RM Final

• Parametric

  • Data is normally distributed (the majority of data falls around average value)
  • Independent data collected from a population (one data point is not related to another)
  • Homogeneity of variance in groups (groups have similar spreads of data)
  • Interval scale: identity, order, & equal distance between numbers
  • Ratio scale: interval scale characteristics plus true zero; weight & height
  • More powerful for detecting differences
• Nonparametric


  • Doesn't make assumptions
  • No shape distribution; no normality
  • No population parameter
  • Nominal & ordinal scales: categorical data
• Parametric tests


  • Independent T (A2)
  • Dependent T (A2)
  • One-way ANOVA (A3)
  • Two-way ANOVA (A3xB2)
  • Pearson Product Moment
• Nonparametric tests


  • Mann-Whitney U
  • Wilcoxon Matched Pairs
  • Kruskal-Wallis
  • Friedman's Matched Groups
  • Spearman Rank Order
• Eye color and hair color relationship


  • PPM – Spearman Rank Order
• Vision vs no vision effect on anticipation time, within-subjects


  • Dependent t – Wilcoxon Matched Pairs
• MCAT scores of female and male APK majors


  • Independent t- Mann-Whitney U
• VO2 max and three types of anaerobic exercise

  • One-way A3 ANOVA – Kruskal-Wallis
• Young, middle-aged, and elderly tested before and after a treatment

  • A3xB2 ANOVA – Friedman's Matched Groups
• What is the effect of four dosages of glucocorticoids on muscle wasting?


  • One-way ANOVA -Kruskal Wallis
• Does standing at a desk help people lose weight? Same people in standing and sitting groups

  • dependent t – Wilcoxon Matched pairs
• Does standing at a desk help people lose weight? Different people in standing and sitting groups


  • Independent t – Mann-Whitney U
• How much protein (15 vs 30 g) should people drink after moderately exercising for an hour? Compare baseline 30 min post, and 45 min post

  • Two-way ANOVA – Friedman's Matched Pairs
• ANCOVA


  • Parametric statistical analysis for removing the effects of a variable that is highly correlated with the dependent variable
  • Determine group differences beyond covariate
  • Removes covariate effects
• Steps involved in ANCOVA


  1. Calculate correlation of variable with outcome measure (r2)

  2. Report calculated correlation

  3. Report original means

  4. Report covariate means
• Hick-Hyman Law


  • Movement decisions vary how we process information
  • IF YOU HAVE MORE OPTIONS = YOU TAKE MORE TIME
  • Reaction time (outcome measure) increases by 150ms
  • Uncertainty increases by 1 bit (bit = binary digit)
  • Simple choice = 0 (2^0=1)
  • 2 choice RT = 1 bit (2^1 = 2)
  • Three exceptions: Incompatible stimulus-response alignment, Highly practiced tasks, Stimulus-response alignment
• Four Criteria Review for Research Methods


  • Scientific quality
  • Logical organization
  • Writing clarity: coherent, clear, concise
  • Experimental design
• Scientific quality


  • Is the problem unique?
  • Is knowledge gap identified and answered?
  • Has published literature been critically analyzed?
  • Has literature been integrated and synthesized
  • How important is problem
  • Will study contribute to the literature
• Logical organization


  4 sections of a paper: Intro/Abstract, Methods, Results, Discussion
• Experimental design


  • How many factors, how many levels of each factor? Between or within subjects?
  • True experiment needs manipulation and randomization of IV and a control group
  • Threats to internal and external validity
• Internal validity


  • The degree to which casual-effect is determined between IV and DV. You must minimize threats to internal validity. No confounding variables
  • Maturation, history, testing, selection, regression, mortality, instruments
• External validity


  • Generalizability of results
  • Selection x Rx, Setting x Rx, History x Rx, Multiple Rx interference
• Relationship between science, statistics, experimental design and variability
• Explaining variability is the goal of science. Statistics are used to evaluate patterns of availability. Variability arises from IV manipulation and human error
• ORCID RECORD All funded UF researchers are required to share their ORCID record

  o Three reasons for including ORCID:

  § Journal submissions.

  § Reviewer recognition

  § Funding agencies

  § Minimize burden of PI (added by UF)
• Oaths for Scientist


  What is our goal as young researchers?
• Five Institutional Review Boards at UF


  • IRB-02: Behavioral and non-medical studies
  • Rights and welfare of human participants must be upheld
• Hawthorne effect


  Subject's behavior is altered by observation
• Rosenthal effect (Pygmalion)


  Higher expectations placed upon people lead to higher performance
• Multivariate Statistics


  • Multiple regression and correlation
  • Meta-analysis
  • Canonical Correlation
  • Discriminant function analysis
  • Factor analysis commonalities
  • MANOVA
• Regression Analysis


  • What is the relationship between Y and X?
  • Simple regression: one predictor variable
  • Residual: how much the actual data point deviates from the model?
  • R^2: coefficient of determination
  • How much variance is explained
  • Closer to 1 = more variance explained
  • Multiple regression: Two or more IV on one DV
• Probability plots


  • Ideal: data matches assumed distribution (behaving as expected)
  • Light-tailed: lower probability of extreme outcomes
  • Heavy-tailed: higher probability of extreme outcomes (higher level of uncertainty)
  • Positive skew
  • Negative Skews

  
• Patterns for residual plot


  • Satisfactory: assumptions of the regression model are met
  • Funnel: variability changes across predicted values (you can't conclude anything)
  • Double blow
  • Nonlinear

  
• r tells you how closely two variables are related, while R2 tells you how well a set of variables can predict another variable in a regression model
• Fitt's Law


  • Movement time is a function of the amplitude or distance of the required movement & the accuracy required at the time of movement
  • Speed accuracy trade off. Faster = less accurate
• Research Approaches


  • Bench to Beside: taking studies from the lab and test treatments on human health protocols
• Clinical Trial Stages


  • Purpose: Rx evolution, Rx efficacy, Rx effectiveness, Delivery method
  • Participants: Small groups; within subjects; repeated measures, Medium group designs; randomized groups, Large group designs; Multisite RC trials, Large group designs; population and institutions
  • Outcomes: Individual Rx; Performance based; Quantify Rx effect, Health status and impact on society, Institutional and impact on society
  • Limitations: High impact publication; less rigorous, Problem identifying outcome measures, Vulnerable to political policy; Data misused
• Ethical transgressions


  • Plagiarism
  • Fabrication
  • Falsification
• Writing Persuasive Arguments (MAGIC)


  • Magnitude
  • Articulation
  • Generality
  • Interestingness
  • Credibility
• Young Scientist Responsibilities


  • Ask critical questions: What experiment could disprove your hypothesis? What hypothesis does your experiment disprove?
  • Thoughts are principled arguments. Make an argument with supportable premise