Biostats Quiz 2

Info so far

What is Truth

What is Truth

Something is objectively true when it corresponds to nature and character of reality

Two Types of Truth

Objective and Subjective Truth

Objective Truth

It is true of the object itself

Subjective Truth

relative truth; it is true based on the subject’s preferences

Ways of knowing Truth

Authority, Reasoning, Intuition, Scientific Method

Authority

We consider something to be true based on long-held traditions or because some person of distinction says its true

Reasoning

We consider something to be true because it makes sense to us

Intuition

Immediate knowledge of truth of a proposition (truth of matter is self-evident)

Scientific Method

Combination of all other ways of knowing to form a hypothesis about what is true, but then relies on the collection of information from reality to see if the hypothesis corresponds to reality

Issue with Authority

Appeal to authority is logical fallacy, can be unreliable

Issue with Reasoning

One’s reasoning can often fail due to failure to logic, bias, and/or unsound assumptions

Issue with Scientific Method

Subject to quality of data, measurements, and human error/bias

"Science Never Says Anything, ___ Do"

Flashcard:

Term: Science Never Says Anything, Scientists Do

Scientists

Statistics can help distinguish . . .

most important or essential part, or help deal with complexity

Sample size (n)

The total number of observations (typically number of people from whom you collect data)

Sampling error

How different the sample is from the population

Sampling bias

the researcher does not select the sample randomly, but with a bias toward a particular characteristic

Statistics help deal with variation

stats such as standard deviation can help understand variation in data among groups

Accuracy

The closeness of a measure value to a standard value (correctness of a measure)

Validity

The extent to which a measurement measures what it is suppose to measure

Precision

the closeness of two or more measurements to one another

Reliability

the repeatability of a measurement

Purpose of Hypothesis

looking for differences or cause/effect

Confounders or covariants

variables that impact variable of interest and effectors (IV)

Independent Variable

Effector (cause)

Confounder

example: age, impacts independent variable but can be controlled for

Covariate

related to dependent variable, influences outcome

Observational Study Designs

researchers only observe and don’t intervene, takes less time and less costly; although cannot establish cause/effect and full of confounds

Case Study

an account of some clinical situations, typically rare, new, or unusual

Advantages of Case Study

opportunity for intensive study, stimulate new research

Disadvantage of case study

cannot be replicated, Hawthorne effect (observation behavior changes), researcher bias, limited to descriptive stats

Cross Sectional Study

investigates an issue by collecting data at a timepoint in the lives of study participants, usually 2 or more groups based on IV, used to describe or compare the dependent variables among individuals or between groups

Advantage of Cross-Sectional Study

relatively quick and inexpensive, examine “links” or “relationships” between variables

Disadvantages to Cross-Sectional Study

susceptible to confounding variables, cannot establish cause-effect relationship

Case-Control Study

Groups are selected to be as similar as possible with regard to characteristics and past exposure to risk factors (presumed cause) is ascertained from interviews, medical records, labs, etc. and conclusions drawn regarding a link between exposure and outcome

Advantages to Case-Control Study

takes less time, less costly, examine links between risk factors and outcomes

Disadvantages of Case Control Study

cannot establish cause-effect, susceptible to confounding variables and recall bias

Good indication of Case-Control

If there is a group with the condition and one without gives a clue

Prospective Cohort Study

the specific cohort is divided into presume risk factors and followed over time and notes the number of incidents; they don’t have the condition to start and they track development in each group (i.e. lean v. obese nurses at Harvard developing cancer)

Advantage of Cohort Design

large sample size, good design for rare exposure, examine links between risk factors and outcomes, less susceptible to bias

Disadvantage to Cohort Design

takes a long time, cannot establish cause-effect, susceptible to confounders, may have reduce generalizability

Retrospective Cohort Study Design

a cohort may have been exposed to a risk factor (exposed v. not exposed) and compare incidents of death/disease

Ecological Studies

No individual level data is reported only group level data and compared across IV’s

Ecological Fallacy

Applying group level data to individuals

Interventional Study Design

the researchers intervene or “treat” the individuals in the study in order to determine a cause-effect relation

Randomized Control Trials (RCT)

sample is divided into at least 2 groups (control and treatment) with random assignment and change in the outcome variables in measured such that conclusions can be drawn regarding the effectiveness of the treatment drawn

Randomized Cross-Over Trial

more powerful that RCT where sample is divided into at least two groups and treatment is given to treatment group, then a “wash out” period occurs before the control is then given the treatment and treatment is now control

Advantage of Randomized Cross-Over Trials

smaller sample size, less confounders

“Blind”

Refers to the fact that the study participants don’t know if they are in the treatment or control group

“Double Blind”

refers to the fact that neither participants of researchers know the assignment of participants so that the researchers knowing can’t change treatment of the participants or how they analyze the data

Inclusion Criteria

attributes of the study participants that are essential for their selection

Exclusion criteria

this minimizes the influences of specific confounding variable

Scales of Data

Categorical & Metric

Types of Categorical Data Scale

Nominal (name) or ordinal (order/ hierarchy)

Types of Metric (measured) Data Scale

Interval (continuous) and Ratio (continuous or discrete)

Ratio Scale of Data

Zero represents the complete absence of attribute measure, values can be compared as ratio or percentage (i.e. age, income, time, speed, height)

Interval as Data Scale

Equal differences between numbers represent equal differences in the attribute being measured (i.e. year, temperature)

Ordinal as Data Scale

Numbers represent rank order of the variable being measured (Pain, SES)

Nominal as Data Scale

Numbers distinguish among the categories; numbers do not represent quantity or degree, assignment of numbers to groups is arbitrary

Variables

A trait of factor that can exist in differing amounts or to differing extents within and between individuals

Randomness

an unpredictable event; an event without a proximal cause

Two Ways to Depict Information

Tables and Figures (Graphs)

All key ____ is displayed in tables and/or figures

data

Each figure to relay a _____ of the story

point/part

Tables Should . . .

only include data referenced in text, have a clear explanation of the data readers are looking at

Types of Figures

Bar graph, histogram, box plot (box and whiskers), Scatterplot

Bar Graph

Order of independent variable doesn’t matter

Histogram

order of independent variable matters (ordinal or metric)

Box Plot (Box and Whiskers)

most commonly and appropriately used with metric data

Scatterplot

both independent and dependent variables are continuous