregiver supplies most of the regulatory scaffolding, and the infant's independent capacity emerges gradually across the first years. Treating an infant's regulation as a purely intra-individual trait, without accounting for the dyad, is a conceptual error that reviewers flag.

Temperament and effortful control

Temperament is early-appearing, biologically based individual differences in reactivity and regulation, relatively stable and present before socialization has had much time to act. Effortful control, in Rothbart's model, is the temperamental regulatory system: the capacity to voluntarily inhibit a dominant response, activate a subdominant one, and focus or shift attention. Effortful control is generally considered to come online later, since it depends on maturation of attentional and executive systems, so in early infancy attention-based regulation (orienting, disengaging) is the more developmentally plausible precursor.

Distress tolerance

Distress tolerance is the capacity to withstand aversive emotional states without escalating or resorting to maladaptive escape. In infants it is usually operationalized through recovery: how quickly and how completely the baby returns to baseline after a distress elicitor, and whether the baby can remain organized while distressed rather than becoming fully dysregulated.

Foundational Theoretical Frames

Functionalist view of emotion

The functionalist perspective, associated with Campos, Barrett, and colleagues, treats emotions as adaptive processes that establish, maintain, or change the relationship between the person and the environment in service of goals. Emotions are not noise to be suppressed. They organize behavior. Under this frame, regulation is not simply dampening negative affect. It is modulating emotion to serve the infant's current goals, which reframes what counts as a "good" regulatory outcome.

Dynamic systems and developmental science

The dynamic systems perspective, drawn from Thelen and Smith's work on development, treats regulation as an emergent property of many interacting components (physiological, attentional, motor, dyadic) rather than a single module. Behavior self-organizes in real time out of these interactions. This frame is useful when you want to talk about within-dyad variability, moment-to-moment state transitions, and why the same infant regulates differently across contexts.

Attachment theory and regulation

Attachment theory (Bowlby, Ainsworth) connects to regulation through the idea of the caregiver as a secure base and a regulator of infant arousal. Sroufe's developmental work, including the Minnesota longitudinal study, positioned emotional regulation as growing out of the history of dyadic regulation in the attachment relationship. Secure relationships are theorized to support the development of flexible regulation. The Strange Situation is the classic attachment paradigm, and its regulatory relevance lies in how infants manage distress across separation and reunion.

Physiological and polyvagal ideas

Porges's polyvagal theory ties parasympathetic (vagal) control of the heart to the capacity for calm engagement and flexible regulation, and to social engagement more broadly. In practice this line of work motivates the use of respiratory sinus arrhythmia (RSA) and vagal tone as physiological indices of regulatory capacity, including vagal withdrawal during challenge and recovery afterward. Treat the broader theory and the specific measures as separable. You can use RSA as an index without committing to every theoretical claim.

The Mutual Regulation Model

Tronick's Mutual Regulation Model holds that infant and caregiver form a communicative system that continually moves between coordinated (matched) and miscoordinated (mismatched) states, with repair of mismatches being central to healthy development. This is the theoretical home of the Still-Face Paradigm and a natural frame if your data involve dyadic interaction and disruption.

Seminal Researchers and Lines of Work

• Ed Tronick. Still-Face Paradigm and the Mutual Regulation Model. Dyadic coordination, mismatch, and repair.
• Ross Thompson. Conceptual and developmental work on the emergence of emotion regulation, including influential definitional writing on what regulation is and how it develops.
• Mary Rothbart. Temperament theory (reactivity and self-regulation), effortful control, and the Infant Behavior Questionnaire tradition.
• Nathan Fox. Temperament, EEG frontal asymmetry, and the physiology and development of reactivity and regulation, including longitudinal work on early temperament.
• Jerome Kagan. Behavioral inhibition, the reactive infant profiles, and the biology of temperamental fear and restraint.
• Megan Gunnar. Cortisol, the HPA axis, and the developing stress-response system in infancy and early childhood.
• Alan Sroufe. Organizational perspective on emotional development and the attachment roots of regulation.

Classic Paradigms and Measures

Behavioral paradigms

• Still-Face Paradigm. Three episodes of normal interaction, an unresponsive still face, and reunion. The still-face effect (increased negative affect, gaze aversion, self-soothing) and reunion behavior are the workhorse indices of dyadic regulation and repair.
• Strange Situation. Structured separations and reunions used primarily for attachment classification, with regulatory relevance in how distress is managed and resolved.
• Lab-TAB style episodes. The Laboratory Temperament Assessment Battery provides standardized elicitors of fear, anger, joy, and other states, with detailed behavioral coding of intensity and regulatory behavior. It is the standard for eliciting temperament and emotion under controlled conditions.

Coding and questionnaires

• Microanalytic behavioral coding of affect, including facial coding systems and second-by-second coding of gaze, affect, and regulatory behaviors. This is where inter-rater reliability and coding-scheme transparency get scrutinized.
• Infant Behavior Questionnaire (IBQ and IBQ-R). Rothbart's parent-report instrument for infant temperament, capturing dimensions of reactivity and, in the revised version, regulatory and orienting capacities. The Early Childhood Behavior Questionnaire and Children's Behavior Questionnaire extend this tradition to older ages.

Physiological measures

• Cortisol. Salivary cortisol indexes HPA-axis stress reactivity and regulation. Sensitive to timing, diurnal rhythm, and context.
• RSA, vagal tone, heart rate variability. Parasympathetic indices of regulatory capacity, including baseline tone and vagal withdrawal or augmentation during challenge.
• EEG frontal asymmetry. Relative left versus right frontal activity, linked to approach and withdrawal motivation and to individual differences in reactivity and regulation.

Key Journals to Know

• Child Development and Developmental Psychology. Flagship general developmental outlets.
• Infancy and Infant Behavior and Development. Specialist infancy venues, close to this exact topic.
• Emotion. Home for affective-science framing of your constructs.
• Developmental Science. Mechanistic and often physiological or cognitive developmental work.
• Development and Psychopathology. Where regulation connects to risk, resilience, and later outcomes.

Live Debates and Gaps a Strong Dissertation Might Engage

• Measuring regulation in preverbal infants. With no self-report, every regulation index is an inference from behavior or physiology. Debates continue over which behaviors are genuinely regulatory versus merely reactive, and how to validate them without circularity.
• Disentangling reactivity from regulation. The two are statistically and observationally entangled. Work that offers a defensible analytic separation, or that models them jointly rather than pretending they are independent, is doing something the field values.
• Cultural variation. Much of the canonical work rests on Western, often North American samples. Norms for caregiver responsiveness, acceptable infant distress, and what counts as adaptive regulation vary across cultural contexts, and the generalizability of paradigms like the Still-Face and instruments like the IBQ is an open question.
• Individual differences and trajectories. Stability, change, and prediction from early regulation to later outcomes, including who moderates whom in the dyad, and whether early regulatory profiles forecast later psychopathology or competence.

Your data is in. The work now is to treat it with the care it deserves: understand its shape before you model it, choose analyses that respect its structure, and document every decision so the pipeline holds up under scrutiny. Infant and dyadic data are messy in specific, predictable ways. This section is organized around those realities.

Get to Know Your Data Before You Model It

Before any inferential test, spend real time looking at the raw data. For observational infant work this means opening the coded files, plotting distributions, and reading a handful of cases end to end. You are checking that the numbers match what you remember from the sessions.

Data cleaning and screening

Screen for impossible values (a heart rate of 400, a coded state that does not exist in your scheme), duplicated rows from double entry, and out of range timestamps in event coded data. For infant studies, note attrition and its reasons separately from missing data on a present infant. A baby who fussed out at minute two is a different kind of missing than a sensor that dropped.

Missingness

Think about why data are missing before you decide how to handle it. The standard conceptual distinctions:

• MCAR (missing completely at random): missingness unrelated to anything, for example a lab computer crash. Listwise deletion is unbiased here but still costs power.
• MAR (missing at random): missingness depends on observed variables, for example younger infants fuss out more and you measured age. Multiple imputation or full information maximum likelihood (FIML) recovers unbiased estimates.
• MNAR (missing not at random): missingness depends on the unobserved value itself, for example the most dysregulated infants terminate the still-face before you can code recovery. This is the hard case and it cannot be fully fixed statistically. Name it, reason about the likely direction of bias, and consider sensitivity analyses.

Coding reliability for observational data

If your outcomes come from behavioral coding, reliability is not a footnote, it is a load bearing claim. Establish it on a meaningful subset (commonly 20 to 25 percent of cases) double coded by independent raters blind to hypotheses and condition.

• Cohen's kappa for categorical codes (affect category, gaze on/off, discrete behavioral states). Kappa corrects for chance agreement, which raw percent agreement does not. For ordered categories use weighted kappa.
• Intraclass correlation (ICC) for continuous or rating scale codes (global affect intensity, sensitivity ratings). Specify which ICC you used. The choice among one-way versus two-way, random versus mixed, and agreement versus consistency depends on your rater design, so report it explicitly rather than just quoting a number.
• For time locked microanalytic coding, consider whether you need reliability on the presence of an event and on its precise timing. Kappa on time binned streams is common for still-face and free play coding.

Descriptives and visualization

Report means, SDs, ranges, and distribution shape for every variable that enters a model. Plot before you test: histograms for skew, boxplots for outliers, and for repeated measures data plot individual infant trajectories (spaghetti plots) so you can see heterogeneity that group means hide. Physiological and cortisol data are often skewed and benefit from transformation, which you should decide on and justify before modeling, not after seeing p values.

Choosing an Analytic Approach

Match the method to the structure of the data, not to habit or to what the last paper used.

• Correlation and regression. Fine for one observation per infant and a continuous outcome. The moment you have multiple observations per infant, ordinary regression violates independence and understates your standard errors.
• ANOVA. Still reasonable for clean factorial designs with a single outcome per cell. Repeated measures ANOVA handles within infant designs but is rigid about missing data (it drops any infant with a missing cell) and assumes sphericity. Mixed models usually do the same job with fewer casualties.
• Multilevel / mixed-effects models. The workhorse for infant data. Observations nested within infants, infants nested within families or dyads, repeated still-face episodes within a session. Random intercepts (and where justified, random slopes) account for the dependence and let you keep infants with partial data under a MAR assumption. This is usually the right default for your kind of data.
• Growth curve modeling. For genuinely longitudinal data (the same infants across ages), model the trajectory itself. You can specify linear or nonlinear growth, estimate individual differences in intercept and slope, and predict them from covariates. Implementable as multilevel models or in an SEM framework (latent growth curves).
• Structural equation modeling (SEM). When constructs are latent and measured by multiple indicators, or when you have mediation or a system of relations to test simultaneously. Report fit indices (for example CFI, RMSEA, SRMR) rather than leaning on chi-square alone.
• Actor-Partner Interdependence Model (APIM). Built for dyadic caregiver-infant data. It separates the effect of a person's own predictor on their own outcome (actor effect) from the effect of the partner's predictor on their outcome (partner effect), while respecting the nonindependence of the dyad. If your question is about mutual influence in the dyad, this is the framework.
• Time-series and microanalytic coding. For still-face and other second-by-second data, the temporal dynamics are often the point. Depending on the question, this can mean modeling recovery slopes, examining lagged contingency between infant and caregiver behavior (for example sequential analysis, cross-lagged or dynamic models of moment-to-moment influence), or state space and time-series approaches. The still-face paradigm's classic signature (the still-face effect and the carryover into reunion) lives in exactly this kind of within-episode change.

Physiological Measures at a High Level

If you collected physiology, the analytic decisions happen before the modeling, in how you reduce the signal.

• RSA (respiratory sinus arrhythmia). A common index of parasympathetic (vagal) regulation. The two quantities usually of interest are baseline RSA and RSA reactivity, the change from baseline to a challenge such as the still-face. Withdrawal (a decrease) versus augmentation matters conceptually, so define your change score direction clearly and decide how you handle baseline dependence.
• Cortisol. Salivary cortisol is slow moving, so sampling schedule and timing relative to the stressor are everything. It is typically right skewed and log transformed. For repeated sampling across a session, area under the curve is standard: AUC with respect to ground (total output) and AUC with respect to increase (reactivity relative to the first sample) answer different questions. Choose deliberately and report which.

Statistical Rigor

• Effect sizes and confidence intervals over bare p-values. Report the magnitude of what you found with an interval around it. With the modest samples typical of intensive infant work, a nonsignificant result is often an imprecise one, not a null one, and a CI shows that honestly.
• Power and precision. Even post hoc, be candid about what your N can and cannot detect. For mixed models, the effective sample size depends on both the number of infants and observations per infant.
• Confirmatory versus exploratory. Draw a bright line. State which analyses tested your a priori hypotheses and which are exploratory follow-ups. Both are legitimate. Presenting an exploratory finding as if it were predicted is not.
• Avoid p-hacking and HARKing. Do not try covariate sets until significance appears, and do not rewrite the hypothesis to match a surprise result. If you found something unexpected and interesting, label it exploratory and flag it as needing replication.

Software

R is the recommended environment for this work. The relevant pieces: lme4 (and lmerTest for p-values) for mixed and growth models, tidyverse for cleaning and reshaping, lavaan for SEM and latent growth curves, and psych or irr for reliability including ICC and kappa. Mplus is strong for SEM, complex latent variable models, and some dyadic designs. SPSS works for ANOVA, regression, and mixed models through a menu driven interface if that is your existing fluency.

How Claude Code helps here

Claude Code is good at the mechanical and translational parts of analysis: drafting a cleaning script, reshaping wide to long for mixed models, writing lmer syntax, debugging a convergence warning, or explaining what an SEM fit index means. Use it to move faster on code, and keep the statistical judgment yours.

Open Science and a Reproducible Pipeline

• Confirmatory versus exploratory, on the record. Where a hypothesis was set in advance, a preregistration (for example on OSF or AsPredicted) makes the confirmatory or exploratory distinction credible rather than a claim after the fact. Since your data are already collected, be precise about timing: preregister analyses you have not yet run, and be transparent that collection preceded it.
• Share code and materials. Coding manuals, the analysis scripts, and (where consent and IRB allow for human infant data) de-identified data or synthetic equivalents. Infant video data usually cannot be shared, but coding schemes and derived numeric data often can.
• Document the pipeline. Someone should be able to go from raw file to final table by running your scripts in order. Number your scripts, keep raw data read-only and untouched, write cleaning and analysis as separate steps, and record package versions. A short README that names each file and what it produces saves you and your committee weeks later.

You have the data. The dissertation is now an argument you build in prose, chapter by chapter, from a defensible question to defensible conclusions. This section covers the shape of the document, the APA mechanics that reviewers notice, and a workflow that gets words on the page without stalling.

The five-chapter anatomy

The traditional psychology dissertation runs five chapters. Each has a job, and each has a way of going wrong.

Chapter 1: Introduction

This orients the reader to the problem and states what you did and why it matters. It is a funnel: the broad significance of emotional regulation in infancy, narrowing to the specific gap your study addresses, ending with your research questions and hypotheses. Keep it shorter than you think. The full literature lives in Chapter 2.

Common failure: a mini literature review that duplicates Chapter 2, or a vague problem statement that never commits to a specific question. Another is stating significance in general terms ("regulation is important for development") without connecting it to what your particular study will show.

Chapter 2: Literature Review

This builds the case that your question is worth asking and has not yet been answered. It should be organized thematically or theoretically, not as an annotated list of studies. For a developmental regulation dissertation, that often means threading together the major frameworks, the paradigms used to measure regulation in infants, and the specific findings that leave your gap open.

Common failure: the "study X found... study Y found..." march with no synthesis. The reader should see where the field agrees, where it conflicts, and where it is silent. The silence is your gap.

Chapter 3: Method

Written so another researcher could replicate you. Standard subsections: participants (recruitment, sample size, demographics, attrition, power analysis justifying N), design, measures and instruments (with reliability and validity evidence), procedure, and analytic plan. For infant work, include coding schemes, inter-rater or inter-coder reliability (report the statistic, kappa or ICC as appropriate), and how you handled fussy-out, missing trials, and exclusions.

Common failure: under-specifying the coding and reliability procedures, or describing an analytic plan that does not actually match the analyses that appear in Chapter 4. Write this chapter in past tense since your data are collected.

Chapter 4: Results

Report what you found, not what it means. Lead with descriptive statistics and preliminary checks (assumptions, distributions, correlations among key variables), then present inferential results organized by research question or hypothesis, in the same order you posed them. Report exact statistics: test statistic, degrees of freedom, exact p value, effect size, and confidence intervals.

Common failure: interpreting in Results ("this suggests infants regulate better when...") or burying the primary analyses under exploratory ones. State the outcome of each hypothesis plainly.

Chapter 5: Discussion

Covered in depth below, since it is where most dissertations lose altitude.

The three-paper (manuscript-style) alternative

Some CUNY programs allow a three-paper format: a general introduction, three publishable-quality manuscripts as the body, and a general conclusion that integrates them. Each manuscript has its own method, results, and discussion. This suits candidates aiming to publish quickly, but it demands three coherent studies and an integrative frame that shows they belong together.

APA 7th edition essentials

Reviewers read for content, but formatting errors signal carelessness and cost you goodwill. The durable ones:

• Headings. Five levels, each with a fixed format. Level 1 is centered, bold, title case. Level 2 is flush left, bold, title case. Level 3 is flush left, bold italic. Levels 4 and 5 are indented, bold (4) or bold italic (5), run into the paragraph. Do not skip levels.
• In-text citations. Narrative: Author (year). Parenthetical: (Author, year). Three or more authors use "et al." from the first citation. Match every in-text citation to a reference entry and vice versa.
• References. Hanging indent, alphabetical by first author surname, DOIs as full https links. Include the DOI whenever one exists.
• Reporting statistics. Italicize statistical symbols (M, SD, t, F, p, r, N). Report exact p values (for example p = .03) rather than p < .05, except report p < .001 for very small values. Include effect sizes and confidence intervals. Use a leading zero for numbers that can exceed 1 and omit it for numbers that cannot (correlations, p values).
• Tables and figures. Numbered in order of mention, each with a bold number, an italic title, and a note where needed. Every table and figure must be referred to in the text. Do not present the same data in both a table and a figure.
• Bias-free language. Use person-first or identity-first language per current guidance, report participant characteristics specifically, and describe your infant sample and caregivers with precise, respectful terms.
• Title page and abstract. Follow your program's manual for the title page (it often overrides APA's student defaults). The abstract is one paragraph, usually 150 to 250 words, stating problem, method, key results with numbers, and conclusion.

Building the argument

The whole document is one argument, and the seams between chapters are where coherence breaks. The test: your Chapter 1 research questions, your Chapter 3 analyses, your Chapter 4 results, and your Chapter 5 interpretations should all name the same constructs in the same order. If hypothesis 2 is about caregiver contingency and Chapter 4 never reports a caregiver analysis, the thread is broken.

Write your research questions and hypotheses so they are directly testable and map one-to-one onto analyses. For a quantitative design, each hypothesis should imply a specific statistical test and a specific direction. Number them, and reuse those numbers in Results and Discussion so the reader can follow the thread without effort.

The Discussion chapter

This is the chapter committees scrutinize hardest, because it shows whether you understand your own findings. Structure it as: a brief restatement of purpose and key findings (one paragraph, not a Results rerun), then interpretation organized by research question, then limitations, implications, and future directions.

Interpret, do not restate

Results says "regulation scores were higher in condition A, t(48) = 2.30, p = .03." Discussion says what that means: which theoretical account it supports, whether it replicates or contradicts prior work, and what the mechanism might be. If you find yourself repeating numbers, you are restating.

Situate in the literature

Every key finding should be placed against Chapter 2. Does it converge with prior findings, extend them to a younger age, or conflict? When it conflicts, offer plausible reasons (sample, paradigm, measurement) rather than dismissing the other work.

Limitations, done honestly

Name the real constraints: sample size and composition, generalizability, the fact that your regulation measure captures one facet and not the whole construct, any confounds you could not control, and correlational versus causal reach. Honest limitations build credibility. Boilerplate ("future research with larger samples") reads as evasion. Say specifically how each limitation bounds your conclusions.

Implications and future directions

Distinguish theoretical implications (what this means for how the field models infant regulation) from applied ones (what it might mean for caregivers, clinicians, or intervention), and be careful not to overclaim applied impact from a single study. Future directions should follow from your specific limitations and findings, not be a generic list.

Managing your committee and advisor

• Send clean, self-contained drafts. Number pages, use tracked changes when revising, and tell your chair what kind of feedback you want (structure, or line edits, not both at once). Send a chapter when it is genuinely ready, not a rough patch.
• Incorporate feedback without losing your voice. Distinguish substantive concerns you must address from stylistic preferences you can weigh. When you disagree with a committee member, respond with a reasoned case in a memo, not silence and not capitulation. A short response document listing each comment and how you handled it makes revisions visible and defensible.
• Track revisions. Keep a change log per chapter mapping each reviewer comment to your action. This is invaluable at the defense and when a second reader raises something the chair already approved.
• Defense preparation. Prepare a tight talk that covers question, method, key results, and contribution, and rehearse the likely hard questions: why this design, why this sample size, what a null result would have meant, and the boundary of your causal claims. Reread your own limitations section, since that is where questions cluster. Know your numbers cold.

Practical workflow

1. Outline first. Draft a heading-level skeleton of the whole chapter before prose. Every research question gets its slot in Method, Results, and Discussion.
2. Write badly, then revise. A first draft exists to be fixed. Separate drafting from editing so you are not polishing sentences you will cut.
3. Version control your drafts. Use dated filenames or a git repository for the manuscript so you can recover an earlier phrasing and see what changed between committee rounds. Never overwrite a version your chair has commented on.
4. Reverse-outline to check structure. After a draft, write one sentence per paragraph capturing its point. Read those sentences in sequence. If the logic jumps or a paragraph has no clear point, the structure needs work, not the prose.

Being stuck is not a character flaw and it is not a sign the project is doomed. It is a signal, and different signals need different responses. The most common mistake is applying a writing fix to a thinking problem, or a motivation fix to a decision problem, and then feeling worse when the effort does not work. So before you reach for a technique, spend two minutes naming what kind of stuck you actually are.

Diagnose the type of stuck first

There are four common flavors, and they feel similar from the inside but respond to completely different moves.

• A thinking problem. You are not clear on the idea itself. You cannot write the paragraph because you do not yet know what you believe. The tell: when you try to write, you keep circling, contradicting yourself, or writing a sentence and immediately deleting it because it is not quite right. The fix is not more willpower. It is thinking out loud, mapping the argument, or going back to your data and notes until the claim sharpens.
• A writing problem. You know exactly what you want to say, you could explain it to a colleague over coffee in ninety seconds, but the words will not come out on the page. The tell: you can talk about it fluently but freeze when the cursor blinks. The fix is to lower the quality bar and get an ugly version down, or literally transcribe what you would say out loud.
• A decision problem. You have too many options and cannot commit. Which framing, which studies to foreground, which way to structure the discussion, whether to report the null result first or last. The tell: you feel busy and anxious but produce nothing, because every path forward forecloses another. The fix is to make the decision provisional, pick one, and let writing reveal whether it was right.
• A motivation or energy problem. You know the idea, you can write, you have decided, and you still cannot start. The tell: you are avoiding, you feel tired before you begin, the dread is bigger than the task. This one is often physical and emotional, not intellectual. The fix is smaller tasks, momentum systems, rest, and sometimes stepping away entirely rather than grinding.

Unblocking techniques

For a thinking problem: talk it out

The fastest way through a fuzzy idea is to explain it to someone. You do not need a person available at 11pm. Talk out loud to an empty room, or type a rambling explanation to Claude Code and ask it to reflect the argument back to you. The act of explaining forces the idea into linear order, which is exactly what was missing.

For any stuck: freewriting and the brain dump

Set a timer for ten minutes and write about the section with the editor turned completely off. No backspace, no rereading, no fixing typos. If you get stuck, write "I am stuck because" and keep going. The rule is that your hands do not stop moving. This is not the draft. It is the raw material you mine for the draft, and it works because it separates generating from judging, which is the whole game.

The ugliest possible first draft

Give yourself explicit permission to write badly. Write "the infants did the thing and I think it means the boring obvious thing" if that is what comes. An ugly draft you can revise beats a blank page you are ashamed of. Anne Lamott's phrase for this is the shitty first draft, and it exists as advice precisely because every serious writer needs it. You revise a draft. You cannot revise nothing.

Write the sentence you can write

You do not have to write the section in order. If the opening paragraph is blocking you, skip it. Write the one sentence you are sure of, wherever it lives in the section. Then the next one you are sure of. Leave brackets like [transition here] and [need the stat] and keep moving. Order and polish are cheap later. Momentum is expensive now.

Reverse-outline what you already have

If you have a messy draft and cannot see why it is not working, reverse-outline it. Go paragraph by paragraph and write one short line stating what each paragraph actually does, not what you meant it to do. Now you can see the real structure, the repeats, the gaps, and the paragraph that is secretly two paragraphs. This turns a vague "this is not working" into a specific list of moves.

Shrink the task

"Write the discussion" is not a task, it is a mountain. "Write one paragraph explaining the main finding" is a task. When you are stuck, the task is almost always too big. Shrink it until it feels almost embarrassingly small, then do that one thing. A dissertation is written one paragraph at a time, and the paragraph you can start is worth more than the chapter you are dreading.

Momentum systems

Motivation is unreliable. Systems are not. The point of a system is that it decides for you, so you spend your willpower on writing instead of on the daily negotiation about whether to write.

• Timeboxing and Pomodoro. Twenty-five minutes of writing, five-minute break, repeat. The timer makes the commitment small and finite. You are not writing until it is good, you are writing until the timer rings. Two or three good pomodoros is a real day of dissertation progress.
• A daily minimum. Pick a number so small you cannot talk yourself out of it. Two hundred fifty words. Some days you write exactly the minimum and stop. Many days the minimum gets you past the friction and you keep going. Both are wins.
• Touch it every day. Open the document every single day, even for ten minutes, even to fix one sentence. The cost of a cold start after a week away is enormous. Keeping the project warm is the single highest-leverage habit for a long project.
• Write before email. Do your writing before you open your inbox or your messages. Your freshest attention should go to the hardest work, not be spent on everyone else's requests before you get to your own.
• Track a streak. A simple chain of days you showed up is oddly powerful. You are not trying to be brilliant each day, you are trying not to break the chain.
• Separate writing from editing. These are different jobs done by different parts of your brain, and doing them at once is why the page freezes. Draft in one session with the critic off. Edit in a later session with the critic on. Never both at the same time.

Perfectionism and the inner critic

You are an experienced, careful researcher, and that carefulness is a strength in your methods and a trap in your first drafts. The standard that makes your work rigorous is the same standard that freezes the page, because nothing you write on the first pass will meet it. The skill is knowing when to hold the standard and when to set it down.

Perfect is the enemy of done, and for a dissertation, done genuinely beats perfect. A finished, defended, good-enough dissertation moves your career forward. A perfect unfinished one does nothing. The messy middle, where the draft is ugly and the argument is half-formed and you cannot yet see the shape, is not a sign of failure. It is the normal, universal texture of the middle of a long project. Everyone who has finished one passed through it feeling exactly as you feel.

Comparison is a particular trap. You are comparing your rough draft and your private doubt against other people's published, polished, revised final products. That is not a fair fight and it tells you nothing true. The published paper you admire went through drafts as ugly as yours.

The perpetual literature trap

Reading feels like work and it is safe, which makes it the most seductive form of avoidance available to a scholar. There is always one more paper. The field keeps producing them. If you wait until you have read everything on emotional regulation in infancy, you will never write, because that day does not exist.

Your data collection is done. That is the fact that should reset your relationship to the literature. You are no longer reading to decide what to study. You are reading to situate what you found and to explain what it means. That is a narrower, finishable job. Read to answer specific questions your results raise, not to achieve completeness.

A useful discipline is to write from what you know now, mark the gaps with brackets like [check whether this replicated], and batch those lookups into a separate reading session later. This keeps reading from swallowing your writing time, and it keeps you honest about how much of the "I need to read more" feeling is real and how much is fear wearing a scholarly costume.

Wellbeing and the long game

A dissertation is a marathon, and you cannot sprint a marathon on no sleep and no rest without paying for it later, usually in the form of exactly the stuck feeling that brought you here. Protecting your body and mind is not a break from the work. It is part of how the work gets done.

• Sleep is a cognitive tool. The tired brain cannot form the sharp sentences a discussion section needs. When you are stuck at night, the honest move is often to sleep and return in the morning, when the paragraph that was impossible becomes merely hard.
• Move. A walk is one of the most reliable ways to loosen a stuck idea. Many problems that will not yield at the desk solve themselves on foot, because a different mode of attention lets the background of your mind work.
• Take real breaks. Rest that you feel guilty through is not rest. A true break, away from the screen, restores the attention that the work depends on.
• Keep your support network close. Isolation makes every problem heavier. Other people who are writing or have written understand the specific loneliness of this stage. A writing group, a friend on the same road, a standing check-in with a peer, all of it helps carry the weight.
• Celebrate small wins. Finished a rough section. Cracked a paragraph that fought you for a week. These deserve to be noticed. A long project with no acknowledged progress is demoralizing by design, so acknowledge the progress.

Hold onto the real proportion here. Finishing matters more than flawlessness. The goal is a strong, complete, defended dissertation, not a perfect one, because the perfect one is imaginary and the complete one is what changes your life.

When to talk to your advisor versus push through alone

Not every stuck point needs your advisor, and not every stuck point should be suffered alone. The line is worth knowing so you neither waste their attention nor waste your own weeks.

Push through alone when it is a writing problem, an energy problem, or the ordinary friction of drafting. Your advisor cannot write your sentences, and a week of feeling stuck on prose is normal and yours to work through with the techniques above. Bringing every daily block to your advisor trains them to see you as fragile and burns goodwill you will want later.

Go to your advisor when it is a genuine thinking or decision problem that touches the direction of the work. If you are unsure how to frame a finding, whether an analysis is defensible, how to handle a result that does not fit your hypothesis, or how to scope a chapter, that is exactly what they are for, and pushing through alone risks writing thirty pages in the wrong direction. Also reach out when you have been stuck for more than a couple of weeks with no movement, when you cannot tell what kind of stuck you are in, or when the stuck feeling has tipped into something heavier that is affecting your health. A short, specific email with two or three concrete questions gets you a better response than silence followed by a crisis.

You have done the hard, patient part already. The data is collected. What remains is telling the field what you found, and that is a task made of paragraphs, each one small enough to start. The path back to momentum is not a burst of inspiration. It is the next small honest paragraph, today, on a project you keep warm. You are entirely capable of this.

Your data is collected and the hard part is behind you. This section is for the moments when you have energy left over and want the dissertation to do more than earn the degree. Everything here is optional. None of it is busywork. Each item raises the odds that your work gets read, cited, replicated, and remembered.

From dissertation chapter to journal article

A dissertation chapter and a journal article are different documents with different jobs. The chapter proves to your committee that you can do rigorous, exhaustive work. The article persuades a busy reader that one specific finding is true and matters. The chapter can afford to be long, cautious, and complete. The article has to be tight, argued, and pointed.

When you convert a chapter, expect to cut heavily. The literature review that ran twenty pages becomes three or four paragraphs that set up exactly the gap your study fills. The methods section keeps everything a reader needs to replicate you and drops the tutorial explanations. The results report the analyses that answer your question and move the rest to supplementary material. The discussion makes one clear argument instead of surveying every possible interpretation.

Choosing a target journal

Match the paper to the audience before you write the cover letter. In infant emotional regulation, the natural homes are worth knowing by their character, not just their name.

• Child Development and Developmental Psychology are the flagship general outlets. They reward strong theory, clean design, and a finding that speaks to a broad developmental audience. A single well-powered study can land here if the contribution is clear.
• Infancy (the journal of the International Congress of Infant Studies) is the specialist home for infant work. Reviewers know your paradigms cold, so methodological detail is expected and rewarded rather than glossed.
• Emotion (APA) fits when your central contribution is about affect, regulation processes, or measurement of emotion, and the developmental angle is one facet rather than the whole story.
• Also worth knowing depending on framing: Developmental Science, Developmental Psychobiology (if you have physiological measures such as heart rate variability or cortisol), and Developmental Cognitive Neuroscience if there is a neural component.

Read the aims and scope page, then read three recent articles in your candidate journal and ask whether your paper would sit comfortably beside them. That single check saves months of misdirected submission.

Open science as a strength, not a chore

Open practices are the cheapest credibility you will ever buy, and for a project where data collection is already done, several of them are still fully available to you.

• Preregistration on OSF. You cannot preregister a study you have already run, but you can transparently distinguish confirmatory from exploratory analyses. If you have a chapter or a follow-up study still in the analysis stage, preregister the analysis plan before you look. Honest labeling of what was planned versus what you found after the fact is itself a mark of rigor.
• Registered Reports. Several developmental journals now offer this format, where peer review happens on the introduction and method before results exist, and acceptance is granted based on the question and design. It is a strong option for your next study, not this dataset, but worth knowing exists.
• Sharing data, code, and materials. A deidentified dataset, your analysis scripts, and your stimuli or coding scheme posted to OSF make your work reusable and far more citable. With infant data, deidentification and consent language matter, so check what your IRB consent covered before posting video or identifiable material. Aggregate data and code can almost always be shared even when raw video cannot.
• Open science badges. Many journals award small badges for open data, open materials, and preregistration. They are visible signals that reviewers and readers notice.

Conferences and scholarly community

The dissertation is a private accomplishment until you put it in front of the people who study what you study. Two meetings matter most for this work.

• SRCD (Society for Research in Child Development) holds a large biennial meeting and is the central gathering for developmental science. A poster or talk here reaches the broadest relevant audience.
• ICIS (International Congress of Infant Studies) is the specialist infancy meeting, biennial, and the room where the people who will review your Infancy submission are standing. For infant regulation work, this is the highest-density audience you can find.
• APS (Association for Psychological Science) and, depending on your angle, meetings focused on emotion or affective science, give you cross-disciplinary reach.

Submit a poster even if a talk feels premature. Posters are where you get unhurried, one-on-one feedback from people who will later cite you, and where postdoc and faculty conversations quietly begin. Bring a short, honest version of your finding you can say in two sentences, and a business card or a slip with your name, email, and OSF or ORCID link.

Reproducibility and craft

An old-school researcher's instinct for careful records translates directly into modern reproducibility. The goal is that a stranger, or you in three years, can rerun your analysis and get the same numbers.

• A clean pipeline. Keep raw data untouched and read-only. Every transformation lives in a script, not in a spreadsheet you edited by hand. Cleaning, then analysis, then figures, in that order, each in its own file.
• Documented code. Comments that say why, not just what. A README that states how to run everything from a fresh start and what each file produces.
• A public repository. A version-controlled repo (GitHub, or OSF for archival) with a clear structure and a license. This is exactly the kind of work Claude Code is good at helping with: refactoring a tangled script into named steps, writing a README, adding comments, and checking that the pipeline runs top to bottom.
• Transparent reporting. Report exact sample sizes, all exclusions and why, effect sizes with confidence intervals, and every measure you collected. Consult reporting guidance such as the APA Journal Article Reporting Standards (JARS). If your work is longitudinal or has a flow of participants through stages, a participant flow diagram helps the reader.

Making a real contribution

Above and beyond means the field is different because your paper exists. Contributions come in a few honest forms, and you do not need all of them.

1. A genuine gap. Not "no one has studied X in this exact sample" but a real hole in understanding that your data fills. State it as a question the field cannot currently answer.
2. Methodological value. A better coding scheme, a cleaner paradigm for eliciting or measuring regulation, or a measurement approach others can adopt. Sharing the instrument multiplies the contribution.
3. Replication value. If your finding confirms or fails to confirm a prior result, say so plainly. Direct and conceptual replications are undervalued and badly needed in developmental science, and a clean one is a real gift to the literature.
4. Theoretical clarity. Sometimes the contribution is sharpening a fuzzy construct or adjudicating between two accounts of how infant regulation develops.
5. Translational reach. Implications caregivers and clinicians can actually use. If your finding tells a pediatrician or an early-intervention team something concrete, name it in the discussion. That is often what gets a paper covered and taught.

The scholar's toolkit

A few pieces of durable infrastructure make you findable and let your work accumulate over a career.

• ORCID. A free, permanent identifier that ties every paper, dataset, and review to you and never breaks when you change names or institutions. Set it up once and put it on everything.
• Google Scholar profile. Auto-collects your work and tracks citations. Set it to verify additions so it does not attach papers that are not yours.
• Citation-tracking and alerts. Create Google Scholar alerts for your key constructs, your own name, and two or three researchers whose work is closest to yours. This keeps you current without hunting.
• Peer review and mentoring. Reviewing for a journal (ask a mentor to recommend you, or accept invitations that follow your first publication) teaches you the standards from the inside and builds your reputation with editors. Mentoring a junior student does the same in reverse.

Career runway

Everything above compounds into options. A dissertation that produced one or two clean, well-communicated papers, with shared data and a talk at SRCD or ICIS, is what makes a postdoc application competitive, because it shows you can finish and publish, not just collect. The same record supports faculty applications, where a clear line of contribution and evidence of open, reproducible practice increasingly matter to hiring committees. And the translational framing, the part that speaks to caregivers and clinicians, is exactly what opens doors in applied and clinical research settings, industry research, and policy-facing work.

You do not have to do all of this. Pick the one or two items that fit the energy you have and the paper in front of you. A single strong article in the right journal, with the data and code posted and a poster at the right meeting, already puts you well past the minimum. The rest is there for when you want it.