Research and Statistics Flashcards
steps of research continuum
- identify a relevant topic
- develop a research question
- develop a hypothesis
- prepare research protocol/methodology
- organize methods and materials
- collect and analyze data
- study results/make decisions/draw conclusions
- study design and checklist
types of hypotheses
null: no relationship exists
alternative: some relationship exists
types of research
descriptive: can only determine association
analytical
- experimental: can determine c + e
- observational: can only determine association
describe qualitative research
collects qualitative information
concerned with particular interests rather than measuring numbers
describe case report/case study
observing a pt with a particular condition to further understand the mechanism of this conditions
describe correlation/ecological study
to determine if there is a relationship between naturally occurring variables
and if there is, what is it?
experimental vs. control group
experimental: receives treatment
control: does not receive treatment
parallel design experiment
participant A and B either get the treatment or the control
crossover design experiment
participant A will get both control and treatment at some point
2 period crossover design experiment
participant A will get both control and treatment with a washout period in between
quasi experimental
measures something before and measures the same thing after to compare the results of a change
cohort study
prospective: follows a group of healthy people to determine who develops the dx and information based on that
retrospective: looks back on a group with a dx to determine information
case control study
comparing a group of pts with a dx to a group of pt without the dx to determine risk factors and information (does not look at mechanism of development)
nominal variable
variables that fit into categories without order (gender, race, marital status)
rank order/ordinal variables
variables that fit into categories with order (stage 1, 2, or 3 of cancer)
numerical discrete
numbers that take on countable and distinct values (number siblings, age)
numerical continuous
numbers that take on decimals points without a distinct value (height, weight)
validity
can the test measure what it intends to measure
internal validity
is the difference between the two groups real?
external validity
can the difference between the groups be extended to a larger population
reliability
are your results reproducible?
sensitivity
sensiTivity
identifies those who are True, with the disease
specificity
speciFicity
identifies those who are False, without the disease
inferential statistics
makes inferences or conclusions from observed data
(probability, hypothesis testing, variance)
descriptive statistics
summarizes and describes small amounts of numerical information
(mean, median, mode; range; standard deviation)
standard deviation
indicates the dispersion of data about the mean or average
tells you how spread out the numbers are from the average
__% of observations lie within +/- 1 SD
68
__% of observations lie outside +/- 1 SD
32
__% of observations lie within +/- 2 SD
95
range is ___
+/- 1 standard deviation
positive correlation means that
as A increases, B increases
OR
as A decreases, B decreases
negative correlation means that
as A increases, B decreases
OR
as A decreases, B increases
r-values that indicate relationships
<0.4 weak
0.4-0.7: moderate
>0.7: strong
a negative r value indicates
a negative correlation
a positive r value indicates
a positive correlation
p-values that determine statistical significance
p<0.05 - significant difference
p<0.01 or <0.001 - very strong significant difference
p>0.05 no significant difference
prevalence
the number of existing cases over a period of time
incidence
the number of new cases over a period of time
accuracy
the degree to getting the correct answer
precision
the degree to getting the same answer over and over
observer effect
the researcher’s or interviewer’s body language or intonation affects the way the participant responds
social desirability bias
the participant responds in a way that they think they should
selection bias
the people you decide to include in your study are not representative of the population you are aiming to study
sampling error
occurs at the stage of analysis
when the data points used in the analysis do not represent the entire population of data
reporting bias
when the direction or statistical significant of results influences whether or how the research is reported
publication bias
when papers finding a null result are not selected, published, or reviewed
measurement error
the different between a measured quantity and the true value
misclassification error
a type of measurement error that applies to variables that are categorical or binary
non-response bias
with low response on a survey, you may not be capturing all the data
health volunteer basis
people who volunteer for research trials or respond to surveys may not be representative of the population (i.e. healthier, higher socioeconomic status)
EAL bases their grading on
quality of study - scientific rigor and validity, design and execution
quantity of studies and subjects - number of studies, number of subjects in studies
consistency of findings across studies - magnitude of effect
clinical impacts - importance of studied outcomes
generalizability of findings - to the population of interest
