G*Power: โปรแกรมทางเลือกส าหรับค านวณขนาดตัวอย่าง

1 G* power : Kittisakdi Choomalee, Supeecha Rungruang: Holistic Health and Safety Management in the Community Research Unit (HASM), PSU application n4studies smartphone/ tablet i OS Android G* power Application Application OS X Windows G* power Application n4studies Download G* power Heinrich-Heine-Universit t D sseldorf Download Windows version OS X Version G* power 1. Tests 2. 3.

2 Logistic regression Test family 1. Test family family (Distribution-based approach) Logistic regression Logistic regression family Z tests 2. Logistic regression 3. (A Priori: Compute required sample size given , power and effect size) 4. review ( Odds ratio, alpha, power , event rate H0 (Null hypothesis) (x) binomial (x parm = balanced design ) ( ) ( ) (x) (normal distribution)) alpha ( ) power ( power = 1- ) Type I error Type II error 5% 20% R default 0 ( covariates covariates correlation covariates ) 5.

3 Calculate 1. options 2. effect size Odds ratio probability (event rate ) 3. power 2 enumeration procedure for the Wald-test and the Likelihood ratio test (Lyles et al. (2007)) large sample approximations for a Wald-type test enumeration power large sample approximations 2 Demidenko (2007) Hsieh et al. (1998) Hsieh et al. (1998) (x) binomial Normal Distribution Demidenko (2007) Binomial Normal Distribution Demidenko (2007) power Hsieh et al. (1998) variance enumeration Monte-Carlo simulations accuracy large sample approximation enumeration accuracy N >200 4.

4 Ok event rate Odds ratio 1. Determine 2. event rate (H0 H1) Odds ratio risk factor protective factor 3. calculate 4. calculate and transfer to main window Odds ratio Pearson s correlation H0: r = 0, H0: =0 H1: = c, H1: r = c c 0 Cohen, 1977 Effect size r c H1 (conventions - r ) = -small = -medium = - large 1. Test family Exact 2.

5 statistical test Correlation: Bivariate normal model 3. Type of power analysis A prio: Compute required sample size given , power and effect size 4. effect size (medium) err prop = power = (H0: ) 0 5. calculate Pearson's correlation r Pearson (small letter rho) default G* power = / =1 ( = 1) = power = (1- , 1 - = ) Chi-square (contingency table) Chi-square test 2 Goodness of fit test , Independence test (association) macular degeneration (AMD)

6 Chi-square Degree of Freedom (df) df= (r-1)*(c-1) r= c= AMD df= (2-1)*(2-1) =1 Effect size (w) w effect size Cohen(1977) conventional effect size -> small -> medium -> large Chi-square 1. Test family 2 test 2. statistical test Goodness-of-fit-test: Contingency tables 3. Type of power analysis A priori: Compute required sample size given , power and effect size 4. (effect size, , power , df) 5. calculate Chi-square Point biserial correlation (rpb) Point biserial Correlation Pearson s product moment correlation ((interval or ratio scale) 2 (dichotomous) 0 1 Point biserial correlation Effect size (| |) conventions effect size (Cohen(1977)) -> small -> medium -> large Alpha ( ) power (1-b) (tail(s)) ,one tail/ two tail 1.)

7 Test family t tests 2. statistical test Correlation: point biserial model 3. Type of power analysis A Priori: Compute required sample size, given , power , and effect size 4. 5. calculate Point biserial correlation Faul, F., Erdfelder, E., Lang, , & Buchner, A. (2007). G* power 3: A flexible statistical power analysis program for the social, behavioral , and biomedical sciences . Behavior Research Methods, 39, 175-191. Download PDF Faul, F., Erdfelder, E., Buchner, A., & Lang, (2009). statistical power analyses using G* power : Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149-1160. Download PDF family test Logistic regression Linear regression Pearson s correlation Cohen, J.

8 (1988). statistical power analysis for the behavioral sciences . Hillsdale, New Jersey: Lawrence Erlbaum Associates. Demidenko, E. (2007). Sample size determination for logistic regression revisited. Statistics in Medicine, 26,3385-3397. Hsieh, F. Y., Bloch, D. A., & Larsen, M. D. (1998). A simple method of sample size calculation for linear and logistic regression. Statistics in Medicine, 17, 1623-1634. Lyles, R. H., Lin, , & Williamson, J. M. (2007). A practial approach to computing power for generalized linear models with nominal, count, or ordinal responses. Statistics in Medicine, 26, 1632-1648.