Exam 2 Flashcards
The less precise a measurement is, the lower association between:
that variable and any other variable
What can effect whether or not the strength of a relationship is underestimated?
Imprecision
A strong association may be found if a measure is:
Reliable and Fairly Precise
What must you consider when interpreting and applying the results of research into the association between variables?
Whether the sample studied truly represents the range of the population of interest
Generalization of results into practice is a judgment of the clinician, but the investigator must provide sufficient description
***What is the fundamental statistic for the analysis of interval and ratio data to estimate the strength of the association between variables?
Regression Analysis
***What is a special case of regression?
ANOVA
***What is the simplest form of regression?
Simple Linear
***What is used to measure the strength of the association between pairs of data that are interval or ratio?
Pearson r Value
***What can you use to estimate the association between multiple variables? And what test is better suited for this?
Multiple Regression; ANOVA
What is the value of a multiple regression test?
R2 Value
What does R2 measure?
The variance in the criterion variable (Y) explained by a predictor variable (X) or predictor variables
What is the difference between normal values for R and R2?
r: (-1 to 1)
R2: (0 to 1)
What value is indicative that the differences observed when studying a sample are reflective of true population differences?
P-Value
***What are the guidelines for interpreting r values?
Little, if any correlation: .00 to .25; 00 to -.25
Low Correlation: .26 to .49; -.26 to -.49
Moderate Correlation: .50 to .69; -.50 to -.69
High Correlation: .70 to .89; -.70 to .89
Very High Correlation: .90 to 1.0; -.90 to 1.0
What do Measures of Association not imply?
Cause and Effect
Although measures of association confirm cause and effect, these analyses can build useful what?
Predictive Models
What analysis method is used most often for Ordinal Data?
Spearman Rank Order Correlation “Spearman r” (e.g. Likert Scale)
What is the formula for a Spearman Rank Order Correlation?
r = 1 – ((6∑d2) / n(n2 – 1))
What is the most common approach for Nominal Data?
Cramer’s V Correlation “Chi Square”
What is one example of a Spearman r analysis?
Likert Scale
What is one example of a Chi Square analysis?
Testing the hypothesis that two nominal variables are independent
What can a Cramer’s V analysis identify?
A relationship, but it may have no suggestion of cause and effect
Are Spearman r and Cramer’s V parametric or nonparametric statistics?
Nonparametric
Are parametric or nonparametric statistics more powerful? Why?
Parametric, nonparametric uses less information in its calculation
What scales belongs to which type of statistics?
Nominal and Ordinal are nonparametric
Interval and Ratio are parametric
What type of data will you get with a Pearson r test?
Interval or Ratio
Generalization of results into a clinical practice is a judgment of who, and requires thoughtful decisions on the part of whom?
The clinician; the investigator
Estimating the usefulness of any diagnostic procedure involves comparing the results of a diagnostic test to the results of what?
“Gold” standard
What reasons would a gold standard test not be applied to all diagnoses in question?
Setting, timing, scope, costs, and risks associated with the diagnostic process
The results from studies of diagnostic tests are reported in what forms?
Estimates of Sensitivity and Specificity
Positive and Negative Prediction Values
Positive and Negative Likelihood Ratios
Receiver Operator Characteristic Curves
When comparing the results from a diagnostic test with a dichotomous result of interest to the results of a “gold standard” diagnostic test, how many subgroups will be formed?
4
Sensitivity and Specificity are related to the ability of a test to identify what?
Those with and without a condition
What is sensitivity?
The number of illnesses of injuries that are correctly diagnosed by the clinical examination procedure being investigated (A) divided by the true number of illnesses/injuries (A + C)(Gold standard measure)
What is the sensitivity formula?
A/A+C
What is specificity?
The number of individuals correctly classified as not having the condition based on the test being investigated (D) divided by the true number of negative cases (B+D)(Gold standard measure)
What is specificity formula?
D/D+B
Tests that have a high sensitivity are at what?
Ruling out a condition
Have relatively few false negative findings
Tests that have a high specificity are good at what?
Ruling in a condition
Have few false positive findings
Do tests with high specificity have few or many false positives?
Few
What is the formula for Positive Prediction Values?
A/A+B
What is the formula for Negative Prediction Value?
D/D+C
What is the correlation between prevalence, PPV, and NPV?
As prevalence falls, PPV falls and NPV rises
What are values derived from estimates of sensitivity and specificity?
Likelihood Ratios
Knowledge of the Likelihood Ratio influences ________ that a condition does or does not exist at the end of the examination
The Level of Certainty
What sort of ratio is an impact of a positive examination finding on the probability that the target condition exists?
Positive Likelihood Ratio
What is the formula for a Positive Likelihood Ratio?
Sensitivity/(1 - Specificity)
How do you interpret Positive Likelihood Ratios?
A Large +LR Rules In a disorder
What sort of ratio is an impact of a negative examination on the probability that the condition in question is present?
Negative Likelihood Ratio
What is the formula for Negative Likelihood Ratios?
(1 - Sensitivity)/Specificity
How do you interpret a Negative Likelihood Ratio?
A Large -NR Rules Out a disorder
How do you interpret Likelihood Ratios?
LR > 1: Indicates probability that the target/problem/disease is present; +LR
LR < 1: Indicates a decreased probability that the target/problem/disease is present; -LR
LR = 1: No change in the probability
What is the full LR Table?
> 10: Large and often conclusive increase in the likelihood of disease
5 - 10: Moderate increase in the likelihood of disease
2 - 5: Small increase in the likelihood of disease
1 - 2: Minimal increase in the likelihood of disease
1: No change in the likelihood of disease
0.5 - 1.0: Minimal decrease in the likelihood of disease
0.2 - 0.5: Small decrease in the likelihood of disease
0.1 - 0.2: Moderate decrease in the likelihood of disease
< 0.1: Large and often conclusive decrease in the likelihood of disease
What is the Pretest Odds Ratio formula?
Pretest Probability/ (1 - Pretest Probability)
What is the Posttest Odds Ratio formula?
Pretest Odds * LR
What is the formula for Posttest Probability?
Posttest Odds/ (Posttest odds + 1)
What is the purpose of an ROC Curve?
It is helpful for determining a cut-off score for deciding things like the number of positive special tests are required to determine a diagnosis
How can you tell if a test discriminates between the presence or absence of a disorder?
It examines the trade off between sensitivity and specificity when you select different cut-off values
What are the interpretations of an ROC Curve?
A truly useless test has an area of 0.5 (it is no better than flipping a coin)
A perfect test has an area of 1.00 (a test with 0 false positives or false negatives)
What are the ROC score breakdowns?
.9 - 1.0: Excellent .8 - .9: Good .7 - .8: Fair .6 - .7: Poor .5 - .6: Fail
Which point, that gives a balance between sensitivity and specificity of a test, should be used as a cutoff in an ROC curve?
The point where the curve “turns”
What does STARD stand for?
Standards for the Reporting of Diagnostic Accuracy Studies
How many aspects are looked into for the STARD?
25
What is the framework of risk identification and injury prevention?
- Establishing the extent of the injury problem (Incidence, Severity)
- Establishing etiology and mechanism of sports injuries
- Introducing a preventative measure
- Assessing its effectiveness by repeating step 1
What is the importance of Establishing the extent of the injury problem?
Done through prospective injury surveillance and anecdotal observations
What is the importance of Establishing the etiology and mechanisms of the sports-related injuries?
The cause of a sports injury may be traumatic or atraumatic
Identifying the mechanism of injury: observational research
Identifying risk factors that may be involved in the etiology of a specific injury: collection of baseline data to assess potential risk factors followed by a prolonged period of injury surveillance
What is the importance of Introducing a preventative measure?
The intervention should be based on information gained in the first two steps of the paradigm
Avoid the use of the “shotgun approach” of injury prevention
What is the importance of Reassessing the extent of the injury problem?
If the incidence and severity of injuries have been substantially reduced, the permanent implementation of the intervention is likely warranted
What kind of design is the gold standard for studies of injury risk factors?
Prospective Cohort Design
How does a Prospective Cohort Designed study work?
A large of participants potentially “at risk” for injury of interest are baseline tested. Participants are followed over time to determine if they go on to suffer the injury
Are Case Control studies retrospective or prospective?
Retrospective
How does a Case Control study work?
Individuals who have already suffered the injury are compared to a control group that has not suffered the injury
What are some of the limitations of Retrospective Designs?
Case Control have smaller numbers of subjects compared to prospective cohort studies
Many rely on patient self-report for their injury history
“Risk factor” measure are taken after the injury of interest has occurred, so there is no way to know if the measures were present before the injury occurred
What is the proportion of a sample that has a given injury or illness at a single time point?
Prevalence of an injury
What is the formula for Prevalence?
(# of injured subjects/# of total subjects) * 100
What is the number of new cases of the pathology in a given period of time?
Incidence of an injury
Why is Incidence reliant upon using a prospective study design?
A surveillance system needs to record the number of new cases
What is the number of new cases that occur per unit of person-time at risk?
Incidence Rate
What is the Incidence Rate formula?
of New Injuries/Total Exposure Time
What is the number of newly injured individuals in a defined population over a given period of time?
Incidence Proportion
What is the Incidence Proportion formula?
(# of Newly Injured Subjects/# of Total Subjects Participating)
What is the difference between Injury Risk and Injury Rate?
Injury risk has to do with the probability of a new injury per individual, where injury rate has to do with the number of new injuries per unit of exposure time
The simplest comparison between two measure of injury incidence is to calculate the ______ of the injury incidence between the two groups
Ratio
What is a Prevalence Ratio?
If the ratio of the injury prevalence estimates between two groups is taken
What is a Risk Ratio?
If the ratio of injury risk estimates between two groups is calculated
What is a Rate Ratio?
If the ratio of the injury rate estimates between two groups is determined
_______ _________ provides a proportion of injury incidence between two groups and is identical to the calculation of risk ratio.
Relative Risk
What is Relative Risk Reduction?
The percentage that the experimental condition reduces injury risk compared to the control condition
What is the Relative Risk Reduction formula?
(1 - Relative Risk) * 100
What is Relative Risk Increase?
If the experimental condition is found to lead to heightened risk of injury, the sign is changed to positive and termed RRI
***What is Absolute Risk Reduction?
The difference between the injury rate/risk in the intervention group and the injury rate/risk in the control group
It indicates the reduction in number of injuries per 100 people who received the intervention
What is a Numbers Needed to Treat?
The number of patients that need to be treated with the experimental treatment to prevent one injury compared to receiving the control condition
What is the formula for NNT?
(1/Absolute Risk Reduction)
How do you analyze NNT?
1: The ideal
Infinity: An infinite number of patients would need to be treated to prevent one injury
A clearer estimation of the treatment effect of the intervention becomes evident by calculating what?
Odds Ratio
What is CER, and how do you calculate it?
Control Event Rate = C/(C + D)
What is EER, and how do you calculate it?
Experimental Event Rate = A/(A + B)
***What are the 4 categories of Statistical Data?
Nominal: the assignment of numeric values for analysis of nominal data is arbitrary
Ordinal: data are ordered in a particular and meaningful manner (numeric pain scales)
Interval
Ratio
***What is the difference between Parametric and Nonparametric stats?
Parametric stats are appropriate for analyzing interval and ratio data under most circumstances
Nonparametric stats are used to analyze nominal and ordinal data
What are the 2 measures of dispersion for interval and ratio data?
Variance and Standard Deviation
What are reported for ordinal data and are the mode for nominal data?
Median and Range Value
What do parametric stats analyze?
The distribution of variance, hence the term “analysis of variance” (ANOVA)
What does an ANOVA test measure?
It compares the means between 2 or more groups when the outcome is a continuous variable
What are the different types of ANOVA tests?
One-Way: 1 Independent Variable
Two-Way: 2 Independent Variables
Repeated Measures: Allows us to compare the effect of an intervention over time
What is the result of ANOVA?
F value
What is the formula for an F value?
Mean Square Explained/Mean Square Unexplained
What is the variation from the mean that is attributed to factors beyond the scope of the research design?
Unexplained Variance
How do you interpret F values?
The larger the F value, the less likely that the differences observed were chance occurrences
Researchers are willing to accept less than a 5% risk that an F value obtained is a chance occurrence
What happens when the F value is larger?
The null hypothesis is rejected, meaning that differences are due to the effects of the intervention
Which values specifies the level of accepted risk of incorrectly concluding that observed differences do not reflect true differences in a population of 100?
Alpha Value
What test has to do with cases where more than one dependent measure is analyzed simultaneously?
Multivariant Analysis of Variance (MANOVA)
MANOVA is best applied when the investigator is interested in the effect of what?
The independent variable(s) on the collection of dependent variables
When is a Post-Hoc Analysis done?
When the ANOVA F-test is significant (good method to find group differences)
What does a significant ANOVA F-test imply?
It is evidence that not all group means are equal, but it does not identify where the differences exist
Type I vs Type II Errors
Type I (false positive): occurs when a null is rejected when in face population differences do not exist. The alpha value is really the level of risk of Type I error (The Null Hypothesis is true, but the decision based on random sample is rejected) Type II (false negative): occurs when a null is not rejected yet a study of the population would reveal differences between groups (The Null Hypothesis is false, and the decision based on the random sample is not rejected)
How can you guard against Type I errors?
Selecting the Alpha Level
What is the typical alpha level selected for research?
0.05; the chances are less than 5% that an “oddball” sample or “false positive” result
What is the Beta, and what is it often set at?
The probability of experiencing a type II error; 20% -> there is a 20% chance of finding no effect of treatment when there really was one
What is statistical power, and how do you calculate it?
Measurement of how powerful an experiment is; = 1 - Beta
What can you conclude if a Type II error is made?
The power of the experiment was not adequate
What is called when you estimate sample size prior to the study?
Priori
What does power determine?
The number of subjects in a study to detect a statistically significant difference; Power of 80% means that in 80% of the time, the researcher will avoid a Type II error
What helps to decrease the risk of a Type II error?
Stronger Statistical Power
What 3 factors influence Power?
The mean difference between groups
The variance within groups
Sample Size
Out of the 3 factors that influence Power, which factor is the only one that investigators can control?
Sample Size
What is a Mixed Model?
When a between-subjects variable and a within-subjects variable exist in a research design
Are greater complexity in research designs and data analysis necessarily an indicator of better research?
No, not always
What all does Statistical Significance depend upon?
Magnitude of the treatment effect, sample size, reliability of treatment effect, and reliability of the measurements
ANOVA, t-tests, and nonparametric procedures address the probability that differences observed reflect population differences but do not address what?
The magnitude of differences; reporting confidence intervals solves this
What is a Confidence Interval?
A range of scores with specific boundaries that should contain the population mean
What are the boundaries based on?
The sample mean and its standard error
What would cause a confidence interval to be statistically insignificant?
It contains 0
What does it mean if a 95% Confidence Interval is larger?
More caution is required when using the estimate
What does MCID mean, and what is it?
Minimally Clinical Important Difference; the smallest amount of difference that is required to be considered important/significant (Minimal change that you want in your outcome)
True or False: Use of a ROC analysis is not always preferable.
True
______ is the smallest treatment effect that would result in change in patient management, given the side effects, costs, and inconveniences.
MCID
_____ effects may be useful if outcomes are rare; _______ effects may be meaningful if outcomes occur frequently
Large; Modest
What is a special case of ANOVA in which there are only two sets of data in the comparison?
T-tests
What is the difference between t values and F values?
t values may be positive or negative; F values are always positive
Why should you not conduct a t-test for multiple comparisons?
More tests conducted increases the risk of a Type I error
What is a calculation that shows the typical response to intervention?
Effect Size
What is the most commonly referenced method for calculating effect size?
Cohen’s d
What impact does effect size have on group differences?
The larger the effect size, the greatest differences between the groups
***How do you interpret Effect Sizes?
0.2 - 0.49: small effect
0.5 - 0.79: moderate effect
> 0.8: large effect
Why are nonparametric procedures good?
They test hypotheses about medians or nominal data distribution
What are the 3 most common nonparametric procedures?
Mann-Whitney U
Kruskal-Wallis One-Way Analysis of Variance by Ranks
Friedman Two-Way Analysis of Variance by Ranks
What test is the Mann-Whitney U test analogous to?
Paired t-test; results are designated as a T
In what situations would you use the Friedman Two-Way ANOVA by Ranks?
Analyses in which there are repeated measures within one group
When would the Kruskal-Wallis One-Way ANOVA by Ranks be useful?
When there are more than two groups; the result is an H-Value
True or False: None of the three main nonparametric tests allows for analysis of repeated measures from multiple groups (mixed model design)
True
What is the BESS Test?
Balance Error Scoring System
Measure the number of errors during 20 second time frame for each of the 6 conditions with eyes closed
Double Limb Support (feet together); Firm/Foam Surface
Single Leg Stand: Firm/Foam Surface
Tandem Stance: Firm/Foam Surface
Describe the 10 Meter Walk Test
Time to walk 10 meters (m/s)
Average of 3 trials
Can use assistive device
Allow space for acceleration/deceleration
What do you need to calculate effect size (r)?
Means
Standard Deviations
Cohen’s D
What is BESD?
Binomial Effect Size Display
Helps further interpret clinical significance, when using effect size
1.5x more successful
What can be calculated using the r value?
Relative Success Rate
How do you calculate Success Rates?
50% + r/2 (converted to a %) for the intervention group
50% - r/2 (converted to a %) for the control group
What all does Outcomes research involve?
The study of interventions with the goal to improve care of the patient and ultimately lead to societal benefits of improved health care
What is one type of outcomes research?
Clinical research assesses the effectiveness of therapeutic interventions
***What are the levels of Evidence for Treatment Outcomes?
Systematic Reviews with Meta Analysis Randomized Control Trials Prospective Cohort Case Control Trials Case Series Expert Opinion
***What is the gold standard for research on treatment outcomes?
Large scale, multi-site RCTs
***What are the perks of Smaller Scale RCTs?
More Feasible
Less Expensive
More Common
What is an example of a multi-site study?
Perform the intervention at Creighton, and then replicate it at an outside clinic
What are Clinical Trials?
The term “clinical trial” generally refers to research performed on patients to assess an intervention
In the NIH world, the term refers to a large-scale, often multi-center RCT
What is sampling?
When a researcher/research study selects a subgroup from a population
When does sampling bias occur?
When the subjects over/underrepresent characteristics related to the problem studied
Why would the ability to generalize findings from a sample to a population be limited?
Important biases in the selection of subjects exists
What type of validity is directly related to Generalizability?
External Validity
An “a priori power analysis” to determine the number of subjects involves understanding the outcome data and statistics of what?
The level of significance of the test (a = .05; probability of Type I error)
Statistical Power (B = .80; Probability of Type II Error)
Effect size of the experiment: the smaller the expected treatment effect, the more subjects needed
Expected variability in the data (standard deviation)
What are the inputs for effect size and variability based on?
Pilot data and/or reported findings from the literature
What is CONSORT?
Consolidated Standards of Reporting Trials
Evidence-based, minimum set of recommendations for reporting randomized trials. It offers a standard way for authors to prepare reports of trial findings, facilitating their complete and transparent reporting, and aiding their critical appraisal and interpretation
How many items are analyzed in CONSORT?
22
What is the PEDro Scale?
Physiotherapy Evidence Database
What is the purpose of the PEDro Scale?
To help identify which of the known or suspected RCTs are likely to be internally valid and has sufficient statistical information to make results interpretable
How many items are analyzed on the PEDro Scale?
11