Exam 1 Definitions Flashcards
Observational study
observes individuals, there is no treatment applied.
Experiment
imposes a treatment followed with observation
SRS
sample chosen randomly to avoid bias, uses the laws of probability
Multistage sample
stratify the population then randomly choose a sample from each strata
Stratified random sample
divide into strata by similarity of individuals then apply simple random selection to pick a sample from each strata
Block design
collect individuals in groups by similar characteristics ??
Matched pairs
a block design with two treatments for each individual
or an equally matched pair with random selection for treatment.
Voluntary response
Sample taken from people that chose themselves to respond
double-blind experiment
Neither facilitator nor subject know what treatment is being applied
factor
explanatory variable
confounding
lurking variable that confuses ??
bias
systematic errors in the way the sample represents the population
lurking variable
variable not in study, but has an effect
Informed consent
subjects give permission to collect and use data
convenience sampling
sample collected by taking members of population that are easiest to reach
sampling design
how samples are collected
inference
process of drawing conclusions about a population from a sample
undercoverage
groups in the population but not in the sampling process
nonresponse
the individual chosen can’t be contacted or refuses to participate
response bias
inaccurate responses due to dishonesty, embarrassment, confusion
wording bias
confusing or misleading questions
treatment
experimental condition applied to subjects, each treatment being a combination of values of the explanatory values
randomized comparative experiment
comparison of treatments where subjects are assigned randomly to the treatments
Simple random sample (SRS)
basic sampling design where every possible sample has the same chance to be chosen
Bias
systematic errors in the way the sample represents the population
Failure to use random sampling results in ___.
bias
What samples are prone to a large bias?
Voluntary response bias
Most national surveys are carried out how?
by phone using random digit sampling
A sample survey will gather information about a population without _______ during the process.
disturbing
Experiment
not observational, it imposes a treatment in order to observe the response
Purpose of an Experiment is …
to study whether the treatment causes a change in the response
Which kind of study is a terrible way to gauge the effect of a treatment?
Observational study
Variable are confounded when:
A. their effects on a response cannot be distinguished from each other.
B. lurking variables cause changes in the response
Treatment
combination of values of the explanatory variables (factors)
Experiment design involves
- control & randomization to combat bias
2. using enough subjects to reduce chance variation
simplest control
comparison of two treatments on same subject
Randomization should be used to produce …
2 similar treatment groups
Double blind experiment
uses a placebo to make lack of treatment realistic to control group
plus the mediator doesn’t know which group receives the treatment and which receives the placebo.
Blocks of similar individuals
can be used as a form of control, randomly assign individuals from blocks to treatment group
Matched pairs
Two similar individuals make a pair. Each subject in a pair receives one of the treatments.
All human ______ raise _____ questions
endeavors
ethical
When ___ is lacking it tends to make authors’ _____ look stronger than it should.
data
conclusions
The most complex ethical situation is when …
Treatments are imposed on people
IRB-institutional review board
reviews all planned studies in advance.
It is charged with protecting safety and well-being of subjects.
Informed consent is given
before subjects participate in an experiment
Individual data is kept ____.
confidential
Randomized comparative experiments are
only way to see true effects of new treatments
Clinical trials produce great benefits but …
- most benefits go to future patients
* risks are borne by subjects of trial
The _____of the subject must always _____ over that of science and society
interests
prevail
Data set
contains info on individuals
variables
describe some characteristic of an individual
categorical variable
characteristic that separates each individual into a specific category, like gender
quantitative variable
numerical value measuring some characteristic
Exploratory data analysis
graphs & numerical summaries to describe variables in a data set & relations among them
Distribution
the values a variable takes and how often it takes them
Pie charts and bar graphs are examples of
categorical variable distributions
Histograms and Stemplots
show quantitative variable distributions
Outliers
observations that lie outside the overall pattern of a distribution
A numerical summary of a distribution should
report the center and its spread
mean describes _______
Median describes _______
arithmetic average of the observations
midpoint of values
5 number summary consists of
median, 1st and 3rd quartiles, minimum, max
Boxplots
useful for comparing several distributions.
based on 5 number summary
variance and standard deviation measure
the spread about the mean
Any aspect of a distribution that is unaffected by the size of a population
Resistant measure
An example of a resistant measure:
Median and quartiles
Normal distributions show
symmetry & no outliers
4 steps to organize problem
- state problem
- plan
- solve
- conclude
Measures of central tendency
Mean
Median
Mode
Measures of dispersion
range
interquartile range
variance & standard deviation
coefficient of variation
Statistical calculations extrapolate from _______.
sample to population
Statistical calculations can do:
Statistical estimation
Statistical hypothesis testing
Statistical modeling
Statistical calculation cannot do:
Define a population
Randomly select a sample of subjects
Randomly select treatment groups
Measure a single variable in each subject
Statistical reasoning cannot help with these problems
Population you care about is more diverse than population sampled
You collect from a “convenience sample” rather than from a random sample
Measurements may be incorrect, and assays may measure wrong thing
You need to combine different measurements
Statistics is at the ______ of mathematics and _____.
interface
science
