Menentukan Jumlah Sampel Minimal Penelitian dengan G*Power

Summary

This video tutorial explains how to determine the minimal sample size required for statistical analysis using the free software GPower. The focus is on calculating sample sizes for correlation and t-tests, based on three key parameters: statistical power (typically 0.8), significance level (usually 0.05), and effect size. It provides step-by-step guidance on how to input values into GPower and interpret the results. The video also highlights the importance of ensuring that the sample is representative of the population for generalizability, not just meeting statistical requirements.

Takeaways

• 😀 G*Power is a free software tool that helps determine statistical power, effect size, and minimal sample size for various statistical analyses.
• 😀 To determine the minimum sample size, three key parameters are required: statistical power, significance level, and effect size.
• 😀 A common significance level used in psychology research is 0.05, or P < 0.05, which indicates statistical significance.
• 😀 Statistical power should generally be above 0.8 (or 80%) to ensure reliable results in research studies.
• 😀 Effect size is important for determining the sample size and can be derived from previous research or guidelines like Cohen's recommendations.
• 😀 If prior studies provide effect sizes, they can be used as a reference for the current study's calculations of minimal sample size.
• 😀 For detecting small effect sizes (e.g., correlation r = 0.1), larger sample sizes are required to achieve sufficient statistical power.
• 😀 The minimal sample size for a bivariate correlation test with a 0.2 effect size, 0.05 significance level, and 0.8 power is 153 participants.
• 😀 For a t-test comparing two independent groups, if the effect size is 0.8, an alpha of 0.05, and power of 0.8, the required total sample size is 42 (21 per group).
• 😀 A larger sample size improves the statistical power, but a smaller sample may lead to power below the desired threshold, affecting result reliability.
• 😀 It's crucial that the sample be representative, not just large enough, especially when the goal is to generalize the findings to a broader population.

Q & A

What is GPower software used for?

GPower is a free statistical software used to calculate statistical power, effect sizes, and the minimum sample size required for statistical analyses, such as T-tests, F-tests, or correlation studies.

What are the three key parameters needed to determine minimum sample size?

The three key parameters are statistical power, significance level (α), and effect size.

What does statistical power refer to in research studies?

Statistical power refers to the likelihood of detecting a true effect if one exists. A power of 0.8 or 80% is typically considered acceptable.

What is the significance level (α) commonly set at in psychology studies?

In psychology studies, the significance level is typically set at 0.05, meaning there is a 5% chance of committing a Type I error (false positive).

How is effect size determined for a study?

Effect size can be determined from previous research findings or by using guidelines such as Cohen's for various statistical tests. For example, for correlation studies, a small effect size might be considered 0.1, and for T-tests, a medium effect size could be 0.7.

How can GPower software help when the population size is unknown?

GPower can still be used to determine the minimal sample size even if the population size is unknown or cannot be generalized, making it useful for experimental studies where population data may not be available.

What is the minimal sample size needed to detect a correlation of 0.2 with 80% power and 0.05 significance level?

For a correlation of 0.2 with 80% power and a significance level of 0.05, GPower determines that a minimum sample size of 153 is needed.

How does GPower calculate sample size for an independent group T-test?

For an independent group T-test, GPower calculates the minimum sample size based on the effect size, significance level, and statistical power. For instance, with a medium effect size of 0.7, 80% power, and 0.05 significance level, the required sample size is 42, with 21 participants in each group.

What does a power of 0.8 mean in statistical analysis?

A power of 0.8 means there is an 80% probability of detecting a true effect if it exists. It is the commonly accepted threshold for ensuring that a study is adequately powered.

Why is it important for a sample to be representative of the population in research?

It is important for a sample to be representative of the population to ensure that the findings of the study can be generalized and applied to the larger population, enhancing the validity of the results.