OD 626 Biostatistics Midterm

Morals

of or relating to principles of right and wrong in behavior

Ethics

a set of moral issues or aspects (such as rightness)

What is science based on?

The moral that Telling the truth is "good"/moral/ethical and telling a lie is "bad"/immoral/unethical

What is the purpose of scientific research?

to identify the truth

Statisitcs

a branch of mathematics used for analyzing and interpreting data to reveal patterns, relationships, and trends in data

What allows researchers to draw valid and reliable conclusions from empirical evidence?

Statistics

Empirical evidence

information gathered through direct observation, measurement, or experimentation

Statistics allows us to identify truth with a degree of

certainty

Biostatistics

statistics applied to biology and biomedical research, including epidemiology, clinical trials, and public health studies

statistics facilitates

-Informed decision-making

-Predicting outcomes

-Problem solving

Scientific method

a systematic investigation to identify truth

-7 steps

Observation

step 1

-Analyzing environments and considering cause and effect relationships

Questioning

step 2

-Formulation of a defined and testable question

Forming a Hypothesis

step 3

-a prediction that serves as the basis for experimentation. It is commonly expressed as an "if...then..." statement

Experimentation

step 4

-Experiments are designed and conducted to test hypotheses by controlling and manipulating variables

Data Analysis

step 5

-Experimental data is analyzed to see if is supports or refutes the hypothesis

Conclusion

step 6

-data analysis leads scientists to draw conclusions and accept, reject, or further test their hypothesis

Communication

step 7

-It is important to communicate findings so that others can review, learn, test is it is replicable, or form new ideas. Most commonly done through research papers and/or presentations.

What does the scientific method encourage?

Objectivity, reproducibility, and skepticism to promote reliability and accuracy.

Evidence based medicine

evidence is based on statistical testing. Interpreting and understanding science and healthcare. Good science is evidence based and that is why we do it.

Why is biostats important in Optometry and clinical practice?

Evidence based medicine is based on statistical testing and is good science. It allows us to make educated decisions with logical and statistical support to guide professional preference. Shows us what works for patients.

Hypothesis

a testable statement describing a relationship between variables

Null hypothesis

assumes there is not a statistically significant relationship between variables

Alternative hypothesis

assumes there is a statistically significant relationship between variables

What does a hypothesis provide for ethical research studies?

It provides direction for designing and executing ethical research studies.

When should a hypothesis be formed?

Should always be made before any research is performed

HARKing

hypothesizing after results are known

Anecdotes

individual stories or personal accounts that lack scientific or statistical rigor and are not systematically collected or analyzed. They are highly prone to bias.

What can anecdotal evidence help develop?

a hypothesis

What should a good hypothesis be?

-Clear and Specific

-Testable

-Measurable

-Based on Existing Knowledge

-Focused and Limited Scope

-Includes an Independent and Dependent Variable

-Predicts a Relationship or Outcome

-Ethical, Practical, and Feasible

Study Design

refers to the methodology for testing a hypothesis

What do proper study designs aim for?

Valid, reliable, and unbiased results

Bias

a systematic error in the design, conduct, or analysis of a study that leads to incorrect results

What does bias threaten?

The validity of the conclusion

Study bias

participants are not randomly selected or assigned

Recall bias

arises when participants remember past exposures differently based on outcome status

Observer Bias (Detection bias)

a researcher's expectations influence how data is collected/interpreted

Measurement bias (instrument bias)

results from different tools or inconsistent procedures

Response bias

when participants give inaccurate response (they lie)

Attrition bias (Loss to Follow-Up bias)

large differences in dropouts between groups

Confirmation bias

favoring data that supports pre-existing beliefs

Publication bias

positive or statistically significant studies are more likely to be published

Blinding

a research method to prevent bias by ensuring that participants and/or researchers do not know which group a subject is in

Single-blind

subjects do not know which group they are in

double-blind

subjects and researchers do not know which group subjects are in

Confounding variables

factors associated with the independent variable and the outcome, but are not part of the study design

What do confounding variables often distort?

The true relationship between variables

Why is an adequate sample size important in research?

It ensures results are reliable, reduces random error, and increases the likelihood of detecting true effects (reduces Type II error).

What is statistical power?

The probability of correctly rejecting a false null hypothesis (detecting a real effect). Typically, researchers aim for 80% power.

-avoiding a Type II error

-increases with larges sample size and effect size

What factors influence statistical power?

Sample size, effect size, significance level (alpha), and variability in the data.

Placebos

substances or treatments that have no active therapeutic effect.

What are placebos often used as?

controls

Placebo effect

a psychological or physiological improvement in symptoms despite receiving a placebo

-verified and validated effect

Controls

standardized conditions or groups used to minimize bias and isolate independent variable effects

Negative control

a group that receives no treatment or a placebo

Positive control

a group that receives a treatment with a known effect to ensure the experiment works

Observational study types

researchers observe and analyze natural patterns or behaviors. They do not intervene.

-Often used when intervention is prevented by ethical or practical constraints.

-Can identify associations but do not confidently establish causation

cohort study

a subtype of observational study

-Follow a group over time based on exposure status to assess outcomes

Case-control study

a subtype of observational study

-Starts with outcomes, then looks retrospectively as exposures

Cross-sectional study

a subtype of observational study

-Measures both exposure and outcome at a single point in time

-It cannot determine which came first: exposure or outcome

-Dot does not assess temporal sequence

Experimental study type

researcher assigns an exposure or intervention to study its effect.

-Allows for stronger inferences about causality

Randomized Controlled Trials

an example of experimental study types

-Participants are randomly assigned to an intervention or control group

-The gold standard for testing treatment effectiveness

-Uses placebo or control group

-Often includes blinding

Descriptive study type

observe and describe characteristics or events without manipulating variables. They answer the who, what, when, and where of a population or phenomenon.

-Case reports and case series

-Used to describe or generate hypotheses

-No statistical evidence

Which study type has the least statistical power?

Expert Opinion

Which study type as the most statistical power?

Meta Analysis

Categorical (Qualitative) data

consists of names or labels that are not numbers representing counts or measurements

-Nominal- no intrinsic order. ex. blood type, race

-Ordinal- ranked order. ex. pain scale, cancer stage

Numerical (Quantitative) Data

consist of numerical measurements or counts

-Discrete- countable, integer-based. ex. number of ocular migraines with aura

-Continuous- measurable across a range with fractions or decimals. ex. height, blood pressure

Why does data type matter?

The type determines appropriate descriptive statistics, type of statistical tests, and type of graphs used for visualization

What data classification is BMI?

continuous

What data classification is smoking status?

nominal

What data classification is disease severity score?

ordinal

What data classification is medication adherence (number of missed doses)?

discrete

Statistical tests

mathematical procedures that make inferences about data

-help determine whether difference in your data represent real, meaningful causes or if they are random and due solely to chance

Statistical tests are used to

-Compare groups (e.g. treatment vs. control

-test variable relationships

-determine significance of tends or outcomes in data

-Predict whether sample group assumptions should apply to a population

What do statistical tests rely on?

Assumptions and data types

-Only specific tests should be used per assumptions and data

Data collection

can be collected through a variety of methods:

-surveys, experiments, observational studies, existing datasets

Sample

a subset of the population used for testing

Variables

a characteristic or attribute that can possess difference values.

-can be measured, counted, or categorized

-can be qualitative or quantitative

quantitative

measurable quantities (e.g. height, test scores)

discrete

countable in whole numbers (e.g., number of students)

continuous

measurable in fractions or decimals (e.g., weight)

qualitative

categories or labels (e.g., gender, eye color)

Independent variable

variable that the researcher manipulates or categorizes to observe its effect.

-considered the causes, input, or explanatory variable

Dependent variable

variable measured or observed to assess the effect of the independent variable

-the outcome or result of the independent variable

Reliability

the consistency or repeatability of a measurement

Validity

whether the study or measurement truly assesses what it claims to

Randomization

random assignment of subjects/treatments/etc to reduce bias and confounding

effect sizee

a quantitative measure of the strength of a relationship or difference

-more informative than p-values alone

Confidence interval (CI)

a range of values likely to contain the trye population parameter

Confidence level (CL)

the probability that a CI contains the true population parameter if the study were repeated

e.g., Average IOP was 15 mmHg with a 95% confidence interval of [12, 18]

Generalizability (External Validity)

the extent to which findings apply to populations beyond the study sample

Internal Validity

the degree to which the study design ensures the observed effects are due to the experimental intervention and not other factors

Replication

the ability of a study to confirm its findings when repeated

What are the two main branches of statisitcs?

Descriptive and Inferential

Descriptive statistics

summarizes and describes data

-measures of central tendency, measures of spread, and graphical displays

measures of central tendency

mean, median, mode

measures of spread

range and varaince

range

the difference between the maximum and minimum values

variance

the average squared deviation from the mean

-quantifies how much individual values differ from the average

-measures data dispersion or variability

-can be directly related to independent variables

low variance

data points close to the mean