click below
click below
Normal Size Small Size show me how
Statistics
data analysis types
Question | Answer |
---|---|
Meta Analysis | Examines and summarize many studies that have approached the sameresearch question but may have different outcome variables, inclusioncriteria, research designs and analyses |
Advantages:Meta Analysis | (1) increase power by increasing sample size and therefore detect importantdifferences that may not be present with small samples (type II error)(2) improve estimates of effect size(3) resolve uncertainty when conflicting results occur(4) improve |
Randomized Clinical Trial (RCT)Same as Randomized Controlled Clinical Trial (RCT) | Term used for prospective pretest-posttest design with comparisons betweenan experimental group and a control group design. |
Non-randomized Controlled Study | Same design as above however subjects are not randomly assigned togroups.· Often used to assess community intervention, where the intervention isprovided to well-defined, naturally occurring groups in order to addressprimary prevention. Can be used |
Cohort Study | A cohort is defined as a group of similar individuals who are followed as agroup together over time. Example: Framingham Study, longitudinal studies |
Prospective Cohort Study | They are then exposed to an intervention or otherenvironmental variables and are followed over time to see if they outcomemeasure improve or they developed the disease/disorder. |
Retrospective Cohort Study | Subjects have already been exposed to an environmental variable or anintervention and the outcome determined. Usually designed to identify whichvariables or interventions played a role in the outcome (whether it isimproved function or if they have d |
Case-Control Study | Cases are subjects who have the disorder or desirableoutcome, controls are a comparison group without the disorder or desiredoutcome. The control group is often historical where the investigator looks inthe literature or back over time to see the ou |
Null Hypothesis - AKA Ho. | When the null hypothesis is accepted (p> 0.05) we conclude that there is no“statistically significant differences” between the control and experimentalgroup in the outcome/dependent variable. |
Alternative Hypothesis - AKA Ha | Hypothesis of “difference” By rejecting the null hypothesis we can concludethat differences exist.Can be directional or nondirectional.(generally stated as > greater than or < less than the controlgroup) |
Clinical (Directional) Hypothesis | Indicates the expected direction of difference between sample means |
Independent Variable | A condition, intervention, orcharacteristic that may predict or cause a given outcome.IOW the effects of US on pain. |
Dependent Variable | The outcome variable or measured dependent variable that may change as aresult of an intervention (independent variable). |
Confounding Variable | Uncontrolled extraneous variables that may contaminate the independentvariable and affect the dependent variable, causing a separate effect if notcontrolled.· Example: pain medication, previous therapy |
Power | Power can be thought of as sensitivity to detect change.So if the power of a study is 80% this can beinterpreted as the tool having an 80% chance of detecting true differences. |
Statistical power of a test is a function of 4 factors: | 1.Alpha=The lower the a the less the chance of Type I error.2. Data variance(s2): The < the ariance in a data, the > statistical power.3) Sample size(n):The > the sample, the >the statistical power.4.the effect size (ES): Treatments that result in |
When is Power analysis used? | used to estimate the sample size needed to obtain adesired level of power before data collection and to use this estimatein recruiting subjects.collection and to use this estimatein recruiting subjects. |
p-value (Probability value) | Example: p=.18 means that there is an 18% probability that the differencebetween the means occurred by chance which is an 18% chance ofcommitting a Type I error. |
Alpha (a) | The level of significance set by the research team (usually a=0.05). Itrepresents the maximum acceptable risk (5%) of committing a Type I error(rejecting the null hypothesis when it is true). |
Beta (b) | The probability of committing a Type II error (statistical power).Acceptingthe null hypothesis when it is false.Example: b = .20, 1-b or power = .80. |