STAT 110 Exam 1

Chapter 1-6 & 21 University of South Carolina - Odero

Variable

characteristic of an individual

categorical

places an individual into 1 or more categories

example of categorical variable

subject

ordinal

ordering, ranking of categorical variables

nominal

categories that cannot be ranked

example of ordinal variable

small, medium, large

example of nominal variable

math, English

quantitative variable

numerical values

discrete variable

counting

example of discrete variable

number of students in the class

continuous variable

data in a range

example of continuous variable

height

experimental study

applying treatments to subjects

explanatory variable

what is being manipulated

response variable

result of explanatory variable

observational study

observes individuals and measures variables of interest

are treatments applied in observational studies

no

can you determine cause and effect from an observational study

no

population

the entire collection from which we want data

example of a population

the entire United States

census

study that includes everyone in the population

sample

subset of the population that is actually surveyed

sampling method

way a sample is selected from the population

convenience sampling

sampling individuals that are easy to reach

voluntary sampling

sample chooses itself by responding to a general appeal

_ are the 2 biased methods of sampling

convenience and voluntary sampling

simple random sampling

method in which each individual has an equal chance of being selected

example of simple random sampling

drawing numbers out of a hat

random digits table

method of random sampling

stratified random sampling

taking an srs from strata to form a complete sample

strata

similar groups

cluster sampling

divide samples into clusters, pick a cluster, sample everyone in those clusters

cluster example

New York, South Carolina, and Arizona. sample is United States

systematic sampling

selecting individuals at a regular interval

p

proportion of the population

x

number of individuals in the sample who are in a specified category

n

sample size

p hat

proportion of sample who agree

bias

systematically favoring a certain outcome/individual

something is biased when…

the true population value is over/under estimated

ways to reduce bias

randomization, replication, control

randomization

using srs and blinding

replication

repeating the process over many subjects

control

having a placebo or group that does not receive treatment

voluntary response bias

people with strong opinions are more likely to participate

self-interest bias

people with a stake in outcome have an incentive to use biased methods

leading question bias

questions are worded in a way that prompt a response

non-response bias

certain proportion of the population do not respond

non responders

people who do not give an answer

sampling bias

some members of the populations are more likely to be included

example of sampling bias

convenience sampling

social acceptability bias

people are reluctant to be truthful bc they do not want answers to be reflected onto them

variability

how observations vary person to person within the sample

sampling variability

describes how sampling methods will vary when an experiment is repeated

sampling distribution

spread of values taken in sample

larger sample = _ variability

less

which is more accurate: large or small sample?

larger sample

population should be _ times larger than sample

20

parameters describe a

population

statistic describes

sample

statistical significant

difference in average response among treatment groups is large enough that it is likely not chance

confidence statements describe

probability the statement given is true

confidence level

what percentage of all possible samples satisfy the margin of error

margin of error

value that accounts for error

a larger margin of error means

there is more uncertainty

margin of error formula

1 / sqrt n

sample proportion ( p hat) formula

= count of successes / n (sample size)

what level of confidence should be used in a confidence statement

95%

sampling frame

list of individuals the sample draws from

mean (p)

average of all the samples proportion

standard deviation

how dispersed data is relevant to the mean

standard deviation formula

sqrt P (1-P) / n

shape

shows variation in sample

shape of this graph

skewed right

shape of this graph

normal curve

shape of this graph

skewed left

confidence interval formula

p-hat ± z (sqrt p-hat (1- p-hat) / n)

sampling errors

errors that occur during sampling

random error

deviation between statistic and parameter

frame error

not complete representation of population

undercoverage (sampling error)

excludes people via biased sample methods

incomplete sample (sampling error)

excludes certain people from the population

non sampling error

errors that occur aside from sampling

response error

subject gives a false answer

lurking variable

variable not studied has effect on the relationship among variables in the study

ex: study on effects of diet on blood pressure, but patient smokes

lurking variable

confounding variable

2 variables effect on response cannot be distinguished

ex: effect of kids iq on reading level, not considering socioeconomic status

confounding variable

experimental design

the way a study is set up

randomized comparative experiment

completely randomized design

randomized block

matched pairs

clinical trials

medical experiments

number of subjects in phase 1

20-80

determinants from phase 1

safety, safe dosage, side effects

number of subjects in phase 2

100-300

determinants from phase 2

is it effective, evaluate safety of drug

number of subjects in phase 3

1,000-3,000

determinants from phase 3

confirm effectiveness, side effects, compare to other drugs, determine safety