Exam 1 Flashcards
What is the order of the evidence pyramid from top to bottom?
Systematic review of RTC or N of 1
RCTs
Systematic review of cohort studies
Cohort studies
Systematic review of case control studies
Case control studies
Case study or case series, cross sectional study
Clinical experience, expert opinion, mechanisms based reasoning
What do the higher levels of the evidence pyramid mean?
The higher it is the more trustworthy the study
Efficacy
Benefit of a treatment delivered in a highly controlled and ideal environment
Effectiveness
Benefit of a treatment delivered in a pragmatic manner under real world conditions
Deductive reasoning
Acceptance of a general inference or premise and drawing a specific conclusion
What do you begin with in deductive reasoning?
Known principle and use your observations to confirm, reject, or modify this conclusion
Inductive reasoning
Developing generalizations from specific observations
What does inductive reasoning begin with?
Experience and results in conclusion
What are both deductive and inductive reasoning used for?
To design research studies and interpret the research
What are types of descriptive studies?
Developmental research
Normative research
Qualitative research
Case study
What are types of exploratory studies?
Cohort studies
Case control studies
Correlational research
What are types of experimental studies?
RCTs
Quasi experimental
Single subject designs
What do descriptive studies do?
Describe populations
What do exploratory studies do?
Find relationships
What are the three essential components of an experimental research?
Include a control and comparison group
Independent variable manipulated by the experimenter
Subjects are randomly assigned to groups
Case control study
Retrospective
Rare outcomes
Multiple exposures
Fast
Inexpensive
Weak evidence
Cohort study
Prospective
Rare exposures
Multiple outcomes
Slow
Expensive
Strong evidence
Prospective study
In the future
Retrospective study
In the past
Cross sectional study
In the present
Single subject or N of 1 study
Cause and effect due to rigorous planning, including reliable and valid outcome measures (1 patient)
Case study
Retrospective
Less standardized and controlled
Less internal validity
What is the purpose of case studies?
Providing future research directives
What does it mean if p is less than alpha?
Reject the null hypothesis
What does it mean if p is greater than alpha?
Retain the null hypothesis (fail to reject)
Reliability
Reproducibility and consistency
Validity
Accuracy and correctness
Minimal detectable change (MDC)
The ability of an instrument to detect change beyond measurement error
Minimal clinically important difference (MCID)
Ability of an instrument to detect minimally important change
Responsiveness
Ability of an instrument to detect minimal change over time
SpPin
A test with high specificity
That is positive
Helps rule a condition in
SnNout
A test with high sensitivity
That is negative
Helps rule a condition out
Likelihood ratios
Sensitivity information combined with specificity information
+LR
Sensitivity/ (1-specificity)
-LR
(1-sensitivity)/specificity
What means a large and conclusive shift in probability for +LR and -LR?
> 10 for +LR
<0.1 for -LR
Null value
Does not change the probability at all
What happens to a pre test probability if you have a +LR?
It shifts higher on the post test probability
What happens to a pre test probability if you have a -LR?
It shifts lower on the post test probability
How do the mean and median shift when the data is negatively skewed?
To the left
How do the mean and median shift when it is positively skewed?
To the right
What’s the empirical rule?
In a normal curve 68% will fall within 1 SD and 95% will fall within 2 SD
What are types of quantitative variables?
Discrete and continuous
Discrete variable
Variable with a finite number of values (rolling a 6 sided dice)
Continuous variable
Variable with an infinite number of values (temperature)
Quantitative variable
Measured numerically
Qualitative variable
Allow for classification based on some characteristic (gender)
Independent variable
Any variable being manipulated
Dependent variable
Any variable that is being measured
Nominal
Names (categories)
Sex or gender
Ordinal
Names/categories but they have order to them (salary)
Interval
Numerical data (temperature in degrees) (can go below zero)
Ratio
Numerical data (distance, age, weight, time)
(Cannot go below zero)
Point estimate
Single value that represents the best estimate of the population value
Confidence interval
Range of values that we are confident contains the population parameter
Descriptive statistics
Works with a smaller data set
Process is simpler
Results obtained represent entire data set
Error is usually less
Inferential statistics
Works with large data set
Process is more complex
Results obtained represent a portion of population
Error is usually more
Type 1 error
Claim there is a difference when there is not
False positive
Rejecting null hypothesis when you should not have
Type 2 error
Claim there is no difference when there is
False negative
Failing to reject the null when you should not have
What does it mean if p is less than alpha?
Reject the null hypothesis (there is a statistical significance)
What does it mean if p is more than alpha?
Fail to reject the null (is not statistically significant)
What does it mean when alpha is set at 0.05?
5% probability of a type 1 error
Null hypothesis
There is no change
Alternate hypothesis
There is a change from baseline
What does it mean if you have a 95% confidence interval?
5% chance the results are outside the true range
What does a narrower confidence interval mean?
More precise and less variation
What does a larger confidence interval mean?
Less precise and more variation
What does it mean if the confidence interval goes through 0?
It is not statistically significant
What does being more precise mean in confidence intervals?
Less confident
How does MCID relate to confidence intervals?
If the data is statistically significant than we look at the MCID and see if it met the point estimate (if not we move onto power)
What does it mean if the confidence interval includes the MCID?
Underpowered
What does it mean if the confidence interval does not include the MCID?
Adequately powered
What are the four pillars of power?
Alpha
Effect size
Variance
Sample size
What is the equation for power?
(Sample size)(effect size)(alpha)/variance
What is the relationship of power to type 2 error?
You can find power by doing 1 minus the probability of a type 2 error
What is A Priori estimation?
Sample size estimation (make power 80% and solve for sample size)
Post hoc power analysis
Only done if not statistically significant
Solve for power to see if you hit 80%
What are the basic assumptions we need to meet in order to run parametric statistical analyses?
Samples are random
Normal distributions
Homogeneity of variance within groups
Continuous data (ratio or interval)
When would you use a two tailed hypothesis testing?
Non directional hypothesis (your alpha at .05 means you divide 5% on either side of bell curve)
(Are the results different in any way)
When would you use a one tail hypothesis testing?
Directional hypothesis (greater than or less than 5% all on one side)
(More statistical power)
(Are the results just greater or just less)
What is an independent (unpaired) t test used to do?
Compare two independent groups via pre and post testing
What is a dependent (paired) t test used to do?
Compare one group (measured twice)
What do both independent and dependent t test use to determine statistical significance?
Alpha and p value
Example of an independent t test?
One group gets hand splint
One group gets regular activity
Example of a dependent t test
Same group tested twice once with pillow support once without
Repeated measure factors
Within subjects (dependent)
Non repeated factors
Between subjects (independent)
Is a two tail or one tail more frequently used?
Two tail
