Manuel B. Garcia

Manuel B. Garcia serves as the Senior Director for Educational Technology and Digital Learning at FEU Institute of Technology, Manila, Philippines. Read More

Contact Info

1607, FEU Tech Building,
P. Paredes St, Sampaloc,
Manila, Philippines
mbgarcia@feutech.edu.ph

Follow Me

The Complete Library

All Research Guides

Browse every published guide in the Researcher's Field Guide, including the complete research journey and supplemental collections, listed alphabetically for quick reference.

Complete Library

Every published guide

Showing 2,701–2,750 of 5,065 guides

5,065 guides
PART 04 · SUBPART 42 · GUIDE 483

Should Interview Questions Be the Same?

Should All Research Interview Participants Receive the Same Questions?

Not every research interview participant needs to receive identical questions. Learn when standardization is necessary, when tailored follow-up questions improve qualitative inquiry, and how to maintain analytical coherence across interviews.

PART 04 · SUBPART 42 · GUIDE 482

Interview Sampling and Data Saturation

Should Interview Sampling Continue Until Data Saturation Is Reached?

Data saturation is widely used to justify stopping qualitative interviews, but it is not appropriate for every methodology. Learn what different forms of saturation mean, how to evaluate them, and when other criteria provide a more defensible basis for ending recruitment.

PART 04 · SUBPART 42 · GUIDE 481

How Many Research Interviews to Conduct

How Many Research Interviews Should You Plan Before Data Collection?

There is no universal minimum number of qualitative interviews. Learn how to plan an initial sample using research objectives, participant specificity, analytical requirements, information power, and a transparent strategy for evaluating sample adequacy.

PART 04 · SUBPART 42 · GUIDE 480

Structured vs. Semi-Structured vs. Unstructured Interviews

Structured vs. Semi-Structured vs. Unstructured Interviews: Which Fits the Research Question?

Structured interviews prioritize standardized questioning, semi-structured interviews combine planned topics with responsive probing, and unstructured interviews allow participants' accounts to shape the conversation. Learn which approach best supports your research question and how to justify your choice.

PART 04 · SUBPART 42 · GUIDE 479

Individual Interviews vs. Focus Groups

When Are Individual Interviews Better Than Focus Groups?

Individual interviews may be preferable when research requires private disclosure, detailed personal accounts, or sustained exploration of individual experiences. Learn how to distinguish these purposes from questions that benefit from focus group interaction.

PART 04 · SUBPART 41 · GUIDE 478

Designing Surveys to Minimize Nonresponse

How Do You Design a Survey to Minimize Nonresponse Before Data Collection Begins?

Nonresponse prevention begins before the first survey invitation is sent. Learn how to anticipate participation barriers, design accessible questionnaires, plan recruitment and follow-up, and assess nonresponse without confusing response rates with representativeness.

PART 04 · SUBPART 41 · GUIDE 477

Survey Question Randomization and Rotation

When Should Survey Questions Be Randomized or Rotated?

Randomizing or rotating survey questions may help distribute order effects, but changing question order is not always appropriate. Learn when randomization is useful, when sequence must be preserved, and how to implement and evaluate it without compromising measurement quality.

PART 04 · SUBPART 41 · GUIDE 476

Single-Item vs. Multiple-Item Measures

Should Every Construct Be Measured with Multiple Survey Questions?

Not every research construct requires multiple survey questions. Understand when a single item may be sufficient, why complex latent constructs often benefit from multi-item scales, and how measurement theory should guide the number and selection of items.

PART 04 · SUBPART 41 · GUIDE 475

Survey Length and Response Quality

How Long Can a Survey Be Before Response Quality Suffers?

There is no universal maximum survey length that guarantees reliable responses. Learn how questionnaire duration, cognitive demands, administration mode, and respondent motivation influence data quality, and how to determine an appropriate length for your study.

PART 04 · SUBPART 41 · GUIDE 474

Survey Mode Effects and Response Bias

Can Survey Administration Mode Affect Who Participates and How They Respond?

Survey administration mode can influence both who participates and how respondents answer the same questions. Understand the distinction between selection and measurement effects, why mixed-mode surveys require caution, and how to assess and reduce potential bias.

PART 04 · SUBPART 41 · GUIDE 472

Census vs. Sample Survey

Census vs. Sample Survey: Which Should You Use?

A census attempts to collect information from every eligible member of a population, while a sample survey collects information from a selected subset. Learn when complete enumeration is justified, when sampling is more efficient, and why surveying everyone does not automatically guarantee accurate findings.

PART 04 · SUBPART 41 · GUIDE 471

Cross-Sectional vs. Longitudinal Survey

Cross-Sectional Survey vs. Longitudinal Survey: Which Do You Need?

Cross-sectional surveys provide information about a population during a defined period, while longitudinal surveys investigate patterns across time. Learn how they differ, what each can establish, and how to select the appropriate temporal design for your research question.

PART 04 · SUBPART 41 · GUIDE 470

Survey Design vs. Survey Method

Is a Survey a Research Design or a Data-Collection Method?

A survey may refer to a broader research strategy or to a method of collecting information, depending on how the term is used. Understand these methodological distinctions and learn how to describe your study accurately without confusing the research design with the questionnaire.

PART 04 · SUBPART 41 · GUIDE 469

When to Use a Survey

When Is a Survey the Right Data-Collection Design?

Surveys can efficiently collect standardized information about people's experiences, attitudes, characteristics, and reported behaviors. Learn when this approach fits your research question, when it may produce misleading evidence, and what to consider before choosing it.

PART 04 · SUBPART 40 · GUIDE 468

Important Outcomes in Screening Program Evaluation

What Outcomes Actually Matter When Evaluating a Screening Program?

Screening programs should not be judged solely by how many cases they detect. Learn which outcomes demonstrate meaningful benefit, how mortality and morbidity differ from intermediate measures, why harms and implementation outcomes matter, and how to select endpoints that support credible screening decisions.

PART 04 · SUBPART 40 · GUIDE 467

Diagnostic Accuracy vs. Screening Benefit

When Is a Diagnostic Accuracy Study Insufficient to Justify Screening?

A screening test can have excellent sensitivity and specificity without improving health outcomes. Learn why diagnostic accuracy is necessary but often insufficient for screening recommendations, how disease prevalence and downstream interventions affect usefulness, and what additional evidence is needed before implementing a screening program.

PART 04 · SUBPART 40 · GUIDE 466

Comparison Groups for Screening Benefits

What Comparison Group Is Needed to Evaluate Screening Benefits?

A screening study needs a comparison that shows what would happen without the proposed strategy or under a credible alternative. Learn why comparing screen-detected with symptom-detected cases can mislead, when randomized or observational comparisons are appropriate, and how to align eligibility, time zero, outcomes, and follow-up to estimate screening benefits.

PART 04 · SUBPART 40 · GUIDE 465

Screening Harms vs. Detection Rates

Should a Screening Study Measure Harms as Well as Detection Rates?

A screening program may identify more disease while also exposing people to false-positive results, unnecessary procedures, overdiagnosis, and treatment-related complications. Learn which harms screening studies should measure, how to choose appropriate denominators and follow-up periods, and why detection rates alone cannot establish screening benefit.

PART 04 · SUBPART 40 · GUIDE 464

Accounting for Overdiagnosis in Screening Research

How Do You Account for Overdiagnosis in Screening Research?

Overdiagnosis occurs when screening identifies genuine disease that would never have caused symptoms or harm during a person's lifetime. Learn how to distinguish overdiagnosis from false positives, estimate its frequency using appropriate comparative evidence, address lead-time and competing mortality, and report uncertainty without overstating individual-level conclusions.

PART 04 · SUBPART 40 · GUIDE 463

Length Bias in Screening Studies

What Is Length Bias, and Why Does It Matter in Screening Studies?

Length bias occurs because screening is more likely to detect diseases that remain in a detectable preclinical state for longer. Learn why screen-detected patients may appear to have better prognoses even without screening benefit, how length bias differs from lead-time bias and overdiagnosis, and which research designs help distinguish genuine outcome improvements from selective detection.

PART 04 · SUBPART 40 · GUIDE 462

Lead-Time Bias in Screening Studies

What Is Lead-Time Bias, and Why Does It Matter in Screening Studies?

Lead-time bias can make screening appear to extend survival simply because disease is diagnosed earlier. Learn how the bias arises, why survival measured from diagnosis can mislead, how it differs from length bias and overdiagnosis, and which study designs and outcomes provide more credible evidence of screening benefit.

PART 04 · SUBPART 40 · GUIDE 461

Screening Research: Earlier Detection vs. Better Outcomes

How Do You Design Screening Research to Distinguish Earlier Detection From Better Outcomes?

Detecting disease earlier does not necessarily mean screening improves health. Learn how to design screening research that separates earlier diagnosis from genuine benefits through appropriate comparison groups, patient-important outcomes, follow-up, and safeguards against screening-related biases.

PART 04 · SUBPART 40 · GUIDE 460

Does Earlier Disease Detection Improve Outcomes?

Is Detecting Disease Earlier Automatically Beneficial?

Detecting disease earlier may create opportunities for effective treatment, but it can also increase diagnoses without improving outcomes. Learn how lead-time bias, length bias, and overdiagnosis can make screening appear beneficial, and what evidence is needed to establish that earlier detection actually helps patients.

PART 04 · SUBPART 40 · GUIDE 459

Research Questions in Screening Studies

What Question Is a Screening Study Actually Trying to Answer?

Screening research is not simply about detecting more cases or finding disease earlier. Learn how to distinguish screening accuracy, effectiveness, implementation, and harms, and formulate research questions that evaluate whether screening meaningfully benefits the intended population.

PART 04 · SUBPART 39 · GUIDE 458

Prediction Model Complexity and Practical Usefulness

When Is a Prediction Model Too Complex to Be Useful in Practice?

A more complex prediction model is not necessarily more useful. Learn when additional predictors, nonlinear functions, specialized measurements, and computational requirements undermine implementation, and how to balance predictive performance with reliability, interpretability, cost, and practical decision value.

PART 04 · SUBPART 39 · GUIDE 457

Discrimination, Calibration, and Clinical Usefulness

Should a Prediction Study Prioritize Discrimination, Calibration, Clinical Usefulness, or All Three?

A prediction model can distinguish high-risk individuals without estimating their risks accurately, and even accurate predictions may not improve decisions. Learn how discrimination, calibration, and clinical usefulness answer different questions, how to evaluate each, and when one deserves particular emphasis.

PART 04 · SUBPART 39 · GUIDE 456

Predictor Availability vs. Cost in Prediction Models

Should Easily Available Predictors Be Preferred Over Expensive Ones?

Expensive predictors may improve statistical performance, but their additional information does not always justify the cost or burden of measurement. Learn how to compare accessible and costly predictors, evaluate incremental predictive value, and design prediction models that remain useful in the settings where they will be applied.

PART 04 · SUBPART 39 · GUIDE 455

Candidate Predictors and Prediction Model Sample Size

How Many Candidate Predictors Can the Available Data Support?

The number of candidate predictors a dataset can support depends on more than its participant count or events-per-variable ratio. Learn how outcome frequency, model complexity, anticipated performance, and overfitting influence sample size planning, and why removing predictors after significance testing does not solve inadequate development data.

PART 04 · SUBPART 39 · GUIDE 454

Selecting Predictors Before Data Collection

Should Predictors Be Selected Before Data Collection?

Choosing candidate predictors before data collection can improve measurement consistency, sample size planning, and the credibility of prediction-model development. Learn what should be prespecified, when data-driven selection may be justified, and how to distinguish candidate predictor planning from final model selection.

PART 04 · SUBPART 39 · GUIDE 453

External Validation of Prediction Models

What Is External Validation, and Why Does It Matter for Prediction Models?

A prediction model may perform well during development yet fail when applied to new individuals or settings. Learn what constitutes genuine external validation, how to evaluate calibration and discrimination, why population differences matter, and how to interpret performance before adopting or updating a model.

PART 04 · SUBPART 39 · GUIDE 452

Internal Validation vs. Train-Test Split

What Is Internal Validation, and Why Is a Simple Train-Test Split Often Inefficient?

A random train-test split is familiar, but it can waste valuable development information and produce unstable performance estimates. Learn how internal validation works, why bootstrapping and cross-validation are often preferable, and how to select a method that reflects the complete prediction-model development process.

PART 04 · SUBPART 39 · GUIDE 451

Developing and Validating Prediction Models in the Same Dataset

Should Prediction Models Be Developed and Validated in the Same Dataset?

Prediction models can be developed and internally validated using the same dataset, but evaluating a fitted model on its training observations does not establish independent predictive performance. Learn how bootstrapping and cross-validation estimate optimism, why random splitting is often inefficient, and when external validation is necessary.

PART 04 · SUBPART 39 · GUIDE 450

Prediction Model Development vs. Validation vs. Updating

Model Development vs. Validation vs. Updating: Which Prediction Model Study Are You Doing?

Prediction model development, validation, and updating are distinct research activities with different objectives and analytical requirements. Learn how to identify the study you are conducting, distinguish independent validation from model modification, and avoid presenting apparent performance as evidence of generalizability.

PART 04 · SUBPART 39 · GUIDE 449

When to Use a Prediction Model Study

When Is a Prediction Model Study the Right Design?

Prediction model studies are appropriate when researchers need to estimate individual outcomes using information available at a defined decision point. Learn how to identify a genuine prediction question, distinguish prediction from explanation, and determine whether model development, validation, or updating is justified.

PART 04 · SUBPART 38 · GUIDE 448

Using Existing Data for Prognostic Research

When Is Existing Data Adequate for Prognostic Research?

Existing datasets can support rigorous prognostic research when they capture the right population, starting point, predictors, outcomes, and follow-up. Learn how to assess whether routinely collected records are fit for purpose and when missing information or measurement limitations make prognostic conclusions unreliable.

PART 04 · SUBPART 38 · GUIDE 447

Prognostic Model Development, Validation, and Updating

When Should a Prognostic Model Be Developed, Validated, or Updated?

Developing a new prognostic model is not always the most useful research contribution. Learn when existing models should be externally validated, recalibrated, revised, or replaced, and how to choose the appropriate research strategy based on evidence of performance and intended use.

PART 04 · SUBPART 38 · GUIDE 446

Explanation vs. Prediction in Prognostic Research

Should Prognostic Research Focus on Explanation or Prediction?

Prognostic research can investigate why future outcomes differ or estimate who is likely to experience them. Learn how explanatory and predictive objectives influence variable selection, regression analysis, validation, causal interpretation, and the practical use of research findings.

PART 04 · SUBPART 38 · GUIDE 445

Outcome Events and Prognostic Study Sample Size

How Many Outcome Events Does a Prognostic Study Need?

The number of outcome events often matters more than the total number of participants in prognostic research. Learn how event frequency, precision, model complexity, and research objectives determine sample size, and why the traditional ten-events-per-variable rule should not be treated as universal.

PART 04 · SUBPART 38 · GUIDE 444

Competing Events in Prognostic Studies

How Do Competing Events Affect Prognostic Study Design?

Competing events can prevent an outcome from occurring and fundamentally change how prognosis should be estimated. Learn how to identify competing risks, distinguish cumulative incidence from cause-specific hazards, avoid misleading Kaplan-Meier estimates, and choose analyses that match the prognostic question.

PART 04 · SUBPART 38 · GUIDE 443

Measuring Prognostic Factors Before Treatment

Should Prognostic Factors Be Measured Before Treatment Begins?

Prognostic factors are often measured before treatment to establish their relationship with future outcomes without incorporating treatment-induced changes. Learn when pretreatment measurement is essential, when post-treatment information is legitimate, and how measurement timing affects prognostic interpretation.

PART 04 · SUBPART 38 · GUIDE 442

Follow-Up Duration in Prognostic Studies

How Long Should Participants Be Followed in a Prognostic Study?

Follow-up duration determines which future outcomes a prognostic study can estimate and how precisely those outcomes can be measured. Learn how to select a meaningful prediction horizon, distinguish planned follow-up from observed follow-up, and handle censoring, attrition, competing events, and incomplete outcome information.

PART 04 · SUBPART 38 · GUIDE 441

Defining Time Zero in Prognostic Studies

How Do You Define the Starting Point for Prognosis?

The starting point of a prognostic study determines who enters the risk set, when outcomes begin to count, and which information can be used for prediction. Learn how to define time zero, distinguish diagnosis from treatment or discharge, handle delayed entry, and avoid biases caused by inconsistent starting points.

PART 04 · SUBPART 38 · GUIDE 440

Prognostic Factor vs. Prediction Model Study

Prognostic Factor vs. Prediction Model Study: What’s the Difference?

Prognostic factor studies examine whether characteristics are associated with future outcomes, while prediction model studies combine information to estimate outcome risk. Learn why these objectives require different analytical decisions, how statistical significance differs from predictive usefulness, and when each approach is appropriate.

PART 04 · SUBPART 38 · GUIDE 439

When to Use a Prognostic Study

When Is a Prognostic Study the Right Design?

A prognostic study examines what is likely to happen to people with a defined starting condition or characteristic. Learn when this design is appropriate, how to establish a meaningful starting point, select outcomes and follow-up periods, and distinguish prognosis from causal explanation and prediction-model development.

PART 04 · SUBPART 37 · GUIDE 437

Why Case-Control Designs Exaggerate Diagnostic Accuracy

Why Can Case-Control Diagnostic Designs Exaggerate Accuracy?

Case-control diagnostic studies can report impressive sensitivity and specificity because they compare advanced disease cases with unusually healthy controls. Learn how this sampling strategy can exaggerate accuracy, when it remains useful, and how to obtain more clinically applicable estimates.

PART 04 · SUBPART 37 · GUIDE 436

Case-Control Sampling in Diagnostic Accuracy

Can a Diagnostic Accuracy Study Use a Case-Control Sample?

Case-control sampling can be used in diagnostic accuracy research, but selecting known cases and controls may produce misleading estimates when the groups do not represent the intended testing population. Learn when this design is useful, what limitations it introduces, and how to improve its validity.

PART 04 · SUBPART 37 · GUIDE 435

Choosing Diagnostic Test Thresholds

Should Diagnostic Test Thresholds Be Chosen Before the Study?

Diagnostic thresholds determine which test results are classified as positive or negative. Learn why thresholds should usually be selected before confirmatory evaluation, when data-driven cutoff development is appropriate, and how to avoid optimistic accuracy estimates.

PART 04 · SUBPART 37 · GUIDE 434

Avoiding Verification Bias in Diagnostic Studies

How Do You Avoid Verification Bias in Diagnostic Accuracy Studies?

Verification bias occurs when disease status is established selectively or using different reference procedures, potentially distorting sensitivity and specificity. Learn how to recognize its mechanisms, design defensible verification strategies, and interpret results when complete verification is impossible.