What is an independent variable?
a constant variable that can be manipulated by the researcher
What is a dependent variable?
a variable that is dependent on results which are measured by the researcher
What is a covariate?
a phenomenon that affects the dependent that is not of interest to the researcher, but the researcher is unable to control
What is a extranenous or confounding variable?
a phenomenon that are not of interest but may have an effect on the value of the dependent variable
What is intra and inter rater reliability
how consistent one person is at measuring something vs how consistently a group of individuals can consistently measure something
What is ANOVA (analysis of variance) used for?
statistical procedure to estimate reliability
What is randomized, repeated measures, and factorial research designs
randomized: everything is randomized; repeated: all subjects measured against the same variables; factorial has multiple independent variables and measures all subjects against all variations
What is used to calculate the association between two variables?
What statiscal test is used to predict?
What statiscal test is used to see if a Tx has an effect?
chi square, ANOVA, or T-test
Study types in a hierarchial manner?
RCT, cohort study, case-control study, case series, descriptive study
What is a meta anaylsis?
a variety of systematic reviews (a thorough review and summary of the research)
How to remember specificity and sensitivity
Sp In Sn out
Stability, repeatability, and dependability of a test or measure
Consistency of a measure made by the same person
Consistency of a measure made by 2 or more people
Appropriateness, truthfulness, authenticity, effectiveness of a test or measure
The appearance of a test or measure
The extent to which a tool measures what it claims to measure
Does the test truly sample the behavior that I wish to study
The relationship between test scores of what is being measured (test #1) and what it is being compared to (test #2)
can a future prediction be based on a measure made today
refers to the relationship between the dependent variable and the independent variable. A study will have high internal validity when all the factors that might cause a change in the dependent variable are controlled
the generalization from the sample studied to the population
the “intervention” being studied
the outcome measure used to study the independent variable
the probability of reaching a correct decision. Most statistical tests should have a power of at least 0.70
the prediction that there will be no statistically significant difference between data sets
Type 1 Error:
Stating there is a difference between two treatments when there is not; rejecting the null hypothesis when we should accept it (backing a loser) - The greater the power of a test, the less likely we are of having a Type 1 Error
Type 2 Error:
Reject the null hypothesis and stating there is no difference when in fact there is (missing a winner). - Usually occurs due to a small sample size
SnNout: high sensitivity (sn), when negative (n), rules out (out) the pathology
SpPin: high specificity (sp), when positive (p), rules in (in) the pathology
a systematic review that uses statistical modeling to summarize the quantitative results from similar studies. Studies must have a high degree of homogeneity in design, methods and data reporting. These are limited representations of the literature that are primarily designed to show either the positive or negative direction of research on a given topic
When is Kappa coefficient (very prevalent in the literature) used
Nominal: Name only data
When is Spearman Rho used?
Ordinal: a scale that implies a rank order (intervals between scores are not even
When is Pearson correlation coefficient used?
Interval/Ratio/Metric: a rank order with consistent intervals between scores
What is a regression analysis used for?
Does a treatment have an effect: nominal data what statistical test is used:
Chi squared - yes/no question
Does a treatment have an effect: ratio data what statistical test is used?
ANOVA or t-test (do 2 or more groups differ on the outcome variable?)