hello all,
I am using g power to calculate sample size. the research will be correlational study under cross sectional design. and my sample is young men and women. I am new to g power. which test family and statistical test should i use to calculate sample size? i am way consfused.
It depends on the type of correlation coefficient that you plan to use. The typical one for continuous variables would be Pearson's product-moment correlation (r). For that, you could use the option "Correlation: Bivariate normal model" under "Test family: Exact"
Do you want to check if the correlation (or slope in regression) is different for men and women, since you explicitly mentioned it. If "yes" maybe a moderation analysis would be more appropriate than comparing correlation coefficients.
Rainer Duesing I am new to g power so i might not be able to get any points. but let me mention that my hypothesis is directional that stating that young men would be more likely to be satisfied with their relationships than young women.
Christian Geiser I have no idea which type of correlation coefficient I'd use because this is my first research and i don't know about g power from the scratch
It sounds like you would be testing for mean differences in relationship satisfaction between males and females (assuming "relationship satisfaction" is a continuous/metrical/interval-scale variable). Although this could be expressed in terms of a point biserial correlation (between the continuous [relationship satisfaction] and the binary [gender] variable), it is more common to use an independent samples t test and compare the means across groups. Both options are equivalent statistically and are available in G*Power under t tests:
- "Correlation: Point biserial model"
- "Difference between two independent means (two groups)"
Christian Geiser Okay. what should I put in allocation ratio and power (1-beta error prob)?
You have not mentioned any things about population size and its nature. So it is difficult to suggest your sample size determination exactly. If the population is very large and heterogeneous in case stratified random sampling method is applied to select units of the population in the sample. If the population is not large simple random sampling method is used to select units of population in the sample. There are other different sampling techniques for selecting units of population in the sample.
Sample size determination depends upon the margin of error and confidence level so there are different methods of determination of sample size according to your margin of error and confidence level and a population size-appropriate method should be chosen. For g power, the following statistical tests are generally used to analyze the statistical data.
For many, different t-tests, F-tests, chi-square tests, z-tests, and certain exact tests, the tool G*Power can compute statistical power analyses. Additionally, G*Power can be used to calculate effect sizes and graphically present the outcomes of power analyses.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running