STATS REVIEW FACTS Flashcards
Critical Z-score for confidence level of 90%
1.65
Critical Z-score for confidence level of 95%
1.96
Critical Z-score for confidence level of 99%
2.58
How do you determine a confidence interval?
Mean +/- z-score(SE)
How do you determine standard error?
Standard deviation/ square root of the sample size (N)
What is special about a z-distribution?
It has an idealized mean of 0 and a standard deviation of 1
Type 1 Error
Null is True, but rejected the null
*Said there was a difference when really there wasn’t one!
Type 2 Error
Null is false, but fail to reject the null
*Said there was no difference but there actually was!!!
What are the 3 main parts of the classical experiment?
- Independent and dependent variables
- Pre-testing & post-testing
- Control & Experimental Groups
Phase 1 of clinical trial
Safety & Side Effects
Phase 2 of clinical trial
Ideal dosing
Phase 3 of clinical trial
Experimental vs. Control
Phase 4 of clinical trial
Continued evaluation of FDA-approved therapy
What is the hawthorne effect?
Research attention affects behavior
What does double blind mean?
Both the researcher and respondents don’t know which person in control or experimental
External Invalidity
Applicability to the real world
Internal Invalidity
Anything other than the stimulus that influences results
- history
- Maturation
- Testing
- Instrumentation
Simple Random Sample
Need a list/Sample frame
Assign number
Select a random number
Systematic Random Sample
- Determine number of elements needed
- Divide population by sample number desired = “k”
- List and number elements
- RANDOMLY select start point
- Select every K-th element within groups
Stratified
Random sample from sub populations
Cluster
Cluster = groups of elements Multistage = listing and sampling
**increases sampling error potential - two samples double the error opportunity!
Two ways to reduce sampling error
- Increase sample size
- Increase homogeneity
What percent of the population falls with-in 1 standard deviation of the mean?
68%
What percent of the population falls with-in 2 standard deviations of the mean?
95%
What percent of the population falls with-in 3 standard deviations of the mean?
99.7%
Prevalence
ppl w disease/ # of ppl at risk
Incidence
new ppl w disease/ 3 of ppl at risk (make sure to subtract ppl who previously contracted the disease)
Attack Rate
new cases/ # of ppl exposed
Secondary Attack Rate
new cases / (# exposed - # of primary cases)
Sensitivity
TP/ (TP + FN)
Specificity
TN/ (TN + FP)
Positive Predictive Value
TP/ (TP + FP)
Negative Predictive Value
TN/ (TN + FN)
CER
Control Event Rate = proportion of control with bad outcomes
EER
Experimental Event Rate = proportion of experimental with bad outcomes
Absolute Risk
Difference btw CER and EER
ARR
When CER > EER
Increased rate of adverse outcomes in control group
ARI
When CER
Relative Risk
EER/CER
Relative Risk Reduction/Increase
1 - RR
Number Needed to Treat (NNT)
NNT = 1/ARR
Number Needed to Harm (NNH)
NNH = 1/ARI
Nominal Measurement
Measures a mere difference
your hometowns
Ordinal Measurement
Measures differences + rank order
(Place finish in race- first, second, third)
Socioeconomic status- low or high
Interval Measurement
Measures difference + rank order + equal distances between values of variable
(Temp)
Does not have true meaningful zero, even when it is 0 degrees outside, there is still a temperature
Ratio Measurement
Measures differences + rank order + equal distances between values of the variable + true meaningful zero point
(Height & weight)
Categorical measurements
Nominal
Ordinal
Continuous measurements
Interval
Ratio
If you have 2 categorical measurements, what statistical test would you use?
Chi-squared
If you have one nominal and one interval-ratio measurement, what test would you use?
T-test
What is an odds ratio?
The “odds” that some outcome would occur in the presence of a predictor vs. absence of the predictor
Odds ratio > 1
Exposure associated with higher odds of outcome
Odds ratio
Exposure associated with lower odds of outcome
Odds ratio = 1
Exposure does not affect odds of outcome
How do you calculate odds ratio?
AD/BC