Stats Terms

Entire gorup of individuals that we want information about

Small gorup chosen from population we examine to gather informaiton about

People choose themselves (Web poll, call in poll)

Choosing people who are easist to readch

Each member of the population has an equal chance to be chosen

Choose every nTH itme (EX: Everyt 10th)

Population divided into groups. Randomly select clusters, measure all individuals in those clusters (ALL FROM SOME)

Population divided into groups. Take SRS from every group (SOME FROM ALL)

Entire population targeted is not included in design of sample

Individual cannot be selected or refused to cooperate

Individual's response influenced in some way by the interviewer

Looks backwards in time

Looks forewards in time

1.) CONTROL the experiment conditons for all treatment groups to assure all lurking variables do not bias the results
2.) Experimental units must be RANDOMLY assigned to treatements
3.) REPLICATION music be used to reduce chance of variations in results
4.) Placebo or control MUST be used as one of the treatements
5.) Medical experiments MUST be double blind

Indviduals you are studying in the experiment

Entire gorup of individuals that we want information about

Explanatory (Independent) variables (maybe have several) thought to influence response (outcome / dependent) variable studied. Combine specific values (levels) of each factor to form a treatment

Variable that is not a,ong explanatory/ response variables, but influence interpretation of the relationationship

Addition explanatory variable that affects response but that is not considered when exploring explanatory/ response relationship

Dummy Treatment

Experiment where neither patient not researcher knows who had which treatement

Experiement where only participants do not know which treatment they got assigned

Observed effect so large tthat it would rarly occur by chance

Compeltely random

Subjects paried accoridng to variables that affect response and then randomly assaigned treatment

Blocks of similar subjects are formed and then randomly assigned to treatment groups

Histograms, stem plots, split stem plots, time plots

Shape, center, spread, outliers

Symmetry, skewness, bimodal, unimodal, bell-shaped

Correlation does not imply causation

Middle Value

Measures spread or variability of the data.

Middle value of the smallest half fo data

Middle vlaue of largest half of data

Minimum
Q1
Median
Q3
Maximum

Mean and standard deviation to measure center and Spread

Median and IQR

Used to compare values from two different normal distributions

Displays relationship between two quantitative variables

1.) Form: Linear, curved, clusters
2.) Direction: Positive or negative association
3.) Strength: How close points lie to a simple form (Such a line)
4.) Outliers: Extreme Observations

Measures strength and direction of a linear relationship between two variables

Square of correlation coefficient Measures the predictive power of the equation

Observed y - Predicted y
Negative Residual: Prediction too high
Positive Residual = Prediction too low

Making predictions outside the range for which you have data

Correlation does not imply causation