Intro to Biostats lecture 29-34 Flashcards
Study Subjects:
Sample
- a subset or portion of the full population “representatives”
- useful when studying the complete population is not feasible
- random processes commonly utilized to draw a sample
- hope that info generated from sample is generalizable to entire population
Study Measurements
- data will be collected on desired variables
- dependent variables= outcome variables
- independent variables
- comparisons= statistical analyses
- inferences= made about measurements and their comparisons in relation to the null (also be made about full population of similar subjects= generalizability)
Null Hypothesis (Ho)
- researchers either accept or reject based on data analysis
- states there will be no true difference between groups being compared
- most conserved/commonly used
- perspectives: superiority, non-inferiority, equivalency
Alternative Hypothesis (H1)
-a research perspective which states there will be a true difference between the groups being compared
3 attributes of data measurement:
- magnitude
- consistency of scale
- rational/absolute zero
- Magnitude
- dimensionality
- bigger/taller/shorter/stronger
- does the data have magnitude?
- Consistency of Scale
- fixed interval
- equal, measurable spacing between units
- height= inches, feet, cm
- does the measurement being acquired have consistency of scale?
3 categories for data based on the 2 key attributes:
- nominal
- ordinal
- interval/ratio
- Nominal
- dichotomous/binary
- non-ranked, named categories
- no magnitude/ no consistency of scale/ no rational zero
- labeled variables w/o quantitative characteristics
- all data pushed into 2 categories
- ex: gender? hair color? occupation?
Study Subjects:
Population
- all individuals
- not to be confused with “study population”
- Ordinal
- ranked categories, non-equal distance
- yes magnitude/ no consistency of scale/ no rational zero
- all pain scales (even numbers are ordinal because numbers are place holders)
- Interval/Ratio
- order and magnitude and equal intervals of scale (units)
- yes magnitude/ yes consistency of scale
- no rational zero= interval
- yes rational zero= ratio
ex) number of living siblings, personal age, blood sugar, labs
how can you change levels of measurement data?
-after data is collected we can appropriately go down in specificity/detail of data measurement but never go up!
ratio>interval>ordinal>nominal
*can make interval ordinal/nominal but not vice versa
what are descrete variables?
- categorical in nature
- nominal and ordinal
what are continuous variables?
- have scale
- interval/ratio
what are the measures of central tendency
- mean/median/mode
- min/max/range
- interquartile range
- mode useful for nominal/ordinal/interval data
- mean/median/ range useful for interval only
what is variance and standard deviation?
- variance= differences in each individual measurement value and the groups mean
- interpretation of the spread of the data, tells us how uniform or not the data is
-SD= square root of variance value (restores units of mean)
Normally Distributed data
- symmetrical shape
- parametric tests most useful
- mean/median/mode are equal or near equal
- equal dispersion of curve tails to both sides of the mean/med/mode
what is the % of the population compared in 1, 2, 3 standard deviations?
1 SD= 68%
2 SD= 95%
3 SD= 99%
Positively Skewed data
- asymmetrical w/ tail pointing to the right
- skewed anytime the median differs from the mean
- when the mean higher than the median
Negatively Skewed data
- asymmetrical distribution with one tail pointing toward the left
- distribution is skewed anytime the median differs from the mean
- when the mean is lower than the median
what is skewness?
- a measure of the asymmetry of a distribution
- perfectly normal distribution is symmetric and has a skewness value of 0
- number closest to 0 is normally distributed and as you move away from 0 (3,4,8,20) seeing more skewness
what is kurtosis?
- a measure of the extent to which observations cluster around the mean
- normal distribution the value is 0 (near 0)
- positive kurtosis= more cluster
- negative kurtosis= less cluster
*discrete data usually has positive kurtosis because more cluster around
what are required assumptions of interval data?
- for proper selection of a parametric test data must be
1. normally distributed
2. equal variances
3. randomly derived and independent
how can we tell if variances are equal?
- use levene’s test
- in SPSS test to see if variances are equal
how do we handle interval data that is not normally distributed?
- use a statistical test that does not require the data to be normally distributed (non parametric tests)- treat as if ordinal or nominal
- transform data to a standardized value (z-score or log) and hope that makes it normally distributed
**always run descriptive stats and graphs to see if normal distribution
what is Power?
1-beta
- ability of the study design and selected stats test to detect a true difference if one truly exists between group comparisons and therefore is the level of accuracy in correctly accepting/rejecting the null
- ability to find the difference if there is one
how does sample size affect power?
- the large the sample size the greater the likelihood of detecting a difference if one truly exists
- increase sample size= increase power
what determines sample size?
- minimum difference between groups deemed significant
- smaller the difference between groups necessary to be significant the greater the number needed - expected variation of measurement (known or estimated)
- alpha and beta error rates (power)
Null Hypothesis (H0)
- a research perspective which states the will be no true difference between the groups being compared
- most conservative and commonly utilized
- can take various perspectives (superior, noninferiority, equivalency)
Alternative Hypothesis (H1)
-a research perspective which states there will be a true difference between the groups being compared
different ways to explain statistical significance
P value interpretation
- the probability of making a type 1 error if the null is rejected
- the probability of erroneously claiming a difference between groups when one does not really exist
- the probability of the outcome of the groups differences occurring by chance