Power Analysis for Multiple Regression: Examples for
Dissertation Students & Researchers
A first model will be examined using multiple regression. The first model will test whether gender, ACD standard scores, and ECF raw scores (independent variables) predict higher levels of relational aggression (criterion variables). The α for the test of this model will be set at .05. To achieve power of .80 and a medium effect size (f² = .15), a sample size of 77 is required to detect a significant model (F (3,73) = 2.73).
The second model will test whether ECF raw scores (independent variable) predict relational aggression (criterion variable). The third model will test whether ECSCI raw scores (independent variable) predict physical aggression (criterion variable). The fourth model will test whether ACD standard scores (independent variable) predict relational aggression (criterion variable). The fifth model will test whether EVT standard scores (independent variable) predict physical aggression (criterion variable). The α for the test of these models will be set at .05. To achieve power of .80 and a medium effect size (f² = .15), a sample size of 55 is required to detect a significant model (F (1, 53) = 4.02).
A model will be examined using simultaneous multiple regression. The first model will determine the ability of certain variables (accuracy, compatibility, currency, flexibility, locability, meaning, reliability and the right data, training) to predict performance impact (criterion variable). The α for the test of this model will be set at .05. To achieve power of .95 and a medium effect size (f² = .15), a sample size of 138 is required to detect a significant model (F (5,132) = 2.28).
Several models will be examined using simultaneous multiple regression. The first model will test whether certain variables (age, race, marital status, level of education, income, previous of current obesity-related health problems, self-regulatory behaviors, self determination, and elf-efficacy) predict satisfaction with body. The α for the test of this model will be set at .05. To achieve power of .80 and a medium effect size (f² = .15), a sample size of 114 is required to detect a significant model (F (9,104) = 1.97).
The second model will test whether certain variables (age, race, marital status, level of education, income, previous of current obesity-related health problems, self-regulatory behaviors, self determination, and self-efficacy) predict perceived unsuccessful weight management. The α for the test of this model will be set at .05. To achieve power of .80 and a medium effect size (f² = .15), a sample size of 114 is required to detect a significant model (F (9,104) = 1.97).
From a convenience sample of approximately 200 adults from both sites it is hoped that a desired sample size of at least 114 will be achieved for the study. A power analysis was conducted to determine the number of participants needed in this study (Cohen, 1988). A model will be examined using simultaneous multiple regression. The first model will test whether certain variables (enter your 9 variables) predict the dependent/criterion variable. The α for the test of this model will be set at .05. To achieve power of .80 and a medium effect size (f² = .15), a sample size of 114 is required to detect a significant model (F (9,104) = 1.97).
|
