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;
74 Cards in this Set
- Front
- Back
ANCOVA |
Parametric Reject Null if F value above critical value Analyzing covariates (independent variables) effect on dependent variables through and F test (see ANOVA) |
|
Chi Squared |
Parametric "Goodness of Fit" Reject Null if Chi Squared is greater than critical vlaue |
|
Kruskal-Wallis |
Non-Parametric Reject null if KW value > Chi Squared value Compare medians of groups within independent variable, similar to ANOVA but nonparametric |
|
Mann Whitney U Test |
Non-Parametric Reject null if U value lower or equal to critical value Determines differnces in median (same shape) or distribution (different shape) in data of Ivs |
|
T Test |
Parametric Reject null if test value is greater than t-value Compare means of groups for difference based off variability though a T-value(variation between groups/variation within groups) - similar to ANOVA but 2 groups |
|
Wilcoxin Signed Rank Test |
Non-Parametric Reject null if test value is less that critical value Compare sample medians - similar to T-test but nonparametric |
|
ANOVA |
Parametric Reject null if F value above critical value Compares means of 3+ groups for a significant difference in order to reject the null hypothesis (Ho)Use F ratio (variation between groups/variation within groups) to determine |
|
Correlations |
Parametric or Non-Parametric Reject null if test value greater than critical value (spearman's)Magnitude and direction are used as values Determines strength of relationship between variables |
|
MANOVA |
Parametric Reject null if F value above critical value Same as ANOVA but for multpile Dependent Variables |
|
Regression |
Linear and small variance from best fit Magnitude and direction are used as values Similar to Correlation, determines how well a variable predicts an outcome |
|
We use a T-test to compare the data of ankle pain before intervention and ankle pain after intervention in the same subjects. What kind of T-test would we use? |
Dependent Sample |
|
What are the three main types of T-tests? |
Independent Samples Dependent Samples One Sample |
|
When are logistic regressions used? |
When making predictions of dichotomous outcomes(dependent variable) from single or multiple variables (independentvariable). |
|
What does the R^2 value tell us in a regression analysis? |
It tells us how strongly a variable predicts an outcome |
|
If we are testing (ANCOVA) the effectiveness of medication A on reducing headaches, what would a possible covariate be? |
Stress level |
|
The results from your Kruskal- Wallis test are similarly distributed. You can use median data. True or false? |
True |
|
Homoscedasticity |
Same variance across a range of values for an independent variable |
|
Which critical values are used for the Wilcoxon signed rank test? |
W and Z |
|
Covariate |
variable that is possibly predictive of the outcome under study Confounding Variable or interacting |
|
Types of plaigarism |
Copy and Paste Purchased "Paper Mills" Word Switching Style Metaphors/Ideas Self Plaigarism |
|
Don't Need to Cite... |
Myths/Folklore Historical Events Generally Accepted Facts |
|
Freely available EBP resources |
PEDro PubMed Cochrane Turning Research into Practice (TRIP) National Guideline Clearinghouse APTA |
|
Elements of Prognosis |
Possible Outcomes Likelihood that outcomes will occur Time frame for change |
|
Research Designs for Prognosis |
Cohort Designs (Outcome) Case Control Designs (Exposure) Cross Sectional |
|
5 and 20 rule |
Lose less than 5% is very strong Lose more than 20% threatens validity |
|
Crude Case Analysis |
Do not take into account those that are lost to follow up Count the 4 deaths out of 84 instead of 100 when 16 lost 4.8% |
|
Worst Case Analysis |
Assume that the patients lost to follow up got worse / died 16 lost to follow up are counted with the 4 dead out of 100 total 20% |
|
Best Case Analysis |
Assume those lost to follow up got better / Alive Only count the 4 who died as negative outcomes 4% |
|
Kaplan Meier Curves |
Survival curves for the proportion of the sample who has NOT had a specific outcome |
|
Only valid if they fall within the confidence interval If they include 1, they are no better than chance |
Odds Ratio Relative Risks Hazard Ratio |
|
Confidence Interval shouldn't cross... |
Zero |
|
Outcome based Intervention is to... |
Evidence Based Practice |
|
Outcome Measurements Establish... |
What works How well it works Who it works for Economic efficiency |
|
Outcome Measurement Research Designs |
Prospective Cross-sectional or longitudinal pro: over time con: difficult, long, need to find baseline patients Retrospective **Preferred Establishes temporal sequence |
|
Types of Secondary Data for Retrospective Outcomes Measurements |
-Institutional: hospitals, clinics, heath systems (forinternal use) -Commercial: uniform data systems, FOTO, insurance claims -Government: National Center for Health Statistics, Agency for Health CareResearch Quality, Medicare/Medicaidh |
|
Test-Retest Reliability |
Consistencyof repeated measures separated in time, indicates stability over time
|
|
Cross Sectional Validity |
Thereis a difference in scores of the measurement tool between 2 or more groups ofresponses on the reference tool during a simultaneous administration
|
|
Longitudinal Validity |
Thereis a difference in scores of the measurement tool between 2 or more groups ofresponses on the reference tool administered in the future
|
|
Person Level Outcomes |
Activity limitations, Quality of Life, participation restrictions Body Function Activity Participation |
|
Minimal Detectable Change (MDC) |
Smallest amount of change that can be reliably detected outside of measurement Error |
|
Minimal Clinically Important Difference (MCID) |
Minimal amount of change that patients Perceive as beneficial and would justify a change in care |
|
Health Related Quality of Life (HRQL) |
Multidimensional assessment of life satisfaction as it relates to health |
|
Reliability Statistics Cronbach's Alpha |
Internal consistency/reliability Expected correlation of two test that measure the same construct |
|
Reliability Statistics
Intra-class correlation (ICC) Coefficient |
Reproducibility |
|
Reliability Statistics
Kappa (K) |
Agreement among repeated Scores |
|
Reliability Statistics Effect Size and Standardized Response Mean (SRM) |
Responsiveness |
|
Floor and Ceiling Effects |
Failure to fully characterize a group of patients Ceiling = all healthy patients score maximally Floor = cannot differentiate between different elderly populations |
|
Content Validity |
Full content of a concepts definition |
|
Face Validity |
Appears to measure what it intends to measure |
|
Concurrent Validity |
Outcome has a high correlation with gold standard at the same time |
|
Predictive Validity |
Outcome has high correlation with future Gold Standard measure |
|
Criterion Standard Validity |
Compare to gold standard |
|
Reference Standard Validity |
Not gold standard but reasonable |
|
Construct Validity |
Ability of a tool to measure an abstract concept |
|
Convergent Validity |
Outcome correlates with another thought to measure the same construct |
|
Discriminant Validity |
Outcome does Not correlate with with a measure thought to assess a different characteristic |
|
Clinical Prediction Rules are a cluster of symptoms that can be used to... |
Improve diagnosis Refer patients for additional testing Establish Prognosis Create Subgroups |
|
Clinical Prediction Rule Step 1 |
Prospective Cohort Derivation of rule list all possible factors 10-20 analyze which factors are most reliable |
|
Clinical Prediction Rule Step 2 |
Prospective Cohort Validation of rule Narrow = one clinic Broad = multiple clinics |
|
Clinical Prediction Rule Step 3 |
Randomized Control Trial Impact Analysis Rule changes clinician behavior Improves outcomes and reduces cost |
|
Clinical Prediction Rule Level 1 |
Highest Level At least 1 prospective validation and Impact Analysis Randomized Control Trial |
|
Clinical Prediction Rule Level 2 |
Broad Prospective Validation Wide Geographic region Varying confidence but no certainty |
|
Clinical Prediction Rule Level 3 |
Narrow Prospective study to validate Add to practice with caution |
|
Clinical Prediction Rule Level 4 |
Derived but not validated Needs further investigation |
|
Systematic Review Selection Bias |
Cherry Picking No good exclusion/inclusion criteria |
|
Systematic Review Language Bias |
Neglecting none English papers |
|
Systematic Review Including Published and unpublished data |
Should only be published Unpublished is cheating papers in |
|
4 types of threats to Systematic Review |
Subject allocation Groups managed differently Outcome measures blinded Loss of Subjects |
|
Narrative Research |
Focuses on stories of individuals |
|
Grounded Theory Research |
theory is "grounded" in participant data Generate a theory from participant views until you saturate the model |
|
Case Study Research |
Bounded by context or setting Non-Reproducable |
|
Participation Action Research (PAR) Research |
Photovoice Private troubles that participants have in common |
|
Content Analysis Research |
Examination of artifacts of social communication may count phrases or discuss symbolism |
|
Theoretical Saturation |
Data starts to repeat itself Not hearing new information |