Midterm Flashcards
helps us know if true change has occurred between measurements
reliability
measure is consistent when performed multiple times on same patient/participant and construct has not changed
test-retest reliability
measurements obtained by same assessor are consistent
intra rater reliability
measurements obtained by 2 or more assessors are consistent
inter rater reliability
does the instrument or test seem to be a good choice to measure something
face validity
making clinical subjective judgement if something measures what it should measure
face validity
instrument covers all elements of construct being measured and does not include irrelevant elements
content validity
what are the two types of criterion validity
concurrent
predictive
measure of interest and measure with already established validity administered at the same time point produce consistent results
concurrent validity
measure predicts an outcome of interest well
predictive validity
instrument measures what it claims to measure (stats involved)
construct validity
doing stats to establish a relationship between things
construct validity
degree to which results of the study can be attributed to the study intervention and not extraneous factors
internal validity
a __________study is well controlled
internally valid
what does rigor in a study mean
that things are well controlled
dependent on the rigor with which the study was conducted
internal validity
process of selecting subjects leads to sample that is not representative of the target population
selection bias
when participants drop out or do not complete the study
participant attrition
solutions for participant attrition
- ) enroll more subjects
- ) account for in statistical analysis
- ) document drop-outs and reasons
seeing how many people you need in the study to see if an intervention worked
statistical power
events that happen outside the study but influence the results (out of control of investigator)
history
changes over time that are internal to participants that are not related to the study but may affect results
maturation
this would be an example of what: people with Parkinson’s condition fluctuating at different times during the day
maturation
multiple baseline testing can help with what
maturation
used to gather information to better understand a condition, test or t/x
background question
information to guide decision-making when managing a specific patient’s condition
foreground question
what does PICO stand for
P–> patient or problem
I–> intervention
C–> comparison
O–> outcomes
gives a numerical conclusion
meta analysis
synthesize findings from multiple studies to generate summary statistics
meta analysis
gives general statement as a conclusion
systematic review
answers questions by systematically reviewing and describing all relevant available evidence; similar to meta-analysis
systematic review
subjects randomly assigned to groups to compare interventions; gives you cause-and-effect
RCT
not cause and effect; observe b/c it’s unethical to randomize pts to have an injury or illness/disability. Collect data on those who have already experiences that injury or who already have that illness/disability
observation studies
two types of observational studies
- ) cohort
2. ) case control
study of exposure leading to outcome
cohort study
observational study design where ‘cases’ have condition of interest
case control study
include one or just a few patients. Used when intervention is new or novel, or when pt’s condition is superrr rare
case study or case series
best type of study when answering question about a diagnosis
prospective, blind comparison to a gold standard
best type of studies (in order) when answering a question about therapy or t/x options
RCT–>Cohort–> case control–> case series
best type of studies (in order) when answering a question about prognosis
cohort study–> case control–> case series
no mathematical properties, can’t add a value or rank to these, just categories.
nominal data
what stats can we use for nominal data
frequencies and mode
categorical but can rank order to categories. There’s no set distance between categories
ordinal data
stats that can be used for ordinal data
frequencies and modes
pain scale would be example of what type of data
ordinal
stats, eye color, marital status would be an example of what type of data
nominal data
numeric values along a scale with equal set distance between them, but there is no true zero point, can have negative values.
interval data
temperature is an example of what type of data
interval
stats that can be used for interval data
mean, median, mode
numeric values along a scale with equal distances between them and a known zero point. Can’t have negative values.
ratio data
level of assistance is an example of what type of data
ordinal
height, weight, walking speed is an example of what type of data
ratio
show proportionally the number or percent by category
pie charts or bar charts
show distribution of a variable; x axis is your scale and y axis is your frequency
histograms
plot a dependent variable on vertical axis and independent variable on horizontal axis. Each subject is a point on the chart
scatterplots
only appropriate for interval and ratio data
mean
graph’s vertex is to the right, mean shifts to the left
left skewed (neg skewed)
graph’s vertex is to the left, mean shifts to the right
right skewed (pos skewed)
mean shifts to the right when
right skewed
means shifts to the left when
left skewed
measure of how well the mean represents the data
standard deviation
amount of variability expressed as a percentage of the mean
coefficient of variation
are there units associated with coefficient of variation
no
o Compare variability btwn different measures of the same thing; Ex-different devices that measure same strength
coefficient of variation
number of SDs a data pt is from the mean of the sample or population
z score
Compare variability btwn 2 different samples on the same measure; Ex- same measure but looking at younger vs older pts
coefficient of variation
is a measure of ‘precision’ of the sample mean; how well your sample represents your population
standard error of the mean
range of values that contains the ‘true’ value with a given probability
confidence intervals
Can compare data points across data sets
z score
how well the sample represents the population
standard error of the mean
repeated measurements are consistent (synonymous with reproducibility, repeatability, consistency, dependability)
reliablity
