Study Design, Bias, and Data

When you can generalize a population

Samples are random, from, and representative of the population (depending on study may need a control group)

Conditions to prove cause and effect

Compare at least 2 treatment groups, random assignment of treatment to subjects, replicates treatment to enough subjects, controls for potentially confounding variables.

Observational study

When observations are made in a study where no treatment is imposed

Retrospective

Examines past or current data

Prospective

Follows a sample into the future

Experimental study

When treatments are imposed on a sample and observed

This study can conclude cause and effect

An experimental study

This study cannot conclude cause and effect

An observational study

Subjects

The who or what in a sample

Treatments

The thing imposed on subjects

Control group

The base-line data used to compare outcomes among treatment groups

Statistic

A characteristic or value that summarizes a sample (sample mean, sample SD, sample median, etc)

Parameter

A number that summarizes data for a whole population (population mean, population SD, population median, etc)

Random

A process of slection in which each item of a set has an equal

Confounding variables

A variable that influences both the dependent and independent variable (A variable that can make two unrelated things appear to have a cause

Census

Collecting data from every

Simple random sample (SRS)

When everything individual in a population has equal odds of being sampled

Cluster random sample

Dividing a population into groups by proximity and randomly select entire groups to be sampled

Stratified random sample

Dividing a population into groups by similarity and randomly selecting subjects from each group to be sampled

Systematic random sampling

When a random starting point is selected, and subjects are chosen by a fixed periodic interval

What study design balances potentially confounding variables and establishes a cause and effect relationship?

A completely randomized design

What study design reduces variation by ensuring similarity between groups within a population before a randomized treatment is imposed?

A randomized block design

What study design occurs only when there are two treatment conditions, where subjects are grouped into pairs based on some blocking variable?

A matched pairs design

What study design is used to reduce bias where the subject OR researcher is unaware of which treatment has been assigned to what group

A single blind study design

What study design is used to reduce bias where the subject AND researcher is unaware of which treatment has been assigned to what group

A double blind study design

Factor (independent variable)

The variable manipulated to study it’s impact

Response variable (dependent variable)

The subject of change within an experiment

Levels

The different value or amount of a factor

What is the factor (independent variable) in this study “Does fertilizer increase the height of a bean plant?”

The amount of fertilizer

What is the response variable (dependent variable) in this study “Does fertilizer increase the height of a bean plant?”

The heights of the bean plants

What could be a confounding variable in this study “Does fertilizer increase the height of a bean plant?”

Different types of soils used for each bean plant

What is the treatment being imposed in this study “Does fertilizer increase the height of a bean plant?”

The amount of fertilizer used per day

What are the levels in this study “Does fertilizer increase the height of a bean plant?”

The specific amount of fertilizer used