Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
53 Cards in this Set
- Front
- Back
Types of validity
|
Logical (face) validity
Content validity Construct validity Criterion validity Predictive validity Discriminant validity |
|
Logical (face) validity
|
Measure obviously involves what is being measured
Not sufficient to show validity (no statistical evidence) Example: static balance test that involves standing on one foot for time |
|
Content validity
|
Not sufficient to show validity (no statistical evidence)
Example: writing an examination Writing a test that adequately samples the material covered in the course |
|
Construct validity
|
Construct validity refers to the extent to which operationalizations (aka tests) of a construct (i.e. quality of life, function, aerobic capacity) actually measure
Degree to which scores from a test measure a hypothetical construct Used to relate a test results to a behavior Example: sportmanship Certain behaviors are expected of someone with a high degree of sportmanship Compare number of compliments to tennis opponent from people with high and low sportmanship scores |
|
Criterion validity
|
Degree to which results are related to a recognized standard
Criterion: previously validated or accepted measures Degree of the relationship between the 2 measures Used when a shorter, easier test is developed |
|
Predictive Validity
|
e.g. Coronary heart disease risk prediction sheet based on LDL cholesterol
|
|
Discriminant Validity
|
Degree that an instrument can differentiate between different constructs
Example: patient satisfaction survey Environment of care vs. delivery of care Instrument’s ability to differentiate among individuals with differing levels of a variable Example: high functioning elderly vs. lower functioning elderly |
|
Levels of measurement
|
Nominal
Ordinal (0-3 scale of disability) Interval Ratio (1 RM, distance walked in 6 min) Qualitative -> toward more quantitative if it isn't nominal, it's quantitative |
|
Nominal level of measurement
|
Nominal basically refers to categorically discrete data such as name of your school, type of car you drive or name of a book. This one is easy to remember because nominal sounds like name (they have the same Latin root).
|
|
ordinal - level of measurement
|
Ordinal refers to quantities that have a natural ordering. The ranking of favorite sports, the order of people's place in a line, the order of runners finishing a race or more often the choice on a rating scale from 1 to 5. With ordinal data you cannot state with certainty whether the intervals between each value are equal. For example, we often using rating scales (Likert questions). On a 10 point scale, the difference between a 9 and a 10 is not necessarily the same difference as the difference between a 6 and a 7. This is also an easy one to remember, ordinal sounds like order.
|
|
Interval level of measurement
|
Interval data is like ordinal except we can say the intervals between each value are equally split. The most common example is temperature in degrees Fahrenheit. The difference between 29 and 30 degrees is the same magnitude as the difference between 78 and 79 (although I know I prefer the latter). With attitudinal scales and the Likert questions you usually see on a survey, these are rarely interval, although many points on the scale likely are of equal intervals.
|
|
Ratio level of measurement
|
Ratio data is interval data with a natural zero point. For example, time is ratio since 0 time is meaningful. Degrees Kelvin has a 0 point (absolute 0) and the steps in both these scales have the same degree of magnitude.
|
|
“ In order to be an effective measurement tool in clinical practice, a scale should possess the following fundamental properties:
|
(a) validity: that is, it actually measures what it purports to measure;
(b) reliability: that the measure is repeatable and reproducible when measured by single and different observers; (c) sensitivity; (d) specificity; (e) ease of use; and (f) non-specialist dependence.” |
|
Common statistics used to measure reliability
|
ICC: Intraclass correlation coefficient
SEM: Standard Error of Measurement |
|
ICC: Intraclass correlation coefficient
|
Repeated measures ANOVA used to determine if there is a statistical difference between cases in a cluster
Uses mean squares of subjects and error Ratio of between-groups variance to total variance Correlation between 0 and 1.0 which measures agreement *****Interpretation***** 0 to 0.50 = Poor Reliability 0.50 to 0.75 = Moderate Reliability 0.75 to 1.0 = Good Reliability |
|
SEM: Standard Error of Measurement
|
Represents the typical deviation between repeated measures in terms of the original measurement unit
Example: Goniometric measures between 2 PTs for knee flexion ROM SEM = 8° worse reliability SEM = 3° better reliability |
|
Experimental design
Quasi-experimental Non-experimental |
Experimental design
Random assignment Controlled manipulation of an independent variable (e.g. getting a treatment such as a back manipulation), study effect on outcome (dependent variable- e.g. back pain, functional ability etc.) Quasi-experimental No random assignment Controlled manipulation is preserved Non-experimental Neither |
|
External Validity
|
External validity;
Representativeness, Generalizability Sample Size, Population Does sample represent population outside of it? External validity is the validity of generalized (causal) inferences in scientific studies, usually based on experiments as experimental validity.[1] In other words, it is the extent to which the results of a study can be generalized to other situations and to other people.[2] For example, inferences based on comparative psychotherapy studies often employ specific samples (e.g. volunteers, highly depressed, no comorbidity). If psychotherapy is found effective for these sample patients, will it also be effective for non-volunteers or the mildly depressed or patients with concurrent other disorders? |
|
Internal validity
|
Internal validity is a property of scientific studies which reflects the extent to which a causal conclusion based on a study is warranted. Such warrant is constituted by the extent to which a study minimizes systematic error (or 'bias').
|
|
Threats to validity:
|
1. Concepts 1a: Assignment/allocation, randomization, matching
Concept 1b: Selection 2. Attrition, missing data (large drop-out rate for a reason) 3. Resentful demoralization 4. Compensatory rivalry 5. Diffusion 6. Regression to the mean - outliers likely to change 7. Testing - get better at doing the test 8. Maturation - kids get more coordinated as they grow 9. Concept: Instrumentation (using bad instruments, the wrong instruments, using it inappropriately etc.) 10. Confounding - history, you can't control everything |
|
Sensitivity
|
Test’s ability to obtain a positive test when the target condition (disease) is really present
Proportion of individuals who test positive for a condition out of all of those who actually have it True positive rate 8 people with tear test positive with trendelenburg test, 3 people with it test negative, sensitivity is 8/11 = 73% sNOut - with high sensitivity, if someone tests negative, rule them out |
|
Specificity
|
Test’s ability to obtain a negative test result when the condition is really absent
Proportion of individuals who test negative for a condition out of all of those who do not have it True negative rate sPin - with high specificity, if someone tests positive, rule them in |
|
Positive likelyhood ratio
|
How many more times likely a positive test will be seen in those with the disease than in those without the disease
reflects the odds that a person who obtains a score in the “disordered/positive/affected” (positive) range on the test really DOES have the condition = Sensitivity/1- specificity = true positive rate/false positive rate 1 is unimportant, get's more important as it increases |
|
Negative likelihood ratio
|
How many more times likely a negative test will be seen in those with the disease than in those without the disease
Would like this to be low reflects the odds that a person who obtains a score in the “normal” range on the diagnostic indicator really DOES NOT have the disorder Closer to 0 = important, 1 is unimportant |
|
Nomogram
|
Graphical tool
Estimates overall probability of disease based on diagnostic testing results Pretest probability = 20% +LR = 4 Posttest probability ??? Pretest probability - probability that an individual has the target condition before the test is carried out Prevalence of the condition in the clinic Gut instinct about the probability the patient has the condition |
|
What kinds of data can mean, median and mode be used for?
|
Mean - interval or ratio, sensitive to extreme values
median - ratio, interval, ordinal, less sensitive to extreme values Mode - numeric and ordinal data, most commonly ordinal, but only measure for nominal data |
|
What percentage of values fall within 2 standard deviations from the mean?
|
at least 75%, regardless of distribution
34% on each side up to 1st deviation, 13.6% more in 2nd |
|
Parametric vs. non-parametric tests
|
Parametric:
-assumptions about the distribution (randomly selected population from normal distribution, numeric data) - robust to some violations to the assumptions Non-parametric: -distribution-free tests -fewer assumptions - less efficient |
|
Independent t-test
|
difference in means between two independent groups
(experimental and control) |
|
Paired t-test
|
same group is measured twice
mean at baseline and after treatment |
|
One-way ANOVA
|
one independent variable
e.g. shoulder elevation change in ROM in patients after surgery at clinic one, two, and three |
|
Two-way ANOVA
|
multiple independent variables
e.g. shoulder elevation change in ROM in patients after surgery at clinic one, two, and three, plus comparing men vs. women So, looking for effect of clinic, sex, and interaction effect |
|
Repeated measures ANOVA
|
multiple measurements taken on groups over time
physical function in obese older adults 4 groups: control group, diet, physical activity, diet and physical activity |
|
Regression (linear)
|
Relationship between independent variable(s) and dependent variable
Can also predict a dependent variable Can include explanatory variables (covariates) Fit a straight line Can do everything ANOVA can do (and more….) Multiple assumptions to consider Robust procedure |
|
Coefficient of Determination: R2
|
Amount of variability in the outcome (dependent variable) that can be explained by the independent variables
Bone density decreases as people age Age R2 = 30% Age R2 = 85% (R "squared") |
|
Effect size
|
“Statistical expression of the magnitude of the differences between two treatments or the magnitude of a relationship between two variables”
Standardized measure of change from baseline to follow-up measurement Interpretation 0.20 minimal effect 0.50 moderate effect 0.80 large effect >1 BIG effect Used often in meta-analyses Use is expanding into other types of studies |
|
Chi-square (X2)
|
Comparison of the observed with the expected frequency
Assessing an association of one nominal variable with another Uses the chi-square distribution LOTS of uses for the chi-square test |
|
Logistic Regression
|
Used with a binary dependent variable
Dead/alive finds the equation that best predicts the value of the Y variable for each value of the X variable probability of obtaining a particular value of a nominal variable Uses odds rather than probability Can determine odds ratios to explain association between dependent and independent variable(s) And confidence intervals or p-values for statistical significance |
|
Odds ratio
|
We have a horse name Epi
60% probability of winning the race (P) 40% probability of losing race (1-P) So what are the odds of winning? Odds = P/1-P The ratio of the number of ways the event can occur to the number of ways the event cannot occur Prob of Epi winning to prob of Epi losing Odds of winning = 60%/40% - 1.5:1 = 1.5 Odds are different from probability of winning = 60% |
|
Interpreting Odds ratios
If the exposure were not related to the disease, what would you expect the odds ratio to be? |
If the exposure were not related to the disease, what would you expect the odds ratio to be?
Well the odds of cases having been exposed would be the same as the odds of controls being exposed, so……. The ratio of the 2 odds would be 1, so OR = 1 This is the null hypothesis for statistical significance testing Ho: OR = 1 If the exposure was positively related or associated with the disease, then the OR will be….. GREATER THAN 1 Conversely, if the exposure was negatively related or associated with the disease, then the OR will be….. LESS THAN 1 |
|
5 steps for EBP
|
Ask a focused clinical question
Search for the best research evidence Appraise the quality of the research evidence Integrate the research evidence with information about the patient and clinical expertise Reflect on the process to improve in the future |
|
PICO stands for
|
Patient/problem
Intervention Comparison Outcome Focuses the question Directly applies to the patient from which is arose Makes the question easier to search Clarity Specificity |
|
Evidence pyramid
(levels of evidence) |
Cochrane systematic reviews
other systematic reviews and meta analyses Evidence guidelines Evidence summaries randomized controlled trials, case cohorts, control studies Clinical research critiques other reviews of literature case reports, case series, practice guidelines clinical reference texts |
|
Types of quantitative study designs
|
Observational
Experimental Single-subject Review |
|
Systematic review
|
Include
Specific research question Inclusion and exclusion criteria Elaborate and thorough search strategies Standardized data collection from appropriate literature Preestablished quality criteria Meta-analysis Statistical analysis of pooled data |
|
Assess validity: diagnostic test
|
Was the diagnostic test evaluated in a representative spectrum of patients?
Was the gold standard applied regardless of the index test result? Were the individuals performing and interpreting each test’s results unaware of the other test’s results? Are the test characteristics presented? Were the methods for performing the test described in sufficient detail to permit replication? |
|
Components of an EBP Technology Profile
|
Pull information technology
Push information technology Reference management system |
|
Quantitative vs. qualitative
|
Generalizability, Prediction, Causal explanations
The aim is to classify features, count them, and construct statistical models in an attempt to explain what is observed. Underlying premises Social facts have an objective reality Variables can be identified and relationships measured Outsider's point of view 3. Researchers’ role Detachment and impartiality Objective portrayal 4. Approach Quantitative data is able to test hypotheses, but may miss contextual detail. Manipulation and control Uses formal instruments Experimentation Reduces data to numerical indices Clear language in write-up vs. Aims Contextualization, interpretation Understanding actors' perspectives The aim is a complete, detailed description Underlying premises Reality is socially constructed Variables are complex, interwoven, difficult to measure Insider's point of view 3. Researchers’ role Researcher as instrument Personal involvement Empathic understanding 4. Approach Qualitative data is 'rich', comprehensive- ends with hypotheses and theory Observation and recording Researcher as instrument Naturalistic Searches for patterns, complexity Makes minor use of numerical indices, use words, pictures, objects Descriptive write-up |
|
Inclusion and exclusion criteria
|
Inclusion criteria: primary traits of the target and accessible population
Clinical findings, demographic and geographic factors Temporal factors Exclusion criteria: factors that would preclude someone from being a potential participant Factors that may introduce bias in the study (confounders) |
|
Type I error
|
there was no difference, but you found a difference
false positive |
|
Type II error
|
there was a significant difference, but you failed to find it
Influenced by four study components Sample size Chosen type I error Size of the difference between the groups Variance |
|
Confidence Intervals
|
Range of values within which the true score of the variable is estimated to lie
Characterizes statistical significance Also precision and accuracy Often 95% confidence intervals are used Corresponds to p-value of 0.05 |
|
Power
|
Probability of rejecting a null hypothesis when it is indeed false
Greater power reduces chance of type II error Greatly influences sample size Often set at 0.80 |