11) Ch11: Data Analysis Flashcards
Descriptive Stats
Used to summarize patterns/characteristics in the data to describe the distribution of data points
- Include sum, mean, mode, standard deviation, and frequency of nominal variables
- Typically performed on all measured/recorded variables in a study
Correlative Statistics
Used to calculate how variables are similar or relate to each other/ Strength of relationships btwn variables
- Includes Pearson Product Moment Correlation Coefficient, ICC, Spearman Rank Correlation Coefficient
Pearson Product Moment Correlation Coefficient
Used to determine if the scores of 2 variables vary together
ICC
Used to determine if the scores of 2 variables vary together and how closely those scores are matched
Spearman Rank Correlation Coefficient
Used to determine if the scores of 2 variables of ordinal data vary together
- Same as Pearson except used w/ordinal data
Comparative Stats
Used to compare 2+ sample groups on 1+ measured attributes; Describe how variables are different; Determine whether 2+ groups of data are different and suggest cause-and-effect relationships between an intervention and an outcome
- Includes t-tests, ANOVA, ANCOVA, MANOVA (Parametric), Mann-Whitney U Test, and Wilcoxen Signed-Ranks Test(Nonparametric)
Independent T-Test
Compares the means of 2 groups in which there aren’t any repeated measures
- Assumes a normal distribution of the sample and random selection & assignment
Paired T-Test
Compares the means of 2 repeated measures from the same sample group or from matched pairs of subjects
ANOVA
Compares the means of 3+ groups on a single measure or 3+ repeated measures on the same group
ANCOVA
Compares the means of 2+ groups on 2 measures or repeated measures on 2+ groups
MANOVA
Compares 3+ independent variables
Mann-Whitney U Test
Compares the rank order of scores from 2 groups on an independent variable
Wilcoxen Rank Sum Test
Compares the rank order of scores from 2 groups on an independent variable w/sample sizes <30 subjects
- Same as Mann-Whitney but used for sample sizes <30 subjects
Kruskal-Wallis ANOVA
Compares rank or ordinal data from 3+ groups
Clinical Significance
Differences in the measured variable that result in useful changes
Statistical Significance
Occurs when the data analysis results in a number that exceeds level of chance
Sampling Bias
Imbalance of characteristics in a group of pt’s studied so the group is not a true representation of the typical defined subject
Procedure Variability
Range of methods used by clinicians to perform the same task or to achieve the same endpoint
- In clinical practice, interventions are not done exactly the same for each pt bc you have to adjust based on pt-specific factors
Measurement Bias
Imbalance in the data resulting from the choice of a measurement tool, from inherent errors associated w/the application or recording of the measure, from limits of tool construction, and variations in documentation on the measure
- Approach to tx’ing acute LBP → Exam processes from one clinic may yield a broad array of tests and measures whereas another clinic may yield a smaller different set based on their certifications
- Neither way is necessarily better than the other but it might make comparison tougher → Need to interpret results, inputs, and processes w/the understanding that they might be unique and that causes limitations
Historical Bias
Imbalance in the data resulting from an event that happens outside of the pt care interaction, which might influence some aspect of pt care or outcomes
Experimental Bias
Imbalance of data that occurs when participants or investigators have expectations for outcomes bc they know who or what is being studied so their objectivity may be consciously or unconsciously be swayed
- Can influence statistical interpretation bc when a certain outcome is anticipated, the investigator may only see that outcome and fail to recognize other outcomes of the data set
