ICAs

ICA 3: Exploratory research

Purpose is to discover and understand a topic that is not well studied. Verbs: discover / understand. Broad claims, no hypothesis tested, often inductive (specific observations to broader themes).

ICA 3: Descriptive research

Purpose is to describe and measure characteristics, patterns, and trends. Verbs: describe / measure. Answers who, what, when, where, how often, but not why.

ICA 3: Explanatory research

Purpose is to explain and test why something happens. Verbs: explain / test. Tests cause and effect (does X affect Y?).

ICA 3 Q1: Classify the interprofessional communication study

Exploratory. (1) It was not designed to test a hypothesis or fully answer a question; it aims to understand a little-studied topic. (2) It uses inductive reasoning, moving from professionals' specific interview answers to broader themes such as patient involvement and the structure of rounds.

ICA 3 Q1: Classify CDC Health United States 2020-2021

Descriptive. (1) It reports rates, percentages, and trends over time in health outcomes. (2) It describes differences between groups without testing cause and effect.

ICA 3 Q1: Classify the Prenatal WIC study

Explanatory. (1) It asks a cause-and-effect question: does prenatal WIC participation (X) improve child outcomes (Y) such as lower ADHD rates and less grade repetition? (2) It accounts for confounders like maternal age, race, education, and health conditions so the effect can be attributed to WIC. [REFINED: maternal conditions like gestational diabetes are control variables, not the causes being studied.]

ICA 3 Q2: The exploratory study is trying to...

Find out and understand what is happening in a topic that is not yet well studied, without testing a cause or solution.

ICA 3 Q2: The descriptive study is trying to...

Measure and describe characteristics, patterns, and trends in a population by collecting and observing data. [REFINED: removed 'evaluate', since evaluating effects is explanatory.]

ICA 3 Q2: The explanatory study is trying to...

Test why particular results occur, meaning whether X causes Y, and draw conclusions about that relationship.

ICA 3 Q3: Do labels like 'descriptive method of content analysis' or 'explore this question' change the classification?

No. Classification depends on the study's aim and the type of claim it makes, not on the words the authors use or the analytic technique. [REFINED: the original answer said methodology defines the type, which conflicts with the point of the question.]

ICA 3 Q3: Research purpose vs. research method

The same method can serve different purposes. The CDC report (descriptive) and the WIC paper (explanatory) both use statistical comparison, but one describes patterns and the other tests a cause.

ICA 3 Q4: Statement A (professionals identify patient involvement and rounds structure as important)

Evidence comes from the interprofessional communication study. Exploratory, because it identifies themes.

ICA 3 Q4: Statement B (smoking rates differ across education groups)

Evidence comes from the CDC Health United States report. Descriptive, because it reports measured differences between groups.

ICA 3 Q4: Statement C (prenatal WIC improves child outcomes)

Evidence comes from the Prenatal WIC study. Explanatory, because it makes a causal claim.

ICA 3 Q4: How does the claim change from A to B to C?

It moves from identifying themes, to describing measured differences, to asserting cause and effect. Claims get more specific and harder to prove.

ICA 3 Q5: Why is the CDC report descriptive and not explanatory?

It identifies patterns and differences between groups but does not test why they occur. Association does not establish causation.

ICA 3 Q6: Chosen theme

Healthcare professionals recognize patient involvement in the decision-making process and robust communication as key aspects of effective communication.

ICA 3 Q6a: Descriptive research question (refined)

How often, and in what ways, are patients included in the decision-making process during rounds? [REFINED: the original asked whether involvement 'improves patient outcomes', which is about an effect and therefore explanatory.]

ICA 3 Q6b: Explanatory research question (refined)

Does including patients in the decision-making process during rounds improve patient outcomes, such as satisfaction or treatment adherence? [REFINED: the original switched topics to pre-round briefing. The explanatory question must stay on the same topic, patient involvement.]

ICA 3 Q6: Why the original descriptive question was wrong

'When can including patients... improve patient outcomes?' The word 'improve' implies cause and effect, so it is closer to explanatory. Descriptive questions ask who, what, when, where, or how often.

ICA 3 Q6: Why the original explanatory question was wrong

'Does pre-round briefing improve physician outcomes?' is explanatory in form but changes the topic from patient involvement to pre-round briefing. All three questions should share the same general topic.

ICA 6: Structured abstract

An abstract divided by headings (e.g., Objective, Study Setting and Design, Data Sources, Principal Findings, Conclusions). Example: Article 1, Walker et al., Louisiana Medicaid.

ICA 6: Unstructured abstract

An abstract written as one narrative paragraph. Example: Article 2, Jun, dependent coverage mandates and fathers' job mobility.

ICA 6: Recommended components of an abstract (lecture)

Need for the project; objectives; study design/methodology; significance/contributions.

ICA 6 Q1a: Which abstract is easier to navigate quickly?

Article 1, because its subheadings label each component (objective, design, findings). Article 2 is one dense paragraph.

ICA 6 Q1b: Does the structured format make the research clearer?

No. It doesn't change the research; it makes the information easier to find and gives a quick overview.

ICA 6 Q1c: Advantages of an unstructured abstract

It reads as one connected story that links background, question, methods, and findings, and it gives authors more flexibility (common in economics journals). [REFINED: the original said 'more background'. That was this author's choice, not a feature of the format.]

ICA 6 Q2: What can a reviewer find immediately in each abstract?

Article 1: conclusions, study design, and data sources. Article 2: mostly background; the first three sentences are context.

ICA 6 Q2: What takes more effort to find?

Article 2's description of what the study did (design/methods), which is buried mid-paragraph. Nothing in Article 1 is hard to find.

ICA 6 Q2: What is missing or unclear?

Article 1 lacks a clear statement of the need for the study. In Article 2 nothing is missing, but the methods are hard to find. [REFINED: removed the contradiction between 'need is missing' and 'nothing is clearly missing'.]

ICA 6 Q2: Which abstract is clearer after 30 seconds?

Article 1. Each component is separate and concise so readers can scan it. Reading Article 2's first and last sentences tells you very little.

ICA 6 Q3a: Is all of Article 2's background necessary?

No. Keep that mandates let young adults stay on parents' plans and may have unintended effects on parents. Shorten the detail on increased young-adult coverage, since the study is about fathers' job mobility and compensation.

ICA 6 Q3b: Sentence to remove from Article 2

'In the United States, young adults have low rates of health insurance coverage.' The next sentence already implies why the mandates exist, and this is the sentence least connected to fathers' outcomes.

ICA 6 Q3c: Background sentence to add to Article 1

Medicaid pays for about 4 in 10 U.S. births, but low out-of-pocket costs may not protect low-income mothers from financial strain after birth, such as lost income and caregiving demands. Without this, readers may assume Medicaid coverage already protects mothers financially.

ICA 6 Q4: Main findings of each article

Jun: the ACA dependent coverage mandate reduced voluntary job separation among some fathers (0.9 percentage points, suggesting job lock), with weaker evidence of lower compensation. Walker et al.: pregnancy and childbirth were linked to worse finances for non-Hispanic Black Medicaid beneficiaries (about a 4-point credit score drop; +$41 medical and +$44 nonmedical debt).

ICA 6 Q4a: Which abstract separates findings from interpretation better?

Walker et al. (Article 1). 'Principal Findings' reports results and 'Conclusions' interprets them (financial vulnerability; need for broader economic support). Jun mixes numerical results and interpretation in one paragraph, which makes it hard to tell what the study shows from what the author thinks it means.

ICA 6 Q5: Sentence to keep (Article 2)

'These policies increased insurance rates among young adults but may have unintended consequences on the labor market outcomes of their parents.' It is clear and direct and sets up the relationship being tested.

ICA 6 Q5: Sentence to rewrite (Article 2)

Original: 'Despite its statistical insignificance, this may still provide evidence of a tangential effect...' Refined: 'Although not significant on their own, the state mandates may have diluted the ACA's effect, suggesting a possible indirect influence on fathers' job mobility.' It replaces the vague 'this' and states the implication directly. [REFINED: kept the hedge ('may', 'possible') because the result was not statistically significant.]

ICA 6 Q5: Information to add (Article 2)

The abstract says it studies fathers' job separation and compensation but never reports a conclusion about compensation. Add the compensation finding to make the abstract complete.

Experiments ICA: Scenario

A health system tests whether an automated text reminder sent 48 hours before a primary care appointment reduces no-shows among 400 eligible patients.

Experiments ICA: Random assignment

Using chance (lottery or random number generator) to decide who gets the treatment. It makes the groups comparable in expectation on both observed and unobserved traits.

Experiments ICA: Stratified (block) randomization

Group patients by key traits (e.g., prior no-show history, age), then randomize within each group. This improves balance and precision. It does NOT 'increase randomization'.

Experiments ICA: Counterfactual

What would have happened to the treated group without the treatment. It can never be observed directly.

Experiments ICA: Selection bias

A bias that occurs when people choose (self-select) into groups, so the groups differ in ways that affect the outcome.

Experiments ICA Q1: Design the experiment

Randomly assign the 400 patients (lottery or random number generator) to 200 treatment and 200 control. Optionally stratify by traits such as prior no-show history. Confirm both groups have similar baseline no-show rates. Send the 48-hour text reminder to the treatment group only. Record no-shows in both groups and compare no-show rates.

Experiments ICA Q2: Treatment group

Patients who receive the automated text reminder 48 hours before their appointment.

Experiments ICA Q2: Control group

Patients who do not receive the text reminder (usual care).

Experiments ICA Q2: Intervention

The automated text reminder sent 48 hours before the appointment.

Experiments ICA Q2: Outcome (refined)

No-show rate: the share of patients who miss their appointment without canceling. [REFINED: originally 'patients showing up'; restated as a measurable rate that matches the question.]

Experiments ICA Q3: Major components of the experimental design

Random assignment; a treatment group (200) and a control group (200); the intervention (48-hour text); measurement of the outcome (no-shows); and comparison of outcomes between groups. [REFINED: stratification improves balance and does not 'increase randomization'.]

Experiments ICA Q4: Can you observe the counterfactual?

No. We can't see how texted patients would have behaved without the text. The randomized control group imitates the counterfactual because, on average, it is the same as the treatment group except for the text.

Experiments ICA Q5: Estimating the causal effect

Average treatment effect = treatment no-show rate minus control no-show rate. Example: 10% minus 18% = an 8 percentage point reduction. Then test whether the difference is statistically significant. [REFINED: more precise than 'compare attendance'.]

Experiments ICA Q6: Is the control group a good counterfactual?

Run a balance check: compare baseline characteristics (age, sex, insurance, prior no-show history, distance) and pre-period no-show rates across groups. They should be similar. Randomization balances unobserved traits in expectation. [REFINED: clarified wording about 'matching the counterfactual'.]

Experiments ICA Q7: Practical or ethical issue

Some patients may lack a phone or digital literacy because of cost or age, so assigned patients may never receive the text. Other issues: withholding a helpful, low-risk reminder from controls, and wrong numbers or opt-outs (noncompliance).

Experiments ICA Q8: Why voluntary sign-up is weak causal evidence

Selection bias. Patients who opt in may already be more organized or engaged and less likely to miss appointments, so differences can't be attributed to the text. Random assignment removes this confounding and isolates the text as the cause.

ICA 9: Difference-in-differences (DiD)

(Change over time in treatment group) minus (change over time in control group). The first difference is before vs. after; the second is treatment vs. control. Subtracting removes trends that would have happened anyway.

ICA 9: Parallel trends assumption

Without the policy, treatment and control outcomes would have moved the same way over time. Groups may start at different levels. Check this by comparing pre-period trends.

ICA 9: Natural experiment

A real-world event (policy, court ruling, disaster) that sorts people into treatment and control groups without self-selection, imitating random assignment.

ICA 9 Q1: Causal research question (refined)

Did OBBBA's Medicaid work requirements reduce Medicaid enrollment among low-income adults in expansion states compared with non-expansion states? [REFINED: the original ('The impact on Medicaid eligibility after OBBBA') was not phrased as a question and did not name a specific cause or a measurable outcome.]

ICA 9 Q2: Causal factor and outcome (refined)

Cause: implementation of OBBBA's Medicaid work requirements. Outcome: Medicaid enrollment (or the uninsured rate) among adults ages 19-64. [REFINED: eligibility is set by the law itself, so measure enrollment or coverage instead.]

ICA 9 Q3: Natural experiment event (refined)

OBBBA (2025) requires work requirements for the Medicaid expansion adult group, starting by 2027. Only expansion states are affected, which creates a natural split. [REFINED: the original event (ACA 2010 / NFIB v. Sebelius 2012) explains ACA expansion, not OBBBA. The event must match the cause.]

ICA 9 Q3: Alternative if keeping the original event

If you keep ACA / NFIB v. Sebelius as the event, change the question to match: Did ACA Medicaid expansion increase Medicaid enrollment among low-income adults in expansion vs. non-expansion states?

ICA 9 Q4: Treatment and control groups

Treatment: low-income adults in Medicaid expansion states (subject to work requirements). Control: low-income adults in non-expansion states.

ICA 9 Q5: Two dimensions

Time (before vs. after work requirements begin) and state expansion status (expansion vs. non-expansion).

ICA 9: Limitation of the state-based natural experiment

States chose whether to expand, so expansion states may differ in politics, budgets, and populations. Check parallel pre-trends. Post-policy data may also take time to become available.

Grant ICA: How to search NIH opportunities

Use 2-3 keywords from your research question (not the full question). The goal is a reasonably aligned opportunity, not a perfect one.

Grant ICA Q1: Opportunity identified

NIH Exploratory/Developmental Research Project Grant (Parent R21 Clinical Trial Not Allowed), PA-25-304.

Grant ICA Q2: Institute

National Institute on Minority Health and Health Disparities (NIMHD), a participating institute in PA-25-304.

Grant ICA Q3: Funding mechanism

R21: Exploratory/Developmental Research Grant.

Grant ICA: Parent announcement

A broad, investigator-initiated funding opportunity that many NIH institutes participate in, for research on any topic within their missions.

Grant ICA Q4: Why the R21 fits the project

The R21 supports early, conceptual stages of projects and new approaches. Little research looks at how Texas immigrants navigate Medicaid (barriers: strict eligibility rules, fear of deportation tied to documentation, language access). An exploratory study (interviews or a pilot survey) would lay the groundwork for a larger study. It is observational, not a clinical trial, which fits because this opportunity does not allow clinical trials. [REFINED: original wording implied the opportunity requires a clinical trial.]

Grant ICA Q5: Constraints and limitations

Up to 2 years and $275,000 total direct costs (no more than $200,000 in any one year), so the study must be scaled down. Recruiting immigrant participants, including mixed-status families, takes time and care. A student cannot be the PI; the application must come through an eligible institution with an eligible PI (e.g., a faculty mentor). [REFINED: added the per-year cap and how a student could take part.]

Grant ICA: If no opportunity fits

That is a legitimate finding. Report what you searched and why the project may not fit current NIH opportunities.