FRCR ONCOLOGY PART 1 - ALL FREE PRACTICE MCQS: FRCR Oncology Part 1: MEDICAL STATISTICS

In a Cox proportional hazards model, violation of the proportionality assumption is most effectively detected using:
A. Wald test
B. Kaplan-Meier plot
C. Schoenfeld residuals
D. Log-rank test
E. AIC comparison
Which Bayesian concept quantifies the impact of prior beliefs on the interpretation of study data?
A. Posterior probability
B. Likelihood function
C. Predictive interval
D. Bayes factor
E. Null distribution
In competing risks analysis, which estimate best accounts for the probability of a specific event type?
A. Kaplan-Meier survival estimate
B. Hazard ratio
C. Subdistribution cumulative incidence function
D. Cox regression
E. Odds ratio
Which method adjusts for time-dependent covariates in survival analysis?
A. Kaplan-Meier method
B. Cox model with time-varying covariates
C. Log-rank test
D. Parametric Weibull model
E. Nelson-Aalen estimator
A study finds non-proportional hazards between treatment groups. What is the best alternative approach?
A. Increase sample size
B. Use time-dependent Cox model
C. Switch to odds ratio
D. Remove outliers
E. Use Student’s t-test
In adaptive clinical trial design, what is the principal advantage of group-sequential analysis?
A. Control over Type II error
B. Ability to correct confounding
C. Interim stopping rules to preserve ethical oversight
D. Ensures blinding
E. Eliminates the need for randomisation
The false discovery rate is most important in which type of statistical context?
A. Meta-analysis
B. Regression modelling
C. Multiple hypothesis testing
D. Kaplan-Meier survival
E. Case-control studies
What does a high I² statistic (e.g., 80%) in meta-analysis imply?
A. Minimal heterogeneity
B. Homogeneous study designs
C. High between-study variability
D. High within-study variance
E. Small sample size bias
What is a primary limitation of the intention-to-treat analysis in superiority trials?
A. Inflation of Type I error
B. Increased power
C. Underestimation of treatment effect
D. Over-adjustment for confounding
E. Inability to generalise
A study with a large sample and small effect size reports a p-value of <0.001. Which is the most appropriate conclusion?
A. Clinically significant result
B. Large treatment effect
C. Statistically significant but possibly clinically unimportant
D. High power
E. Underpowered
Which method is most appropriate to estimate treatment effect in presence of censoring?
A. Risk ratio
B. Odds ratio
C. Kaplan-Meier survival estimate
D. Incidence rate
E. Attributable risk
A forest plot is used in meta-analysis to show:
A. Time to event
B. Correlation
C. Pooled effect sizes and confidence intervals
D. Effect size only
E. p-values only
Which test best assesses goodness-of-fit in logistic regression?
A. Hosmer-Lemeshow test
B. Wald test
C. Log-rank test
D. Pearson correlation
E. T-test
What defines a non-inferiority trial?
A. Seeks to prove no difference
B. Establishes treatment is not worse than control by a pre-specified margin
C. Tests superiority
D. Compares survival curves
E. Uses odds ratio as primary measure
A Type III error refers to:
A. Accepting a false null hypothesis
B. Rejecting a true null hypothesis
C. Correctly rejecting null hypothesis for the wrong reason
D. Using wrong alpha level
E. Misreporting effect size
What does an ROC curve illustrate?
A. Specificity vs. time
B. Sensitivity vs. specificity
C. Sensitivity vs. 1-specificity
D. p-value vs. power
E. False positive vs. false negative
What is the assumption behind using a log-rank test?
A. Gaussian distribution
B. Equal sample sizes
C. Proportional hazards
D. Independent observations
E. Equal means
The ‘number needed to treat’ (NNT) is:
A. Reciprocal of incidence
B. Reciprocal of absolute risk reduction
C. Absolute risk increase
D. Relative risk
E. Odds ratio
When is Fisher's exact test more appropriate than Chi-squared test?
A. Large sample size
B. Continuous variables
C. Small expected frequencies
D. Multivariate data
E. Gaussian distribution
The purpose of bootstrapping in statistical inference is to:
A. Adjust for confounding
B. Increase sample size
C. Estimate sampling distribution
D. Increase power
E. Perform hypothesis testing
Which concept reflects the probability that a statistically significant result is actually a true positive?
A. Power
B. Posterior probability
C. p-value
D. False discovery rate
E. Type I error
An overfitted model in regression is likely to:
A. Be generalisable
B. Have poor predictive performance on new data
C. Improve confidence intervals
D. Reduce variance
E. Increase bias
Kaplan-Meier estimates assume:
A. Constant hazard over time
B. Independent censoring
C. Exponential distribution
D. Normality of survival time
E. Uniform dropout
What is the primary function of the Bonferroni correction?
A. Reduce Type II error
B. Adjust for cluster design
C. Control family-wise error rate
D. Increase power
E. Adjust for covariates
A Type II error rate of 0.20 implies a power of:
A. 10%
B. 20%
C. 80%
D. 90%
E. 95%
What is the role of degrees of freedom in a t-distribution?
A. Measure of effect size
B. Determines critical values
C. Sets alpha level
D. Describes skewness
E. Corrects for multiple testing
Which test is most appropriate to compare survival between two groups?
A. ANOVA
B. Log-rank test
C. Pearson’s chi-squared
D. Fisher’s exact test
E. Wilcoxon signed-rank
What is the main purpose of a Bland-Altman plot?
A. Assess survival
B. Show risk ratios
C. Assess agreement between two methods
D. Estimate hazard ratios
E. Compare variances
In multivariable regression, the variance inflation factor (VIF) is used to:
A. Estimate R²
B. Detect collinearity
C. Measure effect size
D. Calculate p-value
E. Test interactions
A funnel plot is used to:
A. Assess confounding
B. Evaluate randomisation
C. Detect publication bias
D. Estimate treatment effects
E. Visualise power calculations
What does a 95% confidence interval NOT imply?
A. That the interval contains the true population value 95% of the time
B. That there is a 95% probability the true value lies within this interval
C. That 95 out of 100 such intervals will contain the true mean
D. That the interval is based on a sample
E. That the method is repeatable
Which statistical method estimates relative hazard without assuming constant risk over time?
A. Kaplan-Meier
B. Cox proportional hazards model
C. Logistic regression
D. Paired t-test
E. ANOVA
The median is preferred over the mean when:
A. The data are normally distributed
B. The data are skewed
C. The sample is small
D. The variance is low
E. Data are categorical
The main assumption of ANCOVA is:
A. Non-linearity
B. Homogeneity of regression slopes
C. Random effects
D. Censored data
E. Equal medians
In a multivariate logistic regression model, a large standard error of a coefficient suggests:
A. High precision
B. Low significance
C. Multicollinearity
D. Underfitting
E. Overdispersion
In a trial, the number needed to harm (NNH) is 25. This means:
A. 25 patients benefit from treatment
B. 1 in 25 patients will be harmed
C. The effect is negligible
D. Risk ratio is 25
E. Treatment is safe
The Bonferroni correction becomes overly conservative when:
A. Number of tests is large
B. Sample size is small
C. Effect size is large
D. Using paired data
E. Data are normally distributed
Which measure compares observed to expected event counts in survival analysis?
A. Hazard ratio
B. Log-rank test statistic
C. Kaplan-Meier estimator
D. Schoenfeld residual
E. C-index
What is the primary limitation of the odds ratio as a measure of effect?
A. Only valid in case-control studies
B. Cannot be calculated for rare events
C. Overestimates risk when event is common
D. Cannot be interpreted without confidence intervals
E. Biased by censoring
Which of the following best describes the difference between specificity and sensitivity?
A. Sensitivity measures false positives, specificity measures false negatives
B. Sensitivity measures true positives, specificity measures true negatives
C. Sensitivity and specificity measure the same
D. Sensitivity relates to prevalence
E. Specificity relates to incidence
The principle of randomisation in clinical trials is to:
A. Reduce bias by equalising confounders
B. Maximise power
C. Ensure blinding
D. Reduce sample size
E. Increase effect size
Which of the following statistics is most affected by outliers?
A. Median
B. Mode
C. Mean
D. IQR
E. Range
What is the hazard ratio when the two groups have identical survival?
A. 0
B. 1
C. <1
D. >1
E. Undefined
The null hypothesis in a superiority trial typically states:
A. There is no difference between groups
B. Treatment is inferior
C. Treatment is superior
D. Treatment harms patients
E. Sample size is adequate
In a clinical trial, blinding reduces:
A. Sampling error
B. Selection bias
C. Performance and detection bias
D. Attrition bias
E. Confounding
Which of the following is NOT a measure of central tendency?
A. Median
B. Mode
C. Variance
D. Mean
E. Midrange
Which method assesses the fit of a survival model?
A. Likelihood ratio test
B. Kaplan-Meier curve
C. Cox-Snell residuals
D. T-test
E. ANOVA
The specificity of a diagnostic test is:
A. The proportion of true positives correctly identified
B. The proportion of false positives
C. The proportion of true negatives correctly identified
D. The probability of disease given a positive test
E. The probability of no disease given a negative test
Which of the following can cause bias in clinical trials?
A. Randomisation
B. Allocation concealment failure
C. Blinding
D. Large sample size
E. Prospective design
A Kaplan-Meier curve drops vertically at:
A. Censoring points
B. Time points when events occur
C. Random intervals
D. Times of study entry
E. Follow-up completion

The Pearson correlation coefficient measures:
A. Linear association between two variables
B. Causation
C. Differences in means
D. Probability of occurrence
E. Odds of outcome
Which of the following is NOT a parametric test?
A. ANOVA
B. Student’s t-test
C. Mann-Whitney U test
D. Paired t-test
E. Linear regression
In a meta-analysis, the fixed effects model assumes:
A. Study effects vary randomly
B. Single common true effect size
C. Between-study heterogeneity
D. Variable sample sizes only
E. Bias from publication
The Kaplan-Meier estimator is a:
A. Parametric method
B. Non-parametric estimator of survival function
C. Regression model
D. Hypothesis test
E. Variance estimate
The main advantage of using a matched case-control study is to:
A. Increase sample size
B. Control confounding by design
C. Assess causation
D. Randomise subjects
E. Eliminate bias
The positive predictive value (PPV) depends on:
A. Sensitivity alone
B. Specificity alone
C. Disease prevalence
D. Sample size
E. Study design
What is the consequence of multiple testing without adjustment?
A. Increased Type II error
B. Increased power
C. Increased Type I error

In survival analysis, what does censoring indicate?
A. Event occurrence
B. Loss of follow-up or study end without event
C. Random sampling
D. Treatment failure
E. Data exclusion
Which statistic measures inter-rater reliability for categorical data?
A. Pearson correlation
B. Cohen’s kappa
C. Intraclass correlation coefficient
D. Cronbach’s alpha
E. Spearman’s rho
The hazard ratio of 0.5 means:
A. No difference in risk
B. 50% lower hazard in treatment group
C. 50% higher hazard in treatment group
D. Hazard doubles
E. Hazard is half in control group
The main aim of random effects meta-analysis is to:
A. Estimate a single true effect
B. Account for heterogeneity between studies
C. Increase sample size
D. Remove bias
E. Standardise effect size
In a diagnostic test, likelihood ratio positive (LR+) is defined as:
A. Sensitivity / (1 - Specificity)
B. (1 - Sensitivity) / Specificity
C. Specificity / (1 - Sensitivity)
D. False positive rate
E. Positive predictive value
The p-value indicates:
A. Probability null hypothesis is true
B. Probability of observing the data given null hypothesis is true
C. Probability alternative hypothesis is true
D. Magnitude of effect
E. Probability of Type II error
Which sampling method reduces selection bias by giving all individuals equal chance of selection?
A. Convenience sampling
B. Stratified sampling
C. Random sampling
D. Purposive sampling
E. Systematic sampling
What does the term ‘intention-to-treat’ mean?
A. Only participants completing the study analyzed
B. Participants analyzed as per initial randomization regardless of adherence
C. Exclusion of dropouts
D. Analysis based on treatment received
E. Post hoc subgroup analysis
Which test compares means of three or more independent groups?
A. Student’s t-test
B. Paired t-test
C. ANOVA
D. Chi-squared test
E. Wilcoxon signed-rank test

The null hypothesis in a diagnostic accuracy study typically states:
A. Test has no discriminatory ability
B. Test is perfect
C. Test is better than comparator
D. Test is worse than comparator
E. Test results are correlated with prevalence
In hypothesis testing, what does ‘power’ refer to?
A. Probability of Type I error
B. Probability of rejecting null hypothesis when it is false
C. Probability of Type II error
D. Probability of false positive
E. Probability of false negative
Which measure is most appropriate for assessing agreement of continuous data between two methods?
A. Pearson correlation
B. Bland-Altman plot
C. Kappa statistic
D. Chi-squared test
E. ANOVA
What does the Cox-Snell residuals plot assess?
A. Model calibration in survival analysis
B. Normality of residuals
C. Collinearity
D. Interaction effects
E. Independence of observations
Which approach can address missing data in clinical trials?
A. Complete case analysis
B. Last observation carried forward
C. Multiple imputation
D. All of the above
E. None of the above
The median survival time is best estimated by:
A. Mean time to event
B. Kaplan-Meier survival curve
C. Cox regression
D. Log-rank test
E. Hazard function

Which of the following increases the likelihood of Type I error?
A. Small sample size
B. Multiple comparisons without adjustment
C. Blinding
D. Randomisation
E. Stratification
The Kaplan-Meier method is used to estimate:
A. Risk difference
B. Survival probabilities over time
C. Hazard ratios
D. Odds ratios
E. Regression coefficients
In a forest plot, what does the diamond shape represent?
A. Individual study estimate
B. Pooled effect estimate
C. Confidence interval for one study
D. Sample size
E. Heterogeneity
Which measure is NOT affected by prevalence?
A. Positive predictive value
B. Negative predictive value
C. Sensitivity
D. Specificity
E. Likelihood ratios
Which of the following best describes an intention-to-treat analysis?
A. Analyzes only participants who completed treatment as planned
B. Analyzes participants according to the group they were originally assigned
C. Excludes protocol violations
D. Analyzes according to treatment actually received
E. Uses per-protocol population
Which test is appropriate to compare two paired samples with non-normal distribution?
A. Paired t-test
B. Wilcoxon signed-rank test
C. Chi-squared test
D. Mann-Whitney U test
E. ANOVA
The number needed to treat (NNT) is:
A. The number of patients that need to be treated to prevent one additional bad outcome
B. The number of patients treated in a trial
C. The number of adverse events in treatment group
D. The relative risk reduction
E. The absolute risk increase
In clinical trials, the term ‘blinding’ refers to:
A. Random allocation of participants
B. Keeping participants, clinicians, or assessors unaware of group assignments
C. Sample size calculation
D. Statistical analysis plan
E. Interim analysis
Which of the following describes the hazard function?
A. Probability of surviving past a time point
B. Instantaneous event rate at a given time
C. Cumulative incidence
D. Risk difference
E. Odds ratio
Which of these is NOT a type of censoring in survival analysis?
A. Right censoring
B. Left censoring
C. Interval censoring
D. Random censoring
E. Informative censoring
Which of the following is used to assess heterogeneity in meta-analysis?
A. I² statistic
B. P-value
C. Hazard ratio
D. Kaplan-Meier curve
E. Relative risk
The Wilcoxon rank-sum test is used for:
A. Comparing means of two independent samples with normal distribution
B. Comparing medians of two independent samples without normal distribution
C. Comparing paired samples
D. Testing proportions
E. Regression analysis
The type II error (β) is:
A. Rejecting null hypothesis when it is true
B. Accepting null hypothesis when it is false
C. Probability of false positive
D. Statistical power
E. Confidence interval
Which of the following reflects the spread of data around the mean?
A. Median
B. Variance
C. Mode
D. Range
E. Skewness
What is the primary purpose of stratified randomisation?
A. Increase sample size
B. Balance prognostic factors across treatment groups
C. Blind investigators
D. Reduce Type I error
E. Adjust for multiple testing
The log-rank test is used to:
A. Compare survival distributions between two or more groups
B. Estimate hazard ratios
C. Model time-to-event data with covariates
D. Test for normality
E. Calculate odds ratios
Which term describes a study that follows participants forward in time from exposure to outcome?
A. Case-control study
B. Cross-sectional study
C. Prospective cohort study
D. Retrospective cohort study
E. Randomised controlled trial
What does a 95% confidence interval mean?
A. 95% probability that the true parameter lies within the interval
B. 5% chance of Type I error
C. If the study were repeated many times, 95% of the intervals would contain the true parameter
D. The result is statistically significant
E. None of the above
Which of the following describes an effect modifier?
A. A variable that influences the strength or direction of the association between exposure and outcome
B. A confounder
C. A variable controlled by randomisation
D. A variable that causes bias
E. A variable that is a mediator
The incidence rate is best defined as:
A. Number of new cases per population at risk per unit time
B. Number of existing cases at a time point
C. Probability of disease given exposure
D. Prevalence
E. Relative risk
Which statistical test is used for association between two categorical variables?
A. Student’s t-test
B. Chi-squared test
C. ANOVA
D. Pearson correlation
E. Cox regression
What is a type I error (α)?
A. Accepting the null hypothesis when it is false
B. Rejecting the null hypothesis when it is true
C. Probability of failing to detect a true effect
D. Statistical power
E. Confidence interval
In a survival analysis, proportional hazards assumption means:
A. Hazard ratios are constant over time
B. Survival probabilities are equal between groups
C. Events occur at the same rate
D. Kaplan-Meier curves do not cross
E. None of the above
Which is the best measure of central tendency for skewed data?
A. Mean
B. Median
C. Mode
D. Variance
E. Range
The area under the ROC curve (AUC) measures:
A. Diagnostic accuracy of a test
B. Prevalence of disease
C. Probability of false negatives
D. Probability of true negatives
E. None of the above
The confidence interval that crosses the null value indicates:
A. Statistically significant result
B. No statistically significant difference
C. Large sample size
D. Type II error
E. Bias
Which of the following is a non-parametric test?
A. Student’s t-test
B. ANOVA
C. Mann-Whitney U test
D. Linear regression
E. Paired t-test
In a diagnostic test, a high negative likelihood ratio (LR-) indicates:
A. Test is good at ruling in disease
B. Test is good at ruling out disease
C. High false positive rate
D. Low specificity
E. None of the above

ANSWERS

FRCR ONCOLOGY PART 1 - ALL FREE PRACTICE MCQS

Sunday, May 18, 2025

FRCR Oncology Part 1: MEDICAL STATISTICS - 10 (100 QUESTIONS, ANSWERS BELOW)