Biostats Midterm ANKI

A diagnostic test for a disease has a sensitivity of 90% and a specificity of 80%. What is the positive likelihood ratio (LR+) for this test? Based on your answer, would this be considered strong evidence to rule in the disease?

LR+ = sens/(1-spec) = 0.9/0.2 = 4.5 Moderate evidence to rule in disease but not strong (>10).

A diagnostic test for a disease has a sensitivity of 70% and a specificity of 75%. What is the positive likelihood ratio (LR+) for this test? Based on your answer, would this be considered strong evidence to rule in the disease?

LR+ = sens/(1-spec) = 0.7/0.25 = 2.8 Weak evidence, not strong enough to rule in disease.

Match the study type to its purpose: Study Type: Case-control study Cohort study Randomized controlled trial Diagnostic accuracy study Purpose: Therapy Diagnosis Prognosis Harm

Match the study type to its purpose: Study Type: Case-control study Cohort study Randomized controlled trial Diagnostic accuracy study Purpose: Therapy Diagnosis Prognosis Harm case control - harm cohort - prognosis RCT - therapy diagnostic accuracy study - diagnosis

You create a scatter plot of your data and notice a clear curve instead of a straight line. Should you use Pearson’s correlation or Spearman’s correlation? Why?

Spearman is better because Pearson assumes a linear relationship and this data is curved.

A student collected data on hours studied and test scores. They got an r value of 0.61. What does this tell you about the relationship between hours studied and test performance?

R of 0.61 indicates a strong positive correlation between the variables. As study time increases, test performance tends to improve.

A study is designed with 64 total participants (32 in each group), an alpha level of 0.05, and a medium effect size of 0.5. What is the approximate power of the study, and what does that tell you about its ability to detect a true effect?

32*0.5*0.05 = 0.8 = power. You have an 80% chance of detecting a medium-sized effect

Name three things that affect the power of a study. For each one, explain how you would change it to increase the power of your research.

Sample size: Increase to increase power;

effect size: larger effects are easier to detect so make the study to maximize effect;

alpha level: raising alpha increases power.

A study comparing blood pressure between two groups has a p-value of 0.031 and a 95% confidence interval of [1.2 mmHg, 10.4 mmHg]. What do these results tell you about whether the treatment was effective?

The treatment is effective because the p-value is less than 0.05 and the CI does not include 0.

What type of statistical test would you use for each of the following? a. Comparing IOP before and after treatment in the same patients b. Comparing average exam scores between two different classes c. Looking at the relationship between eye color and diagnosis d. Comparing average BMI across four different diets

a. paired t-test b. independent t-test c. chi-squared d. ANOVA

Differentiate between Bayesian and Frequentist interpretations of probability using a medical example (e.g., diagnosis, risk prediction, or treatment outcome).

A doctor estimates a 50% chance a patient has a retinal detachment based on symptoms (prior), then revises this to 95% after a posterior fundus exam. Frequentist Interpretation: Probability is the frequency of events over time.Example: In a clinical trial, if a drug cures 40 out of 100 patients, the response rate is 40%.

A disease test has 95% sensitivity and 90% specificity. Explain what each of these numbers means in plain language.

95% Sensitivity means the test correctly identifies 95% of those who truly have the disease (true positives).

90% Specificity means the test correctly identifies 90% of those who truly do not have the disease (true negatives).

A rapid test was used on a population consisting of 200 true cases and 300 healthy individuals. The test correctly identified 180 of the true cases and incorrectly identified 40 healthy individuals as positive. Calculate the sensitivity and specificity of the test.

Disease (+) Disease (-) Test (+) 180 40 220 Test (-) 20 260 280 200 300 500 Sensitivity = TP / (TP + FN) = 180 / (180 + 20) = 180 / 200 = 0.90 or 90%Sensitivity = 90%Specificity = TN / (TN + FP) = 260 / (260 + 40) = 260 / 300 = 0.867 or 86.7%Specificity = 86.7%

A patient has a 30% chance of having a disease before testing. After testing positive, the chance increases to 75%. What kind of probability is this? Explain using the Bayesian perspective.

This is Bayesian probability because it updates the prior probability (30%) with new evidence (positive test result) to produce a posterior probability (75%).

What is the primary purpose of scientific research?

To identify the truth in a world of information and misinformation.

Why is ethics essential in science?

Because science is based on the moral principle that telling the truth is good and necessary for objectivity.

What is empirical evidence?

Information gathered through direct observation, measurement, or experimentation.

List the main steps of the scientific method.

Observation, Questioning, Hypothesis, Experimentation, Data Analysis, Conclusion, Communication.

Why does science speak in terms of probability and confidence instead of absolute certainty?

Because scientific conclusions are based on evidence and statistical analysis, not definitive proof.

Give an example of how statistics can be applied in optometry.

Using data to evaluate treatment effectiveness or optimize business strategy.

What can happen in healthcare if someone believes misinformation due to poor statistical understanding?

They may make harmful decisions, misinterpret risks, or reject effective treatments.

How can biostatistics improve decision-making in optometry?

It helps interpret research data to support clinical choices and business strategies.

What are the benefits of knowing the truth, according to the lecture?

It makes individuals more educated, productive, efficient, and happier.

What is the relationship between science and morality?

Science relies on moral principles like honesty to ensure valid and reliable knowledge.

What is a null hypothesis (H₀)?

It assumes there is no statistically significant relationship between variables.

What is an alternative hypothesis (H₁)?

It assumes there is a statistically significant relationship between variables.

Why is it important to develop a hypothesis before collecting or analyzing data?

To avoid HARKing and reduce the risk of false positives.

What does HARKing stand for?

Hypothesizing After the Results are Known.

Give an example of anecdotal evidence.

"""My friend took peppermint oil and their vision improved."""

Why is anecdotal evidence not sufficient in scientific research?

It is not systematically collected or analyzed and is highly prone to bias.

List three characteristics of a good hypothesis.

Clear and specific, testable and measurable, based on existing knowledge.

What is a confounding variable?

A factor that distorts the observed relationship between independent and dependent variables.

What is the difference between observational and experimental studies?

Observational studies do not intervene, while experimental studies involve researcher-assigned interventions.

What are nominal data?

Categorical data with no intrinsic order, such as blood type or eye color.

What are ordinal data?

Categorical data with ranked order, like pain scale or cancer stage.

What are discrete data?

Countable numerical data, such as number of missed doses.

What are continuous data?

Measurable numerical data that can take any value within a range, like blood pressure.

What is the purpose of using controls in a study?

To isolate the effect of the independent variable and minimize bias.

What is blinding in research?

A method to prevent bias by ensuring participants and/or researchers do not know group assignments.

What is the primary goal of statistics in scientific research?

To collect, analyze, interpret, and present data to identify patterns and make informed decisions.

What is the difference between descriptive and inferential statistics?

Descriptive statistics summarize data; inferential statistics make predictions or generalizations about a population.

Define independent and dependent variables.

Independent variables are manipulated to observe effects; dependent variables are the outcomes measured.

What does a 95% confidence interval mean?

It means there's a 95% chance the interval contains the true population parameter if the study is repeated.

What is the empirical rule for normal distribution?

68% within 1 SD, 95% within 2 SD, 99.7% within 3 SD from the mean.

What type of distribution has two peaks?

Bimodal distribution.

What is the purpose of statistical tests in research?

To determine if observed differences are statistically significant or due to chance.

What transformation is useful for right-skewed data?

Log transformation.

What are the measures of central tendency?

Mean, median, and mode.

What does standard deviation represent?

The average distance of data points from the mean.

When should non-parametric tests be used?

When data doesn't meet the assumptions for parametric tests.

How does sample size affect statistical power?

Larger sample sizes increase statistical power.

What is the binomial distribution used for?

To model the number of successes in a fixed number of independent trials.

What is the placebo effect?

Improvement due to expectation rather than the treatment itself.

What does a probability of 1 represent?

An event is certain to occur (100% likelihood).

What is the Bayesian interpretation of probability?

Probability as a measure of belief updated with new information.

What is the Frequentist interpretation of probability?

Probability as the frequency of an event over many trials.

Define sensitivity in diagnostic testing.

The probability that a test correctly identifies someone with the disease (true positive rate).

Define specificity in diagnostic testing.

The probability that a test correctly identifies someone without the disease (true negative rate).

What is Positive Predictive Value (PPV)?

The probability a person actually has the disease if the test is positive.

What is Negative Predictive Value (NPV)?

The probability a person does not have the disease if the test is negative.

What is theoretical probability?

Probability based on models and assumptions without experiments (e.g., dice rolling).

What is empirical probability?

Probability calculated from actual data or experiments.

What is subjective probability?

A belief-based probability that is updated as new data becomes available (Bayesian).

What is the complement rule in probability?

P(not A) = 1 - P(A)

What does the multiplication rule state for independent events?

P(A and B) = P(A) × P(B)

What is conditional probability?

The probability of an event given that another event has already occurred.

What is a p-value?

The probability of observing your result or more extreme, assuming the null hypothesis is true.

What does it mean if p ≤ α?

Reject the null hypothesis; result is statistically significant.

What is the main difference between correlation and regression?

Correlation describes association, while regression models the effect of one variable on another.

What does a Pearson correlation coefficient (r) of +1 indicate?

A perfect positive linear relationship between two variables.

What is the formula for simple linear regression?

Y = a + bX

What does the slope (b) in a regression equation represent?

The expected change in Y for a 1-unit increase in X.

What does R² represent in a regression model?

The proportion of variance in the dependent variable explained by the independent variable.

When is Spearman’s correlation preferred over Pearson’s?

When the data is ordinal or not normally distributed.

Why does correlation not imply causation?

Because two variables can be associated due to a third factor or coincidence, not a causal link.

What is multicollinearity in multiple regression?

When independent variables are highly correlated with each other, affecting the model's accuracy.

Name a key assumption of linear regression.

Linearity between variables.

What kind of data is best visualized with a scatter plot?

Quantitative data showing the relationship between two variables.

What is homoscedasticity in regression?

Equal variance of residuals across all levels of the independent variable.

What does a residual plot help diagnose?

Violations of regression assumptions like linearity or equal variance.

When would you use a boxplot?

To compare data distributions, identify medians, quartiles, and outliers.

What does a line graph best visualize?

Trends over time or model-predicted values.

What does a histogram show?

The frequency distribution of data, useful for assessing normality and spread.

What does sensitivity measure in a diagnostic test?

The probability that a test correctly identifies someone with the disease (true positive rate).

What does specificity measure in a diagnostic test?

The probability that a test correctly identifies someone without the disease (true negative rate).

What is the formula for Positive Likelihood Ratio (LR+)?

Sensitivity / (1 - Specificity)

What LR+ value is considered strong evidence to rule in a disease?

Greater than 10

What is the formula for Negative Likelihood Ratio (LR-)?

(1 - Sensitivity) / Specificity

What LR- value is considered strong evidence to rule out a disease?

Less than 0.1

What does the area under an ROC curve (AUC) indicate?

The accuracy of a diagnostic test

What AUC value represents outstanding test accuracy?

0.9 or greater

Define Absolute Risk (AR).

The actual probability of an event occurring in a group over time.

Define Relative Risk (RR).

The risk in the treatment group divided by the risk in the control group.

What is the formula for Number Needed to Treat (NNT)?

1 / Absolute Risk Reduction (ARR)

What does Evidence-Based Practice (EBP) integrate?

Research evidence, clinical expertise, and patient preferences.

List the 5 steps of Evidence-Based Practice.

Ask, Acquire, Appraise, Apply, Assess

What does critical appraisal evaluate?

Validity, importance, and applicability of research findings.

Name strategies to reduce bias in a study.

Randomization, blinding, or statistical adjustments.