Biostats /sampling Flashcards
How do we describe or quantify something?
Make it into a proportion, %, ratio
What is Nominal scale?
Categories
– positive/negative
–sex
–breed
What is Ordinal scale?
Ranked
– heart murmurs, BCS, pain, lameness
What is Interval scale?
‘True’ numbers
– age, weight, CBC
- Measures of central tendency
- Mean, median, mode
- Measures of Dispersion
- Range, SD, Percentile
Population
The ENTIRE COLLECTION of observations or subjects that have something in common and to which conclusions are inferred
Sample
A SUBSET of the population, ideally selected so as to be representative of the population
Simple random sampling
every subject has equal probability of being selected.
Stratified random sampling
population is divided into subgroups, then random sampling applied within each subgroup
** need to know characteristics about the population
Systematic random sampling
every x-th subject selected
*Must pre-determine the interval you want to select
Cluster random sampling
when sampling unit (e.g., group) is different from study unit (e.g., individual).
What is the diff btwn random sampling and randomization?
Random sampling
-Purpose is to select subjects from a population
Randomization
- after you have picked your sample, then you are randomly assigning them to a group (each group has equal chance of being assigned to the groups)
What is the difference between:
Descriptive vs Inferential statistics
Descriptive statistics apply to the study
Inferential statistics is generalized the results
- -Estimation (used to estimate dz prevalence)
- -Hypothesis testing!
What is Hypothesis testing?
- requires the assumption of a “null hypothesis”
What is a null hypothesis?
- assumes NO difference in groups
- ALWAYS assume the null hypothesis is true
- we NEVER ACCEPT
- Reject or “not reject”
What is the Alternative hypothesis?
“parameterization” of the research hypothesis
** difference in values
What is a p-value?
the probability of observing the data that we actually had OR the data more extreme assuming the null hypothesis is true.
** Does not measure degree of difference, just the probability of obtaining these results or better if null hypothesis is true.
2 factors that influence it:
- Effect/ difference / association btwn groups
- Sample size (precision)
Doe– increased sample size = decrease p-value
How do we decide when a p-value is “high” or “low”?
Alpha level / significance level
= an arbitrarily selected number ** Often 0.05
REJECT null hypoth if p < alpha
FAIL TO REJECT null hypoth if p > alpha
**Not “accept”
What does Alpha represent?
the chance one is willing to take in mistakenly rejecting a true null hypothesis (i.e. making a Type 1 error)
What are the errors that can occur by making inferences about a hypothesis?
What are terms for accurately ID’ing a hypothesis?
Reject the null hypothesis
- Null is True = TYPE 1 error
- reject when its really true –chance of doing this is < 5%
- similar to false positive
- Null is False = Power
Fail to reject null hypoth
- Null is true = Confidence
- Null is False = TYPE 2 Error
- the test did not distinguish btwn 2 groups, but there IS a diff!
- similar to False negative
Does a result that is non-significant imply there is no difference btwn the 2 groups?
NO!
How is it possible that a smaller effect results in more statistical significance?
p-value is a mixture of:
- effect size
= measure to QUANTIFY how DIFFERENT study groups are
AND
- precision
= measure to QUANTIFY VARIABILITY of an estimate.

What is the importance of Confidence intervals?
- a measure of PRECISION!
- deals with the “width” from “Point Estimate”
- PE= number used in study as cutoff
*** larger confidence with larger sample size
