Statistics Flashcards
The terms used to describe how closely your results reflects the truth?
Accuracy and Validity
The terms used to describe the reproducibility of your measurement.
Precision/Reliability/Reproducibility
A hypothesis that says there is no statistical significance between the two variables. This result is the opposite of what researchers wish to prove because the outcome is due to chance.
Null Hypothesis
A hypothesis that says there is a difference (or an effect) between two or more variables. This result is anticipated by the researchers.
Alternative Hypothesis
When we use this term, our evidence has proven that our group results are related and not due to chance (exposure is associated with an outcome).
Reject the Null Hypothesis
When we use this term, our evidence has proven that our group results are due to chance (exposure is not associated with an outcome).
Fail to Reject the Null Hypothesis
This is an error that occurs when we reject a null hypothesis, when the null hypothesis is in fact, due to chance (determining that an effect is real when it is due to chance).
Type I Error (Alpha) or False Positive
This is an error that occurs when we “fail to reject the null hypothesis” - despite the evidence suggesting that the results are related (determining no effect is present when one does exist).
Type II Error (Beta) or False Negative
Outcome variable
Dependent Variable
Variable that impacts the outcome variable.
Independent Variable
A scale used for labeling variables without quantitative value (labels or names).
Nominal
A scale used for the order of values presented but the differences between them in unknown (i.e. satisfaction, happiness, pain).
Ordinal
A scale that is numerical, which we know both the order and the exact differences between the values (i.e. temperature). There is no true zero so it isn’t possible to compute ratios.
Interval Data Scales
A scale that tells us about the order, the exact value between units, AND they also have an absolute zero (i.e. weight and height)
Ratio Scales
The strength between two interval/ratio-level variables.
Pearson’s Correlation
Compares two ordinal-level variables.
Spearman Rank Correlation
Explain some properties of a correlation coefficient (r).
- Ranges from -1 to +1
- A “0” means there is no relationship between the variables.
- (-1) is a perfect negative correlation
- (+1) is a perfect positive correlation
- May be associated with a p-value
If you were given a mean for an independent group (of two), what statistical test would you use?
An Unpaired Two-Sample T-Test
If you were given a mean for an independent group (of more than two), what statistical test would you use?
ANOVA
If you were given a mean for a dependent group, what statistical test would you use?
A Paired Sample T-Test
If you were given a proportion/percentage/raw number/category, what type of statistical test would you use?
Chi-Square (x^2)
The practical importance of a treatment effect, whether it has a real, genuine, palpable, noticeable effect on daily life. No standards for calculating clinically important changes in outcomes.
Clinical Significance
Determined by the p-value & confidence interval (p<0.05 and 95% CI does not contain a 0).
Statistical Significance
What is an indicator of clinical significance?
Effect Sizes
True/False: A greater effect size indicates a larger difference between experimental and control groups.
True
True/False: You can have a statistically significant finding that is not clinically significant.
True
Explain the inverse relationship of significance levels and Type I/II Errors.
The smaller the significance level (a<0.01) decreases Type I error but increases Type II error (more false negatives). The opposite is also true.
Explain what power is defined as in statistics.
The statistical power of a study is the power, or ability, of a study to detect a difference if a difference really exists (the probability of correctly identifying a true positive). Power is dependent on sample size (larger sample = larger power).
Power Equation
1-ß
What increases when power increases?
Type I Errors (alpha) or False Positive
When the sample mean is higher than the population mean, the difference is called?
Sampling Error
What estimates the accuracy and precision of the mean (of a population)?
Confidence Intervals
What makes a confidence interval more precise? How do we increase this precision?
The more narrow, the more precise. To make the CI more narrow, increase the sample size.
What is imprecise data more prone to?
Type II Errors or False Negatives
What would a significant CI value look like?
P<0.05 and the CI would NOT contain a zero.
What does it mean if the p-value is above 0.05?
Acceptance of the Null Hypothesis
What does it mean if the p-value is less than 0.05?
Reject the Null Hypothesis (Alternative Hypothesis is true)
True/False: The p-value is highly dependent on sample size.
True
The probability of an outcome if the null hypothesis is true.
P-Value = Type I Error
Prevalence at a specific point in time; useful for comparing at different time points to spot an outbreak.
Point Prevalence
Prevalence across a span of time; useful for chronic disease.
Period Prevalence
If new cases are greater than death rates or a cure, what is happening to prevalence?
It is increasing.
If death rates or a cure is higher than the new cases coming in, what happens to prevalence?
It is decreasing.
Test correctly identifying condition.
True Positive
Test incorrectly identifying condition.
False Positive
Tests correctly excluding condition.
True Negative
Tests incorrectly excluding condition.
False Negative
The probability that a diagnostic test will correctly identify those with a condition.
Sensitivity
What does SNOUT mean?
A highly sensitive test, when negative, rules out disease.
The probability that a diagnostic test will correctly identify an individual as disease-free when disease absent.
Specificity
What does SPIN mean?
A highly specific test, when positive, rules in disease.
What are sensitivity tests used for?
Diagnostics or Screening
What are specificity tests used for?
Confirmation Testing
True/False: Sensitivity and Specificity are dependent on prevalence?
False, but they can be affected by a disease severity in a population.
The probability that a person with a positive test has the disease.
Positive Predictive Value
Are PPV and NPV dependent on disease prevalence?
Yes
The probability that a person with a negative test does not have the disease.
Negative Predictive Value
As prevalence decreases, what happens to PPV and NPV?
PPV will also decrease, as NPV increases.
A high PPV is associated with a ?
High Specificity
A high NPV is associated with a high ?
Sensitivity
When using ROC plots (Receiver Operating Characteristic Curve) (y-specificity vs x-sensitivity), the closer to the ____ ____ _____ _____, the better the test performance.
Top upper left corner
What is the difference between an observational study and a non-observational study?
An observational study does not manipulate the independent variable, while the non-observational study does.
A brief report of a clinical characteristic or outcome from a single subject. There is no control group.
Case Report or Case Study
A brief report of a clinical characteristic or outcome from a group of clinical subjects. There is no control group.
Case Series Report
Exposure and outcome are evaluated at the same time.
Cross - Sectional Study or Prevalence Study
Identify a disease and try to find out the cause. Here you can compare cases of diseased individuals to controls or non diseased individuals; with their respect to their level of exposure to a suspected risk factor. Controls are used in this study on some factor, and they are RETROspective.
Case-Control Study
Follow a group forward to determine risk factors for an outcome (can be followed through time). Exposed and unexposed individuals are compared in relation to the disease incidence. PROspective.
Cohort Study or Incidence Study or Longitudinal Study
What are three non-observational studies?
- Randomized Controlled Trial (RCT)
- Community Trial
- Cross-Over Study
In this study, you test a treatment against a control. The group members are randomly assigned (“Gold Standard”). PROspective but the exposure is being manipulated.
Randomized Controlled Trial (RCT)
For ethical reasons, no group can be left untreated in this study. All groups get the intervention but at different times (treatment vs placebo effects).
Cross-Over Study
What do we use to measure ratios for Case-Control Studies?
Odd Ratios
What do we use to measure ratios associated with Cohort Studies and RCTs?
Risk Ratios
What do we use to measure risk for clinical trials or cohort studies?
Attributable Risk (AR) or Attributable Risk Reduction (ARR)