Stats Flashcards
Why are stats important?
- some forms of statistics have predictive utility
- future risk
- focused on application
SD:
a measure of variability within the cohort being assessed
2 x SD will capture ____% of the _____ (_____ ) of data.
- ~95%
- range (distribution)
Standard error (of the mean):
- determining whether the mean is applicable for the entire population
- accuracy of the mean (generalizability to similar cohorts or population)
- how much do we trust this number?
- if you use the entire population, would you get a similar number?
Standard error is dependent on…
- the SD
- the sample size
OR & RR used to represent the effect of an _____ on a particular outcome.
intervention
Odds ratios (OR):
what are the odds of something happening given a particular exposure or intervention compared to control?
OR =
odds of an outcome in the exposed group / odds of an outcome in the control group
(Relative) risk ratio:
identifies the probability of an outcome given an exposure or intervention (more intuitive)
RR =
probability of an outcome in the exposed group compared to the control group
OR and RR attempt to describe…
the same effect
OR and RR express the influence of an ____ or _____ in separate ways (but they are _____).
- exposure
- treatment
- related
OR and RR _____ the occurrence of an ____ (due to ____) in reference to a ____ group.
- normalize
- outcome
- exposure
- control
OR and RR are commonly used in ____.
medicine
Validity:
- accuracy
- correctness
- does a test, instrument, questionnaire, etc. measure what it is supposed to measure?
Reliability:
- precision
- repeatability
Types of validity:
- logical (face) validity
- content validity
- criterion validity
- construct validity
Logical or face validity can be claimed when…
the measure appears to obviously assess the target variable or performance
Give an example of logical (face) validity:
- balance test of standing on one foot
- it obviously is measuring balance
Logical (face validity):
- weakest form of validity
- may be difficult to quantify
- no statistical verification
- established by expert opinion or judges
Content validity is similar to _____ validity and usually applies to ____ ____ or _____ that attempt to measure the desired ____ or a defined _____ of ____.
- logical
- written tests or questionnaires
- parameter
- domain of content
In content validity, often a table of ____ and/or ____ are developed to act as a _____.
- specifications
- diagrams
- blueprint
Content validity is established through ____ ____ or _____ _____.
- published literature
- curriculum content
In content validity, often no _____ ____ is usually required.
statistical verification
Give an example of content validity:
visual rating scale for body composition
Criterion validity:
the extent to which the results of a standard test can be compared to or used to predict a practical outcome
Criterion validity can be claimed when…
- a test measure provides an outcome similar to a criterion (standard) or previously validated test measure
- when the measure taken successfully predicts the criterion measure or gold standard
- when the measures permit inferences to be made about an underlying trait
Give an example of criterion validity:
- is a 2000m simulated rowing race on a rowing machine a valid measure of an actual on-water 2000m rowing race?
- predicting body fat from skinfolds
With construct validity, the variable of interest is…
multi-factorial/multi-dimensional
Construct validity requires more _____ statistical procedures such as…
- complex
- factor analysis
- multiple regression
- ANOVA/multivariate ANOVA
Reliability refers to the _____ or _____ of a measurement.
- consistency
- repeatability
5 factors influencing validity/reliability:
- biological variability
- technical variability
- testing variability
- environmental variability
- unknown factors
Biological variability:
- the inherent physiological and psychological fluctuations of the individual
- circadian rhythms, mood etc.
Technical variability:
precision and accuracy of the instruments
Testing variability:
instructions and manner of administering the test
Environmental variability:
- temperature
- humidity
Inter-rater reliability:
comparison of same measure between 2 (or more) testers
Intra-rater reliability:
comparison of 2 (or more) measures made by the same tester (tests the measure-er)
Test-retest reliability:
- repeated testing on 2 or more occasions
- used to test the reliability of the technique (repeatability)
6 keys to repeatability:
- same experimental tools
- same observer
- same measuring instrument, used under the same conditions
- same location
- repetition over a short period of time
- same objectives
Intra class correlation needs _____ measures for x and y (______ data).
- same
- interchangeable
- ex. HR 1 and HR 2
ICC is…
- intra class correlation
- the repeatability of a measure
How to interpret ICC:
- less than 0.40 = poor
- 0.40 - 0.59 = fair
- 0.60 - 0.74 = good
- 0.75 - 1.00 = excellent
3 ways to assess validity/reliability:
- correlation
- regression
- bland-altman
2 factors to consider for correlation:
- there should be physiological basis fundamentally linking the variables of interest
- correlation does not mean causation
Why can HR be used to correlate oxygen consumption?
- fick equation
- HR is in it, fundamentally related
_____ and ______ are often presented at the same time.
- correlation
- regression
VO2 max =
CO x a-v O2 diff.
CO =
HR x SV
Although a correlation describes the ____ of the relationship between 2 variables, it does not necessarily describe _____ of the relationship.
- strength
- pattern
Regression describes…
the numerical relationship between 2 variables
The simplest regression can be…
the linear line of best fit
Multiple linear regression:
multiple factors may also be known to be co-related, influencing the relationship of interest
Example of multiple linear regression:
Ebbeling submaximal treadmill test
How can you assess how well 1 metric compares to the “gold standard”?
bland-altman
Example of bland-altman test:
comparing PA questionnaire and accelerometer
X axis of bland-altman:
average of the 2 values
Y axis of bland-altman:
difference of the 2 values
Average difference of the bland-altman tells you …
if it on average overestimates or underestimates gold standard
Why do you need a large range when using bland-altman?
so that it is applicable to sedentary and active individuals
SD is how ____ the range is…
- big
- how tight our relationship is around offset
- want to minimize the 2 SDs
In the bland-altman, regression line (aka …) is not a _____ but you can get _____ relationship.
- line of best fit
- correlation (no r value)
- y = mx + b
A mean offset of - 2.4 in the bland-altman means…
in order to estimate our gold standard, we have to add 2.4
In bland-altman, a big range of SD is ______.
unacceptable
Line of best fit in bland-altman:
across the entire range of individuals we are working with, sometimes we are underestimating our gold standard, while other times we are overestimating our gold standard
Ideal line of best fit for bland-altman:
- you would have no difference
- you have same difference across entire range
- you have line of best fit that is flat
- your SD is narrow
Where do our guidelines come from?
meta-analysis
What is the condition for being used in a meta-analysis?
all need to be measuring the same thing
Meta-analysis:
- strict, defined process for conducting analysis
- pool data to provide a larger sample size
- the basis for many clinical practice guidelines
In meta-analysis, 0 means…
inconclusive
Meta-analysis is a highly regarding technique for interpreting _____/ ____ in data.
- variability
- controversy
Meta-analysis is used to determine ____ of outcomes across studies.
validity