EXAM III Material Flashcards
Define Population
All individuals
DIFFERENT from study population (the final group of individuals selected for a study)
Define Sample and state when it’s useful to be used
A subset or portion of the full population
“representatives”
Useful when studying the complete population is not feasible
Study measurements of human studies are collected based on:
Desired “variables”
Dependent variable(s) outcome variable
Independent variables
In which comparisons will be made = Statistical analysis
and inferences will be made about the sample-derived measurements and their comparisons (in relation to Null Hypothesis) & to the full population of similar subjects (generalizability)
Define Null Hypothesis
A research prospective which states that there will be no true difference b/w the groups being compared
Most conservative and commonly utilized
What statistical references can be taken by the researcher in a null hypothesis?
Superiority
Noninferiority
Equivalency
Don’t always have to show superiority; sometimes equivalency is good enough
What are the 2 key attributes of data measurement/variables in which help to determine the statistical test?
Magnitude/Dimensionality - i.e. pain level scale, satisfaction, fanciness
Consistency of scale/fixed interval - equal, measurable spacing between units - i.e. date, time, months, age groups, etc.
A 3rd one is rational/absolute zero
What are the 3 categories for data/variables based on the answers to the two key attributes of magnitude and consistency of scale?
Nominal/Dichotomous/Binary; Non-Ranked named categories
Ordinal/Ranked Categories; equal-distance
Interval/Ratio (order & magnitude & equal intervals-of-scale (units))
Define the nominal category for data/variables
Has No magnitude, No consistency of scale, No rational zero
Dichotomous, Non-ranked, Categorical
Simply labeled variables without quantitative characteristics
Descrete = Whole Numbers
No mean - cannot interpret mean
i.e. male/female, hair color, eye color, etc.
Define ordinal level of data measurement and list examples
Yes Magnitude
No Consistency of scale
No Rational Zero
Descrete = whole numbers
May calculate Mean; must be careful tho
i.e. pain level scale, ranking, satisfaction
Define Ratio/Interval level of measurement and list examples
Yes Magnitude
Yes Consistency of Scale
N/Y Rational Zero (N = Interval, Y = Ratio)
Continous = Fractional Numbers
Can calculate mean, median, standard dev.
i.e. age, number of living siblings, anything physiological measured (BP, lipid panel, etc.)
What type of data measurement is applicable while having a Parametric test (normally distributed shape of data distribution)
Interval
MMM are essentially equal
Equal dispersion of curve “tails” to both sides of MMM
What does it mean to have a positively skewed graph?
When the tail points to the right
Mean > Median
What does it mean to have a negatively skewed graph?
When the tail points to the left
Mean < Median
Define skewness, what is the value depicting a symmetric, normal distribution?
A measure of the asymmetry of a distribution
Value of 0 = perfectly normal, symmetric, equal MMM
The further from 0, the more the data is skewed
Define kurtosis and what (+) and (-) kurtosis means
A measure of the extent to which observations cluster around the mean
How well the values cluster around the mean/middle
(+) = more cluster within the graph/bell shape
(-) = less cluster within the graph/bell shape
What are the required assumptions of interval data for proper selection of a parametric test?
Normally distributed
Equal variable = Levene’s Test; used to determine if the interval data is equal; within the SBSS test
Randomly-derived and Independent
ALL must be true in order to pick Interval measurement
What question does Confidence Interval address?
What is the plausible range of possible difference or relationship within which I believe the true difference or relationship may lie?
What question does the p value address from a statistical test?
What is the single measurement value most likely to represent the true (yet unknown) difference or relationship between the groups being compared and what is the probability the difference has occurred by chance?
p value is attributed to Type I or Type II error?
Type I error = False Positive
When a test detects the presence of disease when in fact the person does not have the disease
Define p value
The probability of making a Type I error if the Null Hypothesis is rejected.
What does is mean when a graph has a kurtosis value of 0?
A perfect cluster of obversations at the mean of the bell shaped curve = a normal distribution
Which type of data(s) are discrete and why?
Nominal and Ordinal
Variables are discrete because they do not have a scale and you cannot have something in between the variables, not continuous
What is a Levene’s test and why is it used?
A test that is done when you need to determine if the data has magnitude and a balanced scale
Tells us if the data is interval data, if there are equal variances
Asses for equal variances b/w groups
What requirements must be met before choosing a Parametric test?
Interval data must be:
Normally distributed (bell-shaped curve)
Contain equal variances
Randomly-derived and independent
What type of stat test is used for non-parametric tests?
Descriptive Stats and Graphs
Data is transformed into a standardized value
(z-score or log)
These are stat tests that does not
Define Power (1-beta)
The ability of a study design/methodology/selected test statistic to detect a true difference if one truly exists b/w group-comparisons and the level of accuracy in correctly accepting/rejecting the Null Hypothesis (analogous to sensitivity in screenings)
Value is traditionally set at 20%
How does the sample size impact the power of stat significance?
The larger the sample size, the greater the likelihood/ability of detecting a difference if one truly exists
Increase in Power
Must add in anticipated drop-outs or loss to follow-ups
In determining sample sizes, would you need a larger and smaller number of samples when the differences between groups are smaller?
Need a greater number (N)
Due to the fact that there is a small difference, it will be harder to find differences in a small group, therefore you need more samples
Add in anticipated drop-outs or loss to follow-ups
What is another word for a type I error rate?
p value
If < 0.05 it is statistically significant
Determined before a study begins
Low p values allow more confidence and have a less risk of Type I error
by chance alone
Define Type I Error and state another term for it
alpha
Rejecting the Null Hypothesis when it is actually true and you should have accepted it
There is really NO TRUE DIFFERENCES b/w groups but you in error reject the null hypothesis, thereby stating that you believe there’s a difference b/w the groups when in fact, there really isn’t!
= False Positives
Define Type II Error and state another term for it
Beta
Not rejecting the Null Hypothesis when it is actually false and you should have rejected it
When there really IS A TRUE DIFFERENCE b/w the groups being compared but you in error do not reject the null and state that there is no difference when there actually is
= False Negative
What are the most common selected confidence interval values?
90%, 95%, 99%
Explain the interpretation of a 95% CI
We are 95% confident that the “true” difference (0) or relationship (1) between the groups is contained within the CI range
List the common types of measures of central tendency
Mean, Median, Mode
Outliers
Minimum and maximum range
Interquartile Range
Define the two measures of variation/spread/dispersion of data
Variance (from the mean) - the difference in each individual measurement value and the groups’ mean
Stardard Deviation - The square root of variance value
What are the 3 population percentages comprised within 1, 2, and 3 SDs around the mean of a normally distributed dataset?
1 = 68%
2 = 95%
3 = 99%
Differentiate between dependent and independent data and list the common terms used for dependent data
Dependent data is when you have data from the same/paired groups = before/after, pre/post, beginning/end, start/finish
Independent data is when you have data from different groups
What 4 questions must you ask yourself in while determining a correct stat test to use?
- What type of data is being collected/evaluated?
Nominal/Ordinal/Interval
- What type of comparison/assessment is desired?
Correlation, Survival, Regression = stop here
Frequencies, Proportions, Counts = proceed to Qs 3 and 4
- How many groups are being compared?
- Is the data independent or dependent?
What is a Correlation test, when would you want to use it, what do they tell you and whatare the types of correlation tests for the 3 data types?
Provides a quantitative measure of the strength and direction
When is a partial correlation test used?
When you are performing a correlation test and want to control for confounding variables