Nursing Research Process & Evidence-Based Practice - Cheat Sheet
Your last-minute revision companion for the Nursing Research chapter. Master the 10-step research process, research designs, sampling methods, ethics, statistics interpretation, and EBP framework in 30 minutes. This sheet covers every high-yield concept tested in the PRC NLE and Board of Nursing competencies.
The 10-Step Nursing Research Process
Important facts
- Step 1: Identify and state the problem — define the gap and its significance.
- Step 2: Review the literature — synthesize existing knowledge, locate the gap.
- Step 3: Formulate the theoretical/conceptual framework — the lens guiding the study.
- Step 4: State research questions, objectives, or hypotheses.
- Step 5: Select the research design and methodology.
- Step 6: Identify population and sample; select sampling method.
- Step 7: Collect data using valid, reliable instruments.
- Step 8: Analyze data using appropriate statistics.
- Step 9: Interpret findings and draw conclusions.
- Step 10: Communicate/disseminate findings and recommend application.
Key definitions
Nursing Research
Example
A study examining the effect of early mobilization on hospital-acquired pneumonia rates in post-operative patients.
Systematic inquiry designed to develop, refine, and expand nursing knowledge to improve patient outcomes and inform clinical practice.
Research Gap
Example
No evidence exists on the optimal timing of patient education post-discharge for medication adherence in a specific Filipino population.
The difference between what is known and what needs to be known; the justification for conducting research.
Diagrams to know
- Linear sequence of the 10-step research process (flowchart)
- Relationship between problem statement, literature review, and research questions
Variables and Hypotheses
Important facts
- IV is what you GIVE or MANIPULATE; DV is what you MEASURE.
- H₀ is tested statistically; if p < 0.05, you REJECT H₀ and support H₁.
- A directional hypothesis predicts the direction of the relationship (e.g., 'will increase').
- A non-directional hypothesis predicts a relationship but not the direction (e.g., 'will differ').
- Only quantitative research typically uses hypotheses; qualitative research uses research questions.
Key definitions
Independent Variable (IV)
Example
Type of dressing (wound care intervention) in a study on wound healing rates.
The presumed cause or intervention that is manipulated by the researcher.
Dependent Variable (DV)
Example
Wound healing time (measured in days) as the outcome of different dressing types.
The presumed effect or outcome that is measured as a result of the IV.
Null Hypothesis (H₀)
Example
H₀: There is no difference in infection rates between sterile and non-sterile dressing techniques.
States that NO relationship or difference exists between variables; this is what statistical testing attempts to reject.
Alternative/Research Hypothesis (H₁)
Example
H₁: Sterile dressing technique results in lower infection rates than non-sterile technique.
States that a relationship or difference DOES exist; reflects the researcher's predicted outcome.
Diagrams to know
- IV → DV causal pathway diagram
- Hypothesis testing logic (reject vs. fail to reject H₀)
Quantitative vs. Qualitative Research
Important facts
- Quantitative = numerical, objective, deductive (specific to general); hypothesis-testing.
- Qualitative = narrative, subjective, inductive (general to specific); exploratory.
- Quantitative seeks generalizability to larger populations.
- Qualitative seeks deep understanding of a specific phenomenon or group.
- Mixed-methods research combines both approaches for comprehensive insight.
Key definitions
Quantitative Research
Example
A study measuring mean blood pressure reduction after a 12-week exercise program across 500 participants.
Measures variables numerically; tests hypotheses; seeks objective, generalizable results using deductive logic.
Qualitative Research
Example
In-depth interviews with patients to understand their lived experience of chronic pain management.
Explores meaning, experience, and process through narrative data; uses inductive logic to develop understanding.
Phenomenology (Qualitative Tradition)
Example
Understanding the experience of Filipino nurses working in intensive care units.
Explores the lived experience of a phenomenon; seeks to describe subjective meaning.
Grounded Theory (Qualitative Tradition)
Example
Developing a theory of how nurses cope with moral distress in resource-limited settings.
Develops a theory from data; the theory emerges from analyzing the data inductively.
Ethnography (Qualitative Tradition)
Example
Observing traditional healing practices in a rural Philippine community and their integration with Western medicine.
Studies a culture or subculture; describes beliefs, practices, and social interactions within a group.
Diagrams to know
- Continuum of quantitative vs. qualitative research
- Qualitative research traditions (phenomenology, grounded theory, ethnography, case study)
Quantitative Research Designs
Important facts
- TRUE EXPERIMENT = Manipulation + Control Group + Randomization (all three required).
- RCT is the GOLD STANDARD for establishing causation and sits at the TOP of evidence hierarchy.
- Quasi-experimental allows manipulation but LACKS randomization or control; more practical but weaker causal inference.
- Correlational research does NOT equal causation; r ≠ cause-and-effect.
- Cohort (prospective) follows people forward; case-control (retrospective) looks backward.
- Descriptive research is often the FIRST step; identifies what exists before testing why.
- Cross-sectional is a 'snapshot' at one time; cannot prove causation.
Key definitions
Experimental (True Experiment)
Example
Randomizing patients to receive either new wound care protocol (intervention) or standard care (control), with random assignment.
Has THREE hallmarks: manipulation of IV, control group, and randomization; STRONGEST design for establishing cause-and-effect.
Randomized Controlled Trial (RCT)
Example
RCT testing effectiveness of a new patient education video on medication adherence in hypertensive patients.
The gold standard experimental design; random assignment to intervention or control group; sits at the TOP of evidence hierarchy.
Quasi-Experimental
Example
Implementing a new fall-prevention protocol on one unit (intervention) and comparing outcomes to a unit without the protocol (no randomization).
Has manipulation of IV but LACKS randomization and/or control group; weaker causal inference than true experiment but often more feasible clinically.
Descriptive (Non-Experimental)
Example
Survey describing the prevalence of catheter-associated urinary tract infections (CAUTIs) in Philippine hospitals.
Describes characteristics, prevalence, or frequency of a phenomenon; NO manipulation; answers 'What is happening?'
Correlational (Non-Experimental)
Example
Examining the correlation between shift length and medication errors; correlation does NOT prove causation.
Examines relationships between variables WITHOUT manipulation; determines strength and direction of association.
Cohort Study (Prospective)
Example
Following 1000 nurses exposed to bloodborne pathogens for 5 years to determine infection rates.
Follows groups FORWARD over time; starts with exposure, then observes outcomes; weaker than RCT but stronger than case-control.
Case-Control Study (Retrospective)
Example
Comparing nurses who developed back injuries to nurses without, examining past ergonomic practices.
Compares people WITH an outcome to those WITHOUT, looking BACKWARD for exposure history.
Cross-Sectional Study
Example
A survey measuring stress levels and job satisfaction among nurses on a single day.
Data collected at ONE point in time; provides a 'snapshot'; cannot establish temporal order (cause before effect).
Diagrams to know
- Hierarchy of quantitative research designs (from strongest to weakest for causation)
- Experimental vs. quasi-experimental vs. non-experimental characteristics
Population and Sampling
Important facts
- PROBABILITY SAMPLING (random) supports generalizability; non-probability does NOT.
- Simple Random: every member has equal, independent chance (use random number tables, computer).
- Systematic: every kth member from a list (e.g., every 10th name); easy to execute.
- Stratified Random: divide population into strata, then randomly sample within each; ensures representation.
- Cluster (Multistage): random selection of groups/clusters, THEN sampling within; useful for geographically dispersed populations.
- Convenience: readily available subjects; WEAKEST, but common in qualitative and pilot studies.
- Quota: convenience with preset numbers per subgroup; still non-probability.
- Purposive/Judgmental: hand-picked for a specific characteristic; common in qualitative research.
- Snowball/Network: participants refer others; useful for hard-to-reach populations (e.g., undocumented migrants).
- LARGER SAMPLES reduce sampling error and increase power to detect real effects.
- Sample size should be determined BEFORE the study starts (often via power analysis).
Key definitions
Population
Example
All registered nurses working in public hospitals in the Philippines.
The entire group of interest from which the sample is drawn.
Target Population
Example
Registered nurses in Metro Manila public hospitals.
The group to which the researcher intends to generalize findings.
Accessible Population
Example
Registered nurses working in three specific public hospitals in Metro Manila (the hospitals that gave permission).
The portion of the target population that is actually available for study.
Sample
Example
200 nurses randomly selected from the three hospitals.
The subset of the population that is actually studied.
Sampling
Example
Using random number generation to select 200 nurses from a list of 1500.
The process of selecting a subset (sample) from a population.
Sampling Error
Example
If the population mean is 100 but the sample mean is 98, the sampling error is 2.
The difference between sample statistics and population parameters; ALWAYS present in sampling. Larger samples reduce sampling error.
Probability (Random) Sampling
Example
Simple random, systematic, stratified, cluster sampling methods.
Every member has a known, non-zero chance of selection; SUPPORTS generalizability.
Non-Probability Sampling
Example
Convenience, quota, purposive, snowball sampling methods.
Selection is NOT random; some members have no chance or unknown chance of selection; DOES NOT support generalizability.
Diagrams to know
- Probability vs. non-probability sampling methods comparison
- Population → Accessible Population → Sample relationship
Data Collection Methods & Instrument Quality
Common values
Value
≥0.70
Symbol
α
Quantity
Cronbach's Alpha (minimum acceptable)
Value
≥0.80
Symbol
α
Quantity
Cronbach's Alpha (good)
Important facts
- VALIDITY = Does it measure what it should? RELIABILITY = Does it measure consistently?
- An instrument can be RELIABLE without being VALID (consistently wrong).
- An instrument CANNOT be VALID without being RELIABLE (you cannot be accurate if you are inconsistent).
- AIM FOR BOTH: valid AND reliable instruments.
- Common data collection methods: questionnaires, interviews (structured/unstructured), observation, biophysiologic measures, existing records.
- Questionnaires: quick, economical, but low response rates; self-report bias.
- Interviews: rich data, flexible; time-consuming; interviewer bias possible.
- Observation: see actual behavior; observer bias; Hawthorne effect (behavior changes when observed).
- Biophysiologic measures: objective (e.g., BP, heart rate, blood glucose); equipment-dependent.
- Existing records: convenient, economical; incomplete, not designed for research purposes.
Key definitions
Validity
Example
A pain scale that truly measures pain (not anxiety or depression).
The instrument measures what it INTENDS to measure; it is accurate.
Content Validity
Example
A medication administration knowledge test that includes questions on calculations, drug interactions, and patient safety.
The instrument adequately samples the construct being measured; covers the full domain.
Construct Validity
Example
A stress scale that measures the psychological construct of stress (not just physical symptoms).
The instrument measures the theoretical construct it claims to measure.
Criterion Validity
Example
A depression screening tool validated against psychiatric diagnosis.
The instrument's scores correlate with an external criterion (gold standard).
Reliability
Example
A thermometer that gives the same reading when the patient's temperature is measured twice in succession.
The instrument yields CONSISTENT, reproducible results; it is dependable.
Test-Retest Reliability
Example
A stress scale administered on Monday and again on Friday to the same person should yield similar scores.
Consistency over TIME; the same instrument given to the same person at two time points yields similar scores.
Internal Consistency Reliability
Example
All items on a depression scale should measure depression (not anxiety or pain).
Consistency WITHIN the instrument; all items measure the same construct. Measured by Cronbach's alpha.
Interrater Reliability
Example
Two nurses assessing a patient's pain using the same pain scale should assign similar scores.
Consistency BETWEEN raters; different observers using the same instrument agree on their observations.
Cronbach's Alpha
Example
A scale with Cronbach's alpha of 0.82 has good internal consistency.
A coefficient measuring internal consistency; ranges 0–1. Values ≥0.70 are acceptable; ≥0.80 is good.
Diagrams to know
- Validity and reliability relationship (Venn diagram or 2x2 matrix)
- Types of validity and reliability at a glance
Research Ethics (CRITICAL for NLE)
Important facts
- INFORMED CONSENT is the cornerstone of research ethics; must be VOLUNTARY and INFORMED.
- Consent forms must be written in language participants understand (plain language, native language if appropriate).
- Minors need PARENTAL CONSENT + the minor's ASSENT.
- Pregnant women, prisoners, institutionalized persons are VULNERABLE; require additional protections.
- CONFIDENTIALITY ≠ ANONYMITY: confidentiality means researcher knows but keeps secret; anonymity means researcher doesn't know.
- Research must have MORE BENEFIT than RISK (beneficence > harm).
- Participants have the RIGHT TO WITHDRAW at ANY TIME without penalty.
- Data must be SECURE (locked, encrypted, limited access).
- In the PHILIPPINES, research oversight follows PHREB (DOST) guidelines and institutional ethics committees.
- RA 10173 (Data Privacy Act) mandates protection of personal information.
- Vulnerable groups get ENHANCED protection: extra review, closer monitoring, additional safeguards.
- Children cannot provide informed consent; parents/guardians consent, children assent (if age-appropriate).
Key definitions
Informed Consent
Example
A patient reading and signing a consent form before participating in a medication study, with questions answered.
Voluntary, informed agreement to participate; participant understands purpose, procedures, risks, benefits, and right to withdraw anytime without penalty.
Assent (Pediatric Research)
Example
A 10-year-old child agreeing to participate in a study after having it explained in simple language.
The child's agreement to participate (not legal consent, which comes from parents).
Confidentiality
Example
Data is linked to a subject ID code, and the key linking names to codes is kept secure.
Researcher knows the participant's identity but keeps data private; protects the identity.
Anonymity
Example
A survey with no identifying information; data is returned without names.
Researcher does NOT know the participant's identity; data cannot be traced to any individual.
Belmont Report Principles
Example
Nursing adds NONMALEFICENCE (do no harm) and FIDELITY (loyalty, honesty with participants).
Foundational ethical principles: RESPECT FOR PERSONS (autonomy), BENEFICENCE (maximize benefit), JUSTICE (fair distribution of benefits and burdens).
Institutional Review Board (IRB)
Example
In the Philippines, studies follow PHREB (Philippine Health Research Ethics Board) or institutional ethics committees.
A committee that reviews and approves research involving human subjects; ensures ethical compliance.
Vulnerable Populations
Example
Children, pregnant women, prisoners, cognitively impaired, critically ill patients, low-income individuals.
Groups with limited ability to protect themselves; require EXTRA safeguards.
Minimal Risk
Example
A questionnaire survey; observational study.
Research in which the probability and magnitude of harm is not greater than in everyday life.
Data Privacy Act of 2012 (RA 10173)
Example
Securing participant data, limiting access, obtaining consent for data use.
Philippine law protecting personal data; researchers must comply when handling participant information.
Diagrams to know
- Belmont principles and how they guide research ethics
- Vulnerable populations and their additional protections
- Informed consent process flowchart
Basic Statistics Interpretation
Formulas
Mean = Σ(x) / n
Meaning
Σ(x) = sum of all values; n = number of values; gives the average.
Watch out
Mean is SENSITIVE to outliers; one extreme value can pull the average up or down.
When to use
When data is normally distributed (no extreme outliers); best for interval/ratio data.
Median = middle value
Meaning
The value that divides data in half; 50th percentile.
Watch out
Median ignores values; changes in high or low values don't affect it (good when outliers exist).
When to use
When data is skewed (has outliers); best for ordinal data or non-normal distributions.
Mode = most frequent value
Meaning
The value that appears most often in the dataset.
Watch out
May not exist (all values occur equally) or may be multiple modes; doesn't reflect actual data well in non-categorical data.
When to use
The ONLY measure of central tendency for NOMINAL (category) data.
Standard Deviation (SD) = √[Σ(x - mean)² / n]
Meaning
Measures how spread out data is from the mean; larger SD = more variability.
Watch out
SD is in the SAME UNITS as the original data; don't confuse with variance (SD²).
When to use
To describe variability and identify outliers; used in calculating confidence intervals.
Correlation coefficient (r) ranges from -1 to +1
Meaning
r = -1 (perfect negative), r = 0 (no linear relationship), r = +1 (perfect positive); sign shows direction, magnitude shows strength.
Watch out
CORRELATION ≠ CAUSATION; even a strong correlation does NOT prove cause-and-effect.
When to use
When examining the LINEAR relationship between two continuous variables.
Common values
Value
p < 0.05
Symbol
p
Quantity
Statistical significance cutoff
Value
r = 0
Symbol
r
Quantity
Correlation: no relationship
Value
r = +0.70 to +1.00
Symbol
r
Quantity
Correlation: strong positive
Value
r = +0.30 to +0.69
Symbol
r
Quantity
Correlation: weak positive
Value
r = -0.70 to -1.00
Symbol
r
Quantity
Correlation: strong negative
Value
α = 0.05
Symbol
α
Quantity
Type I error (alpha) default
Value
1 - β ≥ 0.80 (80%)
Symbol
Power
Quantity
Minimum acceptable power
Important facts
- LEVELS OF MEASUREMENT determine which statistics to use: nominal (mode), ordinal (mode/median), interval (mode/median/mean), ratio (all).
- NORMAL DISTRIBUTION: 68% fall within ±1 SD, 95% within ±2 SD, 99.7% within ±3 SD.
- Mean is BEST for normal data; median is BEST for skewed data or ordinal data.
- Mode is the ONLY option for nominal data.
- p < 0.05 = statistically significant; p ≥ 0.05 = not statistically significant.
- Statistically SIGNIFICANT ≠ clinically SIGNIFICANT; small differences can be statistically significant in large samples.
- TYPE I ERROR (false positive) = concluding effect exists when it doesn't (α = 0.05 by convention).
- TYPE II ERROR (false negative) = concluding effect doesn't exist when it does (β; power = 1 - β).
- CORRELATION ranges -1 to +1; r = 0 = no linear relationship; positive = move together; negative = move opposite.
- t-test compares MEANS of TWO groups; ANOVA compares THREE or MORE groups.
- Chi-square tests ASSOCIATIONS between categorical (nominal/ordinal) variables.
- Confidence Interval (CI): range of values likely to contain the true population parameter; e.g., 95% CI = 95% confident the true value is within this range.
Key definitions
Nominal Data
Example
Sex (male/female), blood type (A, B, AB, O), religion.
Categories with NO order (mutually exclusive); only MEASURE: mode.
Ordinal Data
Example
Pain scale (0–10), educational level (elementary, high school, college), Likert scales (strongly agree to disagree).
Ordered categories but UNEQUAL intervals between levels; MEASURES: mode, median (not mean).
Interval Data
Example
Temperature in Celsius (0°C ≠ no temperature), IQ scores.
Ordered with EQUAL intervals between values but NO true zero; MEASURES: mode, median, mean.
Ratio Data
Example
Weight (0 kg = no weight), pulse, blood pressure, height, blood glucose.
Ordered, equal intervals, WITH a TRUE ZERO; MEASURES: all (mode, median, mean); STRONGEST level.
Normal (Bell-Curve) Distribution
Example
Height, IQ, blood pressure in healthy populations often approximate normal distribution.
Symmetrical, bell-shaped curve; mean = median = mode; 68–95–99.7% rule applies.
Skewed Distribution
Example
Income (right-skewed, tail to the right); age at death (left-skewed, tail to the left).
Asymmetrical; mean is pulled toward the tail.
p-value
Example
p = 0.03 means there is a 3% probability the result occurred by chance.
Probability that a result occurred by CHANCE alone; answers 'If the null hypothesis is true, how likely is this result?'
Statistical Significance
Example
If p = 0.03, the result IS statistically significant (reject H₀).
Result is UNLIKELY due to chance; conventionally p < 0.05 is considered statistically significant.
Type I Error (Alpha, α)
Example
Concluding a new drug is effective when it actually isn't.
REJECTING a TRUE null hypothesis; a FALSE POSITIVE; concluding there IS a difference when there isn't.
Type II Error (Beta, β)
Example
Concluding a new drug doesn't work when it actually does.
FAILING TO REJECT a FALSE null hypothesis; a FALSE NEGATIVE; concluding there is NO difference when there is.
Statistical Power (1 - β)
Example
Power = 0.80 means 80% chance of detecting a real effect if it exists; aim for ≥0.80.
The probability of correctly REJECTING a false null hypothesis; ability to detect a REAL effect.
Diagrams to know
- Normal distribution with 68–95–99.7% rule
- Type I vs. Type II error matrix
- Correlation strength interpretation (r values)
Evidence-Based Practice (EBP) Framework
Important facts
- EBP = Best Evidence + Clinical Expertise + Patient Values (all THREE components).
- ASK: Frame a focused PICO(T) question; avoid broad 'yes/no' questions.
- ACQUIRE: Search databases (PubMed, CINAHL, Cochrane); use keywords derived from PICO.
- APPRAISE: Judge study quality (validity, reliability, applicability, relevance); use critical appraisal tools.
- APPLY: Adapt evidence to your setting and patient; consider barriers, resources, patient preferences.
- ASSESS: Evaluate outcomes; did implementing this evidence improve results? Adjust as needed.
- EVIDENCE HIERARCHY (strongest to weakest): Systematic reviews/meta-analyses of RCTs > individual RCTs > quasi-experimental > cohort/case-control > qualitative/descriptive > expert opinion.
- RCT sits #2 in hierarchy because it is a single study; systematic reviews of RCTs are strongest because they synthesize all RCTs.
- QI uses local data for local improvement; research produces knowledge meant to be shared and generalizable.
- Clinical expertise + research evidence = better outcomes than either alone.
- Patient values/preferences are NOT optional; patient-centered care requires honoring them.
Key definitions
Evidence-Based Practice (EBP)
Example
Deciding to use a new wound care dressing based on current research, your clinical experience, and the patient's preference for frequency of dressing changes.
Integration of BEST AVAILABLE RESEARCH EVIDENCE + CLINICAL EXPERTISE + PATIENT VALUES/PREFERENCES to guide care decisions.
PICO(T) Framework
Example
In adult ICU patients (P), does early mobilization (I) versus standard care (C) reduce hospital-acquired pneumonia (O) within 30 days (T)?
Structure for formulating a focused clinical question: Population, Intervention, Comparison, Outcome, (Time).
The 5 A's of EBP
Example
- ASK a clinical question; 2) ACQUIRE evidence; 3) APPRAISE quality; 4) APPLY to practice; 5) ASSESS outcomes.
Ask, Acquire, Appraise, Apply, Assess — the steps for implementing EBP.
Systematic Review
Example
Cochrane review of interventions for catheter-associated urinary tract infection prevention.
Comprehensive, reproducible summary of all available evidence on a topic; sits at the TOP of evidence hierarchy.
Meta-Analysis
Example
Pooling data from 10 RCTs on hand hygiene interventions to calculate a combined effect size.
Statistical combination of results from multiple studies to determine overall effect; often part of a systematic review.
Clinical Guideline
Example
Philippine Health Ministry guidelines on infection prevention; ANA standards of care.
Evidence-based recommendations for clinical practice; developed by experts and organizations.
Quality Improvement (QI) vs. Research
Example
QI: Implementing a checklist to reduce medication errors in your hospital unit. Research: Testing whether a new drug reduces errors in multiple centers.
QI: Uses DATA to improve a SPECIFIC LOCAL PROCESS; does NOT generate generalizable knowledge. Research: GENERATES NEW KNOWLEDGE.
Diagrams to know
- The 5 A's of EBP cycle (circular/iterative process)
- Evidence hierarchy pyramid (strongest at top)
- EBP integration: evidence + expertise + patient values (Venn diagram)
Common Quantitative Tests (Quick Reference)
Important facts
- Parametric tests (t-test, ANOVA, Pearson r) assume NORMAL DISTRIBUTION; use for interval/ratio data.
- Non-parametric tests (Mann-Whitney, Wilcoxon, Spearman rho) do NOT assume normal distribution; use for ordinal/skewed data.
- t-test is for COMPARING TWO groups; ANOVA is for THREE or MORE groups.
- Paired t-test is for SAME group at two times (before/after); independent t-test is for DIFFERENT groups.
- Chi-square is ONLY for categorical data (nominal/ordinal); asks if association between two categories exists.
- Correlation (r) describes relationship strength/direction but DOES NOT prove causation.
- Spearman's rho is the non-parametric version of Pearson's r; used for ordinal data or non-normal continuous data.
Key definitions
t-test (independent samples)
Example
Comparing mean weight loss in a diet group versus a control group.
Compares MEANS of TWO INDEPENDENT groups (e.g., treatment vs. control).
t-test (paired/dependent)
Example
Comparing blood pressure before and after a medication in the same patients.
Compares MEANS of the SAME group at TWO TIME POINTS (before and after).
ANOVA (Analysis of Variance)
Example
Comparing test scores across three different teaching methods.
Compares MEANS of THREE or MORE groups.
Chi-Square Test
Example
Testing if there is an association between gender (male/female) and smoking status (yes/no).
Tests ASSOCIATION between TWO CATEGORICAL (nominal/ordinal) variables; compares observed vs. expected frequencies.
Pearson Correlation (r)
Example
Correlation between age and blood pressure in a sample of hypertensive patients.
Measures LINEAR relationship between TWO CONTINUOUS variables; r ranges -1 to +1.
Mann-Whitney U Test
Example
Comparing pain ratings (ordinal scale) between two groups.
Non-parametric test comparing TWO INDEPENDENT groups when data is NOT normally distributed or ordinal.
Wilcoxon Signed-Rank Test
Example
Comparing stress scores before and after an intervention in the same patients (ordinal data).
Non-parametric test comparing the SAME group at TWO TIME POINTS when data is NOT normally distributed.
Diagrams to know
- Decision tree: which statistical test to use (based on data type and group number)
Philippine Nursing Context & RA 9173
Important facts
- RA 9173 mandates nursing research competence as part of professional practice.
- Philippine nurses must be able to READ, UNDERSTAND, and APPLY research to practice.
- RA 10173 (Data Privacy Act) applies to all research with human subjects in the Philippines.
- Informed consent should be in language the participant understands (Filipino/Tagalog if appropriate).
- Research in Philippine healthcare settings requires institutional approval and ethical oversight.
- Filipino nurses working abroad must still adhere to RA 9173 standards and apply EBP.
Key definitions
RA 9173 (Nursing Act of 2002)
Example
Registered nurses are expected to stay current with evidence and question practices that lack scientific basis.
Philippine law regulating nursing practice; mandates competencies including ability to be a research consumer and apply EBP.
Board of Nursing (BON) Competencies
Example
Nurses must demonstrate ability to interpret research findings and integrate evidence into patient care decisions.
Expected nursing competencies set by the PRC; includes research literacy and EBP application.
Data Privacy Act 2012 (RA 10173)
Example
Securing informed consent in Filipino language; protecting participant identity; limiting data access.
Philippine data protection law; mandatory compliance when conducting research with human participants.
PHREB (Philippine Health Research Ethics Board)
Example
Studies in Philippine institutions must follow PHREB guidelines; some institutions have their own ethics committees aligned with PHREB.
National body under DOST providing ethical oversight and guidelines for health research in the Philippines.
Diagrams to know
- Philippine research ethics framework (PHREB, institutional committees, RA 10173)
Must remember
- A TRUE EXPERIMENT has ALL THREE: Manipulation, Control Group, AND Randomization. RCT is the gold standard (strongest for causation and #2 in evidence hierarchy after systematic reviews).
- INDEPENDENT VARIABLE (IV) = cause/intervention (what you GIVE/MANIPULATE). DEPENDENT VARIABLE (DV) = effect/outcome (what you MEASURE). NULL HYPOTHESIS states NO difference; statistical testing tries to REJECT it.
- PROBABILITY SAMPLING (random: simple, systematic, stratified, cluster) SUPPORTS GENERALIZABILITY. Non-probability (convenience, quota, purposive, snowball) does NOT. Larger samples reduce sampling error.
- VALIDITY = measures what it SHOULD (accurate). RELIABILITY = measures CONSISTENTLY (dependable). An instrument can be reliable without being valid (consistently wrong), but CANNOT be valid without being reliable.
- INFORMED CONSENT is the cornerstone of ethics: VOLUNTARY, INFORMED agreement. Minors need PARENTAL CONSENT + ASSENT. Vulnerable groups (children, pregnant women, prisoners, cognitively impaired, critically ill) need EXTRA protections. Data must be kept CONFIDENTIAL; RA 10173 (Data Privacy Act) applies in the Philippines.
- p < 0.05 = STATISTICALLY SIGNIFICANT (reject H₀). TYPE I ERROR = false positive (rejecting true H₀; α = 0.05 by convention). TYPE II ERROR = false negative (failing to reject false H₀; power = 1 - β; aim for ≥0.80).
- NORMAL DISTRIBUTION: 68% within ±1 SD, 95% within ±2 SD, 99.7% within ±3 SD. MEAN (sensitive to outliers, best for normal data) vs. MEDIAN (best for skewed/ordinal) vs. MODE (ONLY for nominal). Correlation coefficient r ranges -1 to +1: sign = direction, magnitude = strength. CORRELATION ≠ CAUSATION.
- EBP = BEST AVAILABLE RESEARCH EVIDENCE + CLINICAL EXPERTISE + PATIENT VALUES/PREFERENCES. Use PICO(T) to frame the question. The 5 A's: ASK → ACQUIRE → APPRAISE → APPLY → ASSESS. EVIDENCE HIERARCHY (strongest to weakest): Systematic reviews/meta-analyses of RCTs > individual RCTs > quasi-experimental > cohort/case-control > qualitative/descriptive > expert opinion.
- QUALITY IMPROVEMENT (QI) uses data to improve a LOCAL PROCESS (not generalizable). RESEARCH GENERATES NEW KNOWLEDGE meant to be shared. QUALITATIVE RESEARCH (phenomenology, grounded theory, ethnography, case study) explores MEANING and EXPERIENCE through narrative data (inductive).
- In the PHILIPPINES: RA 9173 (Nursing Act) mandates research competence. RA 10173 (Data Privacy Act) governs data protection. PHREB (Philippine Health Research Ethics Board, DOST) provides national ethical oversight. Institutions have ethics review committees. Board of Nursing (BON) expects nurses to be research consumers and apply EBP.
Last minute tips
- EXPERIMENTAL = Manipulation + Control + Randomization (ALL THREE required). If any ONE is missing, it's quasi-experimental or non-experimental. The RCT is #2 in evidence hierarchy (after systematic reviews); remember that systematic reviews SIT ABOVE single RCTs because they synthesize multiple studies.
- When you see 'CORRELATION,' immediately think 'NOT causation.' Even r = 0.95 does not prove cause-and-effect. Only experimental and quasi-experimental designs (with manipulation) can suggest causation.
- For STATISTICS: If the question asks 'which measure of central tendency?'—ask yourself: Is it normal distribution (mean), skewed (median), or nominal/categories (mode ONLY)? For hypothesis testing: p < 0.05 reject H₀; p ≥ 0.05 fail to reject H₀. Don't confuse 'statistically significant' with 'clinically significant.'
- ETHICS QUICK CHECK: Informed consent? Confidentiality/anonymity? Benefit > risk? Vulnerable group protection (minors = parental consent + assent)? IRB/ethics review? If any are missing, it's an ethics violation. RA 10173 is mandatory in the Philippines.
- EBP INTEGRATION: Research GENERATES knowledge; EBP APPLIES knowledge. Remember: BEST EVIDENCE alone isn't EBP—you need clinical judgment AND patient preferences. Use PICO(T) to frame clinical questions. When appraising evidence, systematic reviews of RCTs are your gold standard; expert opinion is weakest. Always assess outcomes after applying evidence.
Comparison tables
Experimental vs. Quasi-Experimental vs. Non-Experimental Designs
| Feature | Experimental (True Experiment) | Quasi-Experimental | Non-Experimental |
|---|---|---|---|
| Manipulation of IV | YES (required) | YES | NO |
| Control Group | YES (required) | OFTEN NOT or weak | NO |
| Randomization | YES (required) | NO (lacks this) | NO |
| Strength for Causation | STRONGEST | MODERATE | WEAK |
| Feasibility in Clinical Setting | OFTEN NOT FEASIBLE (ethical/practical barriers) | MORE FEASIBLE | MOST FEASIBLE |
| Example | RCT of new wound care vs. standard care (random assignment) | Comparing fall rates on unit with new protocol (no randomization) vs. unit without | Survey describing CAUTI prevalence in hospitals |
Probability (Random) vs. Non-Probability Sampling
| Feature | Probability Sampling | Non-Probability Sampling |
|---|---|---|
| Selection Process | Random; every member has known chance | NOT random; selection bias possible |
| Supports Generalizability | YES (strong) | NO (weak) |
| Sampling Error | Can be calculated and controlled | Cannot be estimated |
| Common Types | Simple random, systematic, stratified, cluster | Convenience, quota, purposive, snowball |
| Cost/Time | Often more time-consuming and expensive | Quick and inexpensive |
| Best Use | Quantitative research when generalizability is goal | Qualitative research, exploratory studies, hard-to-reach populations |
Validity vs. Reliability
| Aspect | Validity | Reliability |
|---|---|---|
| Definition | Measures what it INTENDS to measure (accuracy) | Yields CONSISTENT, reproducible results (dependability) |
| Question | Are you measuring the RIGHT thing? | Are you measuring it the SAME way each time? |
| Can Exist Without the Other | NO (cannot be valid without being reliable) | YES (can be reliable but not valid — consistently wrong) |
| Types | Content, construct, criterion validity | Test-retest, internal consistency, interrater reliability |
| Measurement | Expert judgment, correlation with gold standard | Cronbach's alpha (α ≥0.70), correlation coefficients |
Qualitative Research Traditions at a Glance
| Tradition | Focus | Guiding Question | Example |
|---|---|---|---|
| Phenomenology | Lived experience; meaning of a phenomenon | What is the experience of...? | Lived experience of nurses working in ICU; what does it feel like? |
| Grounded Theory | Theory development; process; how behavior is explained | How do people cope with/experience...? | How do nurses manage moral distress? (theory emerges) |
| Ethnography | Culture; beliefs, values, social interactions of a group | What are the cultural norms/values in...? | Healing practices in a rural Philippine community |
| Case Study | In-depth exploration of a single case/small group | What is unique about this case? | One patient's journey through recovery; one unit's improvement process |
Measures of Central Tendency: When to Use
| Measure | Definition | Best For (Data Type) | Advantage | Disadvantage |
|---|---|---|---|---|
| Mean | Sum of all values ÷ number of values (average) | Interval/ratio data; NORMAL distribution | Uses all data; mathematically stable | VERY SENSITIVE to outliers; can be misleading if extreme values exist |
| Median | Middle value; 50th percentile | Ordinal data; SKEWED distributions; any outliers | NOT affected by outliers; best for non-normal data | Ignores extreme values; less precise than mean in normal data |
| Mode | Most frequently occurring value | ONLY FOR NOMINAL (category) data | ONLY measure for categories; simple to understand | May not exist; may have multiple modes; doesn't fully represent data |
p-Value Interpretation & Statistical Significance
| p-Value | Interpretation | Decision | Clinical Meaning |
|---|---|---|---|
| p < 0.001 | Very strong evidence against H₀ | REJECT H₀ | Result is HIGHLY unlikely due to chance; result is very statistically significant |
| p = 0.01 to 0.05 | Strong evidence against H₀ | REJECT H₀ | Result is statistically SIGNIFICANT; likely a real effect, not chance |
| p = 0.05 | Borderline; conventional cutoff | REJECT H₀ (at p < 0.05 convention) | At the threshold of significance |
| p > 0.05 | Weak evidence against H₀ | FAIL TO REJECT H₀ | Result is NOT statistically significant; may be due to chance; no strong evidence of effect |
Type I vs. Type II Error
| Error Type | Definition | What Happened | Example | Consequence |
|---|---|---|---|---|
| Type I Error (α) | Rejecting a TRUE null hypothesis | FALSE POSITIVE; conclude effect exists when it DOESN'T | Study says new drug is effective, but it actually isn't | Implement ineffective intervention; waste resources; potential harm |
| Type II Error (β) | Failing to reject a FALSE null hypothesis | FALSE NEGATIVE; conclude NO effect when one EXISTS | Study says drug is ineffective, but it actually IS effective | Reject beneficial treatment; miss opportunity to help patients |
Diagrams
The 10-Step Nursing Research Process
Linear flowchart showing the orderly sequence of the 10-step quantitative research process, from problem identification through dissemination. This is the foundational framework for all nursing research.
Research Design Hierarchy: Causality Strength
Hierarchical flowchart showing research design strength for establishing causation. True experiments with all three components (manipulation, control, randomization) are strongest. Note: Systematic reviews of RCTs sit at the TOP of the evidence hierarchy.
Sampling Methods: Probability vs. Non-Probability
Mind map contrasting probability (random) and non-probability (non-random) sampling methods. Probability methods support generalizability; non-probability does not. This is a critical distinction for evaluating research quality.
Validity vs. Reliability: Can You Have One Without the Other?
Decision flowchart showing the relationship between validity and reliability. Key insight: An instrument can be reliable without being valid (consistently wrong), but CANNOT be valid without being reliable.
Research Ethics: Protection of Human Subjects
Comprehensive flowchart of research ethics protections. Highlights informed consent, confidentiality, beneficence, autonomy, and ethics review. Notes special protections for minors and vulnerable groups. Includes Philippine regulatory context.
Choosing the Right Measure of Central Tendency
Decision tree for selecting the appropriate measure of central tendency based on data type and distribution. Critical for choosing correct statistical analysis.
Statistical Hypothesis Testing Logic
Flowchart showing the logic of statistical hypothesis testing. Illustrates the decision to reject or fail to reject the null hypothesis based on p-value, and notes the risks of Type I and Type II errors.
Evidence-Based Practice: Integration & Application
Mind map showing the three pillars of EBP (research evidence, clinical expertise, patient values) and the 5 A's framework for implementation. Emphasizes that all three pillars must be integrated for true EBP.
Normal Distribution: The 68–95–99.7 Rule
Pie chart showing the distribution of data in a normal (bell) curve. Critical for understanding where data falls relative to the mean and standard deviation—essential for interpreting research results.