130 likes | 151 Views
Single sample. One-tailed versus two-tailed tests t -test: Use Significance level (type I and II errors) Degrees of freedom (df). Two samples. Unpaired or independent samples t-test Paired or dependent samples t-test. Non-parametric statistics.
E N D
Single sample • One-tailed versus two-tailed tests • t-test: Use Significance level (type I and II errors) Degrees of freedom (df)
Two samples • Unpaired or independent samples t-test • Paired or dependent samples t-test
Non-parametric statistics • Chi-square Test (χ2): Used with nominal variables in two independent categories • Works by comparing the actual scores to probability estimates • Rank-order methods: Scores transformed into ranks • Wilcoxon matched-pairs signed ranks test: For dependent samples • Mann-Whitney U test: For independent samples
Multigroup statistical methods • ANOVA (Analysis of Variance) • Saves time • Reduces probability of significance by chance • Rationale: Between group and within group variance • Is variability between groups > variability within groups • Statistic: F ratio and df
One-way ANOVA: Similar to independent samples t-test • Two-way ANOVA: Two independent variables • Two main effects and one interaction • Multiple comparison methods: Post-hoc tests • Scheffe (unequal sample size) • Tukey (equal sample size)
Randomized blocks ANOVA (RBANOVA): For repeated measures designs • Analysis of covariance (ANCOVA): Controls for variables that might affect data • Non-parametric multigroup tests • Friedman two-way analysis of variance • Kruskal-Wallis one-way analysis of variance
Complex statistical methods • Multivariate ANOVA (MANOVA): More than one dependent variable • Discriminant analysis: Estimates the contribution of each dependent variable in separating groups. • Factor analysis: Organize observations into underlying constructs
Research proposal outline • Title • Abstract • Table of contents, budget, resources, biographical sketches • Research protocol • Specific aims • Background and significance • Methods • Ethical issues: IRB • References
Title page • Title: Short, succint, unambiguous. • “A study to investigate physiological measures of anxiety and their relation to stuttering adaptation during oral reading trials under two levels of audience complexity” • Investigator details (degree, affiliation, contact) • Agency to which being submitted
Abstract • 100-200 words • Summary of proposed research • Budget: Equipment, subject payments, salaries, travel, consultants, administrative costs. • Biography: Education, skills, abilities. Resume, pilot studies, completed investigations, existing grants.
Research protocol • Specific aims: Objectives, rationale • Background and significance: Existing knowledge, Priliminary work, Theoretical and practical consequences, Clinical applications. • Methods: Sampling, subjects, measurements, procedures, data analysis. • Ethical issues: IRB, human/animal subjects protection
References • Appropriate citations • Watson, Jensen, Smith, and Kelly (1987) • Watson et al. (1987) • (Miller, 1991) • (Miller, 1991; Killion, 1992) • Complete references. • Author (year). Title. Ear and Hearing, 36, 776-781.
Writing • Concise, clear, efficient, effective • “ The results of the study indicated that when faced with hyperactive children with short attention spans (AD-HD) who were clearly unmotivated and who eventually became unresponsive to test stimuli because of their distractibility, clinicians were prone to modify their instructional techniques by terminating reinforcement followed by the introduction of a “time out” phase, during which time the training program was suspended for a period of few minutes or more.” • When the rate of the response being trained decreased by 10%, the clinicians suspended the training program for 5 to 10 minutes. • Resources: APA, AIP