Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
201 Cards in this Set
- Front
- Back
- 3rd side (hint)
Accuracy
|
how close the measurement is to the true value. Declines as measured values approach the detection limit of the test
|
|
|
Reliability
|
consistency on repeat measurements
|
|
|
Precision
|
this relates to the limit of detection of the test. The detection limit is the lowest value that can be measured by a test.
|
|
|
Validity
|
whether the test measures what it purports to measure
|
|
|
Screening
|
2ndary prevention, we will target these individuals to identify them sooner before symptoms arise
|
|
|
Surveillance
|
Population level detection. An early warning system that we have a problem. Triggers investigation and probably action. Primary prevention in an indirect sense.
|
|
|
Sensitivity
|
a/a+c
|
|
|
Specificity
|
d/b+d
|
|
|
Positive predictive value
|
a/a+b
|
|
|
Negative predictive value
|
d/c+d
|
|
|
T/F Sensitivity and Specificity are inversely related
|
True
|
|
|
T/F In screening tests we’d rather have more false positives
|
True
|
|
|
In diagnostic tests we’d rather have more false_____
|
negatives
|
|
|
Receiver Operator Characteristics (ROC)
|
Systematic method to determine best single cutoff value when test yields continuous results. Plot sensitivity (true positive rate) on y axis against 1 - specificity (false positive rate)
|
|
|
What is the optimal position in the ROC and why?
|
The best position is the upper left hand corner because it will have the 100% Sensitivity and 0% False Positive Rate
|
|
|
As prevalence goes up what happens to positive predictive value and negative predictive value?
|
Positive predictive value goes up and negative predictive value goes down
|
|
|
Kappa (κ) Statistic
|
A useful measure of inter- or intra-observer agreement. Calculate agreement beyond chance alone. k > 75% Excellent agreement, k < 40% poor agreement
|
|
|
Cumulative Incidence
|
New incidence / person-years
|
|
|
Incidence Density
|
cases/ person-years
|
|
|
Incidence
|
New incidence/ Period of time
|
|
|
Prevalence
|
number of people with disease/ Population at risk at a point in time
|
|
|
Period Prevalence
|
number of people with disease over a period of time / Population at risk at mid-period
|
|
|
What are factors increasing both incidence and prevalence?
|
Greater case ascertainment. Enhanced diagnostic methods, and More liberal criteria in disease definition.
|
|
|
Factors increasing prevalence only:
|
Improved (non-curative treatment), Out-migration of healthy people from the population, In-migration of people with disease into the population
|
|
|
What was Last’s description of epidemiology?
|
studying differences, find out why, and it will be applied to help alleviate health problems.
|
|
|
Miasmatic concept of health
|
nasty airs that were invisible that would settle down in low lying areas, causing disease
|
|
|
“shoe leather” epidemiology
|
get out of the office and see what’s actually going on
|
|
|
What are the three factors that contribute to disease in a host?
|
Agent, environment, vector
|
|
|
Who was the first person to receive an Abiocor transplant and was this cost effective?
|
Robert tool. No because it didn’t improve his quality of life because he stayed in the hospital.
|
|
|
Cost-benefit analysis
|
Measure cost and savings.
|
|
|
Cost-effectiveness analysis
|
Measure cost for a specified outcome
|
|
|
Cost-utility analysis
|
Use QALY or DALY’s as measures of outcome
|
|
|
Primary Prevention
|
stop disease from happening in the first place. Health promotion, prevent onset of disease, always the preferred method
|
|
|
Secondary Prevention
|
identify disease early to improve outcome
|
|
|
Tertiary Prevention
|
do nothing to alter full manifestation of disease or injury, attempt to reduce long-term effects.
|
|
|
Saal brothers, what did they “invent”
|
They invented a device that purportedly reduced lower back pain. It wasn’t really that good.
|
|
|
Cochrane Controlled Trials Register (CCTR)
|
Bibliographic database of controlled trials from systematic search of journals and other sources. Published or underway. Searchable
|
|
|
Death certificates governed by state statutes
How is reporting handled? |
variations exist in reporting requirements and specific terminology.
|
|
|
Immediate (or Principle) Cause
|
Final disease, injury, or complication resulting in death.
|
|
|
Underlying (or Antecedent or Intermediate) Cause
|
the disease or injury that initiated the chain of events that led directly and inevitably to death.
|
|
|
Underlying (or Contributory) Causes
|
The condition present before and leading to the intermediate or immediate cause of death.
|
|
|
Mechanism of Death
|
How death occurred, NOT a disease or injury.
|
|
|
Neonatal maternal ratio
|
Deaths within 4 weeks postnatal
|
|
|
Perinatal mortality ratio
|
Deaths 28 weeks gestation to 1 week postnatal
|
|
|
Direct Standardization
|
Start with death rates in Town A and standardize to Town B’s age
Multiply each stratum’s death rate in Town A by the number of people in that stratum in Town B. We get the death rate given Town A’s death rates but Town B’s age structure |
|
|
Homer paradox
|
Disappearance (reversal) of an observed difference when data is stratified and standardized
|
|
|
Indirect Standardization
|
We can also perform the exact same calculation the other way around
Start with death rates from the standard population and multiple by age structure of study population |
|
|
Mortality Country Group I
|
developing countries with high mortality
|
|
|
Mortality Country Group II
|
developing countries with low mortality
|
|
|
Mortality Country Group III
|
developed / industrialised countries
|
|
|
Survival Analysis
|
Commonly used technique in medical statistics
Take a group of people and follow-up them over a period time We can do this from birth to death used fixed time intervals |
|
|
Survival Analysis
When do we use a Wilcoxon Test? |
If mortality changes over time
|
|
|
Survival Analysis
When do we use a log rank test? |
If mortality is constant
|
|
|
Survival Analysis
What is a regression curve of the survival analysis called? |
Cox or proportional hazards model
|
|
|
Ecological Fallacy
|
Error that we make when we look at group data and two sets of group data, we don’t know any information on individuals, but we make assumptions about the individuals based on the population level information.
|
|
|
Causation
|
Causation refers to a relationship in which one condition precedes and must be present in order for another outcome to occur
|
|
|
Effect Modifier
|
A third variable which is part of the causal chain and which influences the frequency of disease
|
|
|
Inductive reasoning:
|
going from specific observations to generate a general principle.
|
|
|
Deductive reasoning:
|
applying general principles to a specific situation.
|
|
|
Confounder
|
A third variable that is independently associated with both the exposure and the outcome
|
|
|
Bias
|
A systematic error involving one group in the study
|
|
|
Disease Cluster
|
“An unusual number, real or perceived, of health events (for example, reports of cancer) grouped together in time and location." - CDC
|
|
|
Induction period:
|
period of time between appearance of causal action and onset of disease
|
|
|
Latency:
|
period of time between disease occurrence and detection
|
|
|
Hill’s Criteria
|
Temporal relationship
Strength of the association Dose-response relationship Replication of the findings Biological plausibility Consideration of alternate explanations Cessation of exposure Consistency with other knowledge Specificity of the association |
|
|
Possible
|
> 0% likelihood
|
|
|
Probable
|
> 50% likelihood
|
|
|
Descriptive Epidemiology
|
Describes general characteristics of disease
Comparisons to exposure often implicit or indirect |
|
|
Analytic Epidemiology
|
Comparison of disease to exposure is direct and explicit
Case-control, cohort and interventional studies |
|
|
Case report =
|
one case of the disease
Careful description of disease and circumstances May arise from routine surveillance |
|
|
Case series =
|
more than one case of the same disease
Careful description of disease and circumstances May arise from routine surveillance |
|
|
Correlational Studies
|
Compare disease frequency in relation to another factor at a population level
May compare disease rates based on geography, time, occupation, etc. Often called ‘ecological studies’ as in Gordis |
|
|
Cross-Sectional Studies
|
Measure number of cases of disease among individuals at one point in time (prevalent cases) in a defined population
Often called prevalence study Surveys which frequently use interviews or questionnaires |
|
|
Prevalence study Surveys
|
Cross-sectional studies
|
|
|
Correlation vs cross-sectional in terms of data
|
Correlation = aggregate data
Cross-sectional = individual data |
|
|
NHANES
|
Sample of 5,000 people over 12 months continuously since 1999
2 parts: Home interview Health exam |
|
|
NHANES findings
|
Growth charts
Blood lead and gasoline/solder Prevalence of obesity Undiagnosed type 2 diabetes |
|
|
T/F A key difference between cross-sectional studies and correlational studies is that we have individual data in cross-sectional studies
|
True. We have individual data in cross-sectional studies
|
|
|
Ecological or Descriptive Epidemiology
|
Describes general characteristics of disease
Comparisons to exposure often implicit or indirect |
|
|
Analytical Epidemiology
|
Comparison of disease to exposure is more direct and explicit
Two types: 1. Observational 2. Interventional |
|
|
The Framingham Study was this kind of study:
|
Prospective Cohort
|
|
|
Cohort Study
|
Individuals are studied who are disease free
Establish exposure status Follow forward in time to determine disease status May be either prospective or retrospective |
|
|
For a cohort study, the relative risk:
|
Incidence of disease in exposed/ Incidence of disease in unexposed
|
|
|
Prospective Cohort Study
|
Subjects have not developed disease yet when you perform the study
Must follow them into the future to see who gets disease |
|
|
The Framingham Heart Study
|
Sample of 5,127 adult residents in Framingham, Massachusetts taken in 1948.
Standardized biennial cardiovascular examination Daily surveillance of hospital admissions, death information and information from physicians and other sources outside the clinic. |
|
|
Retrospective Cohort Study
|
Subjects have developed disease when you start the study
You use information on exposure status when they were disease free and follow them forward in time (but prior to the start of the study) to see who gets disease |
|
|
Sources of Cohort
|
Can use one large general cohort that contains both exposed and unexposed and compare disease within cohort
Can use a cohort with a very high level of exposure and compare disease either to another unexposed cohort or general population The unexposed and exposed people should resemble each other in every way other than having the exposure |
|
|
What is the hierarchy of preference of studies
|
Clinical Trial
Prospective cohort study Retrospective Case- control |
|
|
Cohort Studies – Disadvantages?
|
Costly
Retrospective – you need to rely on data that is old and it is not designed for your study Logistical – expensive and time consuming. |
|
|
Case-Control Studies
|
Subjects are selected based on having disease under study (cases) or not having disease under study (controls)
Look back in time to see if there are any differences in exposure between the groups Sometimes called retrospective studies |
|
|
What is another name for Case-Control Studies?
|
Retrospective studies
|
|
|
Criteria for cases should be _____.
|
Exclusive
|
|
|
What are the characteristics of a control?
|
They must resemble cases in every way other than developing the disease under study.
Apply same exclusion criteria as cases Appropriate selection of controls is critical step in minimizing bias and confounding |
|
|
T/F Controls are required to be healthy
|
False
|
|
|
Matching
|
selecting the controls so that they are similar to the cases in certain characteristics
Group Individual or matched pair |
|
|
Odds Ratio
|
Calculate odds that someone WILL have exposure relative to NOT having exposure
Express as a ratio of those with disease (cases) compared to those without disease (controls) |
|
|
Nested Case-Control Study
|
A study within a previous study
|
|
|
Synonyms: Correlational study
|
Ecological study
|
|
|
Synonyms: Cross-sectional study
|
Prevalence survey
|
|
|
Synonyms: Case-control study
|
Case-comparison study, case-referent study
|
|
|
Synonyms: Cohort study
|
Longitudinal study, follow-up study, incidence study
|
|
|
Synonyms: Prospective cohort study
|
Concurrent study
|
|
|
Synonyms: Retrospective cohort study
|
Historical cohort study
|
|
|
Synonyms: Clinical trial, intervention study
|
Experimental study, therapeutic trial, randomized [blinded] controlled trial
|
|
|
Chronological Synonyms: Case-control study
|
Retrospective study
|
|
|
Chronological Synonyms: Cohort study
|
Prospective study
|
|
|
Chronological Synonyms: Prospective cohort study
|
Concurrent prospective study
|
|
|
Chronological Synonyms: Retrospective cohort study
|
Non-concurrent study
|
|
|
What are the two major categories of Analytical studies?
|
Observational: Case-control and Cohort
Interventional: Clinical |
|
|
Analytical Studies
Observational. Definition |
Investigator is passive and does not influence exposure
Case-control and cohort studies |
|
|
Analytical Studies
Interventional. Definition |
Investigators allocate exposure and follow-up for disease
Clinical trials |
|
|
What is a clinical trial?
|
A planned experiment of an intervention in humans to investigate:
Efficacy Safety |
|
|
What makes interventional studies the "gold standard"
|
Randomization and blinding make this the gold standard
|
|
|
Volunteer bias
|
health conscious, likely to be compliant to treatment. The treatment is considered the correct solution for the people under study.
|
|
|
Intervention Study
|
Like a cohort study, want a homogenous group without the outcome under study
Clearly defined with explicit exclusion criteria Must agree to participate Volunteer bias |
|
|
Reasons for stopping interventional studies.
|
May stop early:
if clear benefit If clear harm Need to develop guidelines for early termination before study is started Independent data monitoring group |
|
|
Phases of a drug trial
|
Preclinical Testing
Phase I – Safety and Pharmacology Phase II – Pilot Efficacy Phase III – Extensive Clinical Trial Phase IV – Long Term Effects Post-marketing surveillance |
|
|
Drug Trial Phase I
|
Safety and Pharmacology
|
|
|
Drug Trial Phase II
|
Pilot Efficacy
|
|
|
Drug Trial Phase III
|
Extensive Clinical Trial
|
|
|
Drug Trial Phase IV
|
Long Term Effects
|
|
|
Pre-clinical testing
|
~3.5 years
Animal testing File Investigational New Drug Application (IND) with Food and Drug Administration (FDA) |
|
|
Phase I – Safety and Pharmacology
|
~ 1 year
First introduction into humans Usually small number (<100) of healthy volunteers Goal is determine safety and mode of action Study pharmacokinetics, route of administration |
|
|
Phase II – Pilot Efficacy
|
~ 2 years
200-500 volunteers with disease Usually randomized to new drug or existing treatment Demonstrate preliminary safety and efficacy |
|
|
Phase III – Extensive Clinical Trial
|
~ 3 years
1000 – 3000 volunteers with disease, multicenter More complete assessment of safety with longer use and verify efficacy File New Drug Application (NDA) with FDA ~ 2.5 years |
|
|
Phase IV – Long Term Effects
|
After drug registration with FDA
May be required by FDA > 3000 participants Explore longer term effects, specific adverse outcomes More reflective of routine use |
|
|
Postmarketing surveillance
|
Based on reports by health care providers
HealthWatch Not systematic studies |
|
|
Crossover design
|
Study that switches patients from treatment A to B and from B to A allowing the patients to be their own control
|
|
|
What are the advantages and disadvantages of Crossover design?
|
Advantages
Allows each subject to start act as an own control. Most response. Need the fewest number of people because Disadvantage? Effect of B may really be a effect of A |
|
|
Crossover Design
How can we reduce the chance that the the Effect of B is really an effect of A? |
we use washover period - no treatment in the middle between the A and B to limit the disadvantage
|
|
|
Factorial design
|
We can explore synergistic and antagonistic effects
Allows us to study more than one treatment |
|
|
Who is responsible for the creation of Blinding?
|
Benjamin Franklin
|
|
|
Who did Franklin Debunk in the Franklin Commission?
|
Antoine Mesmer – mesmerize patients in order to help them out… not really the case
Franklin Commision. He identified it that it worked essentially because the patients wanted for it to work. The patients were blindfolded and the mesmerizing effect was gone |
|
|
Blinding is the best method to reduce:
|
Chance
Bias Confounding |
|
|
Blinding (aka Masking). Three types. what does it require?
|
Single patient
Double patient and doctor Triple In order to blind people, you must have a placebo or sham for comparison |
|
|
Placebo
|
an intervention designed to simulate medical therapy but not believed by the clinician or investigator to be a specific therapy for the target condition.
|
|
|
Nocebo Effect
|
Patients get worse no matter what
|
|
|
What did Haygarth disprove? What did he invent?
|
the Perkins Patent Tractor, 1799, a metal rod could "cure" people. He used a wooden rod which looked like the tractor to show that people still believed that they were being "cured". Disproved the tractor's efficacy.
the sham treatment |
|
|
Methods to deal with non-compliance
|
Run-in or wash-out period
All participants receive treatment OR placebo before randomization Those who do not comply in pre-randomization phase not included in study Monitor compliance Pill counts Pick high risk patients Frequent contact with participants Financial incentives Testing for presence of drug |
|
|
Who developed intention to treat analysis and what is it?
|
Intention to treat analysis. Once randomized, always analyzed.”
All patients in each group are followed-up and analyzed according to what they were assigned even if they didn’t complete or comply with assigned therapy. Sir Richard Peto |
|
|
Why use Intention to Treat Analysis?
|
1.Guards against conscious or unconscious attempts to influence the results of the study by excluding odd outcomes.
2. Preserves the baseline comparability between treatment groups achieved by randomization. Reflects the way treatments will perform in the population by ignoring adherence when the data are analyzed. |
|
|
Efficacy
|
works under Phases I-III of the drug trial
|
|
|
Effectiveness
|
works in the real world Phase IV
|
|
|
Efficiency
|
cost versus benefit. Number needed to treat helps us with this component.
|
|
|
The Three E’s of an Intervention
|
Efficacy ≡ Does it work under the ideal conditions of a RCT?
(Phases I – III) Effectiveness ≡ Does it work in the real world? (Phase IV) Efficiency ≡ cost versus benefit. Number needed to treat helps us with this component. |
|
|
Number Needed to Treat
|
NNT is the number of people you’d need to treat for the prescribed period of time in order to prevent one episode of disease.
|
|
|
NNT - Example
Adverse outcomes in treated group = 4% (0.04) Adverse outcomes in placebo group = 10% (0.10) calculate it. |
Adverse outcomes in treated group = 4% (0.04)
Adverse outcomes in placebo group = 10% (0.10) Absolute risk reduction is 0.10 – 0.04 = 0.06 1/0.06 = 17 Therefore you would need to treat 17 people to prevent one case of disease *remember that the NNH (number needed to harm) is calculated in a similar fashion. |
|
|
Meta-analysis
what does box size mean? How does CI change when sample size increases? What do the diamonds represent? |
Size box is bigger sample size.
So when sample size Increases we should expect the confidence interval to go down. Combined odds ratio here based off of the meta-analysis |
|
|
Steps in Performing Meta-Analysis
|
Identify studies
Evaluate studies Abstract data Perform statistical analysis |
|
|
Publication bias
|
“Tendency of editors (and authors) to publish articles containing positive findings, especially ‘new’ results, in contrast to reports that do not yield ‘significant’ associations.” (Last)
|
|
|
What indicates a publication bias as seen on a funnel plot?
|
Using meta-analysis funnel plot we see that there is no symmetry.
|
|
|
How do we reduce publication bias when doing a meta-analysis?
|
Researchers tend to choose positive results to report... Must balance these out with negative results - or what they would show
|
|
|
Metaanalysis
Fixed effects model |
conclusions derived in the meta-analysis are valid only for the studies included
|
|
|
Metaanalysis
Random effects model |
assumes that the studies included in the meta-analysis belong to a random sample of a universe of such studies
|
|
|
Metaanalysis
Test of homogeneity what statistic is used? What does homogeneity mean? What test is this similar to? |
Q statistic
Null hypothesis that all studies are homogenous (effect size equal) Analogous to Chi square test |
|
|
Meta-analysis?
Sensitivity Analysis compare studies based on ____ compare results based on _____ |
Compare studies based on quality
Compare results based on model choice |
|
|
Factors
|
are conditions that can possibly affect the outcome, might be controlled by an investigator
|
|
|
Outcome
|
Endpoints or Responses
|
|
|
What kind of study is shown below? Example: Assign one of several dosages (placebo, 10 mg, 20 mg) to each subject.
|
Intervention/experimental study
|
|
|
Name some variable types
|
Binary, nominal, ordinal, continuous, ratio
|
|
|
Binary variable
|
also known as a dichotomous variable one example is gender or the presence or absence of somethings
|
|
|
Nominal scale
|
naming. Is simply a classification scheme
|
|
|
Ordinal variable
|
rank. scale such as from 1-10, or strongly agree -> strongly disagree unlike continuous variables, arithmetic does not make sense
|
|
|
Continuous variable
|
variable that exists over an interval with EQUAL INTERVALS (IQ scores, centrigrade) Order is implied numerically and the arithmetic does makes sense
|
|
|
Ratio
|
continuous scale + a true zero (for instance distance or Kelvin)
|
|
|
Ranked variables
|
are ordinal,
|
|
|
Discrete Variable
|
one for which there is a finite number of potential values which the variable can assume between any two points on the scale. (1,2, 3, 4, etc..) It’s a count!!
|
|
|
Continuous variable
|
one which theoretically can assume an infinite number of values between any two points on the scale. (1.008, 1.009, 1.01, etc..)
|
|
|
Three basic ways to summarize data
|
Distributions, central tendency, and dispersion
|
|
|
Distributions and give an example
|
The distribution of a variable is a graphical or mathematical description of a variable that captures “everything” Example: Range
|
|
|
Real Frequency Distribution. Describe the y and x axis
|
y-axis (vertical) is usually describing the frequency with which the corresponding value of the x-axis (horizontal) are expected to be found
|
|
|
What is the purpose of a normal distribution? What assumption are we making?
|
The normal distribution can help us make inferences about the real distributions, which we assume is a sample from a population with a normal distribution (Chapter 10)
|
|
|
Quantitative variable
|
things with numbers (temperature)
|
|
|
Qualitative
|
descriptive (such as color)
|
|
|
Latent variable
|
variables that are not directly observable (such as motivation and intelligence)
|
|
|
What are the three measures of central tendency?
|
Mean - average Mode - most frequent statistic Median - Middle number
|
|
|
What are the different measures of dispersion?
|
percentiles, mean deviation, variance, and standard deviation
|
|
|
Choosing a statistical test. Nominal
|
Chi square
|
|
|
Choosing a statistical test. Nominal (dichotomous)
|
t-test
|
|
|
Choosing a statistical test Nominal (multiple variables)
|
ANOVA
|
|
|
Parameters
|
Characteristics of populations. (two examples are the mean and standard deviation)
|
|
|
Dependent variables
|
also called endpoints. This changes according to changes in factors/independent variables
|
|
|
Independent variable
|
also called a factor. The dependent variable changes based off of this.
|
|
|
T/F It is reasonable to obtain the entire population data to obtain parameters
|
False. It is not practical to get population data!!! So, parameters are usually unknown and we often need to estimate them from samples
|
|
|
Statistics
|
These are values that we calculate from the samples So, a mean () of a population is a parameter. AND A mean (x) of a sample is “statistic.”
|
|
|
What is the purpose of calculating statistics?
|
We calculate statistics from samples to estimate the parameter of populations.
|
|
|
Mode
|
Measure of central tendency that describes: Most frequent score in the distribution
|
|
|
Median. describe it, where is it's use appropriate
|
Measure of central tendency. The point at which 50% of the scores fall below (when listed in numerical order). Appropriate to use when the data are ordinal or ratio.
|
|
|
Mean
|
Measure of central tendency The arithmetic average of scores in a distribution. Mean is the “average”
|
|
|
What determines the direction of skew? Describe a left and right skew and what this means in terms of the mean and median
|
Direction of tail determines skewness Negative: (left) mean < median Positive: (right) mean > median
|
in terms of mean median and mode
Right/+ skew: mean > median > mode Left/ - skew: mean < median < mode |
|
T/F Continuous variables assume a finite number of intermediate values
|
False - It assumes a infinite number of intermediate values
|
|
|
Overall Range. Description how do we obtain it
|
It is a measure of dispersion or the "spread of the data" Overall Range = Biggest value - smallest value Range is susceptible to the effect of outliers
|
|
|
Sample variance, s^2
|
Sample variance, s2 The average squared difference between the values of a variable and the sample mean.
|
|
|
Standard Deviation
|
An estimate of the average variability (spread) of a set of data measured in the same units of measurement as the original data. It is the square root of the variance.
|
|
|
Estimate
|
Estimates represent our best guesses about the value of a parameter.
|
|
|
Sampling error. What is it? Does a individual sample accurately estimate the population mean?
|
Sampling error = discrepancy Each sample has its own mean, which does not provide a perfectly accurate representation of it population. No Individual sample means tend to under or overestimate the population mean.
|
|
|
Compute the 95% Confidence Interval for the mean based on the following information: The sample mean = 2994 The sample std dev = 800 The standard error is 800 / 78 = 90.5 What does this mean?
|
~ 95% CI for the mean is 2994+2(90.5) 2994+/-181 We are ~ 95% sure that the actual mean of the larger population is between
|
|
|
Do we want a small or large Standard Error? Why?
|
We want a small SE = More confidence in our mean.
|
|
|
Standard Error (general idea, relationship between sample mean and population mean, how to calculate).
|
For each individual sample, you can measure the error (distance) between the sample mean and the population. The standard error provides a way to identify the “average” or standard distance between the sample mean and population mean. The standard error is calculated by dividing the sample deviation by the square root of the sample size.
|
|
|
What is typically used mean or median?
|
Median, because it is not as affected by outliers as the mean. It is more robust
|
|
|
What portion of the population (normal distribution) falls within one standard deviation? How about 2?
|
1 SD = 68.2% of the population 2 SD = 95.4% of the population
|
|