How to Calculate Sample Size with GPower

AFTERS VIDEO: This video will show you how to calculate sample size and power for a research study. This uses an example using information (means and standard deviations) from a previously published study and G*Power. Be sure to see your lecture notes, Canvas, or stop by during office hours if you need help.

[@deepakalive]

How do we calculate sample size for prevalence study from g power? Help appreciated.

[@vz1098]

It'll be a type 1 error as there might not actually be a difference but the outnumbered sample size would resemble a difference that's not true

[@robertlachausse8135]

Yes, I type one error is most likely to occur when the sample size is too big. So for example you might say there is an effect but there’s actually not because the sample size is too big. That is why calculating the exact sample size that you need before you collect data is so important

[@vz1098]

@@robertlachausse8135 your simplification was quite helpful. Thank you

[@deepakalive]

Also what does it mean by "correlation between groups" in the section of effect size calculation

[@robertlachausse8135]

There is going to be a correlation between groups, because in this example were using a one group pretest post test design. So the pretest scores are going to be highly correlated to the post test scores, because each of the two scores come from the same subject.

[@deepakalive]

Why didn't you estimate the sample size with available effect size?

[@robertlachausse8135]

You can use ES if you like, and if you have it already calculated or you can just put in the means of the standard deviations and it calculates it for you and uses that in the sample size calculation.

[@ArjunSharma-wi3jp]

can you help with more papersl ike this

[@ArjunSharma-wi3jp]

i want to calculate effect size and confindence intervals , what should i choose in the gpower - by my understanding i should go for standard deviation pooled and then calculate cohens d ,,,,

[@ArjunSharma-wi3jp]

or shoould i first compute sample size and then use that smaple sizeto compute effect size , your help would be appreciated greately man . thanks

[@robertlachausse8135]

What type of stat test are you doing?

[@ArjunSharma-wi3jp]

@@robertlachausse8135 i am planning to do a priori and then use its sample and calculate post hoc , so i can state that if i do similar kind of study how much sample size would i need for x amount of power

[@robertlachausse8135]

@@ArjunSharma-wi3jp in Gpower, you need to specify the research design and statistical test you will do. For example, if you wanted to look at differences between men and women and the number of times they smoke cigarettes you could compare the sample mean for men to the sample mean for women using an independent samples t-test. In G power, you would specify independent samples, t-test, or it’s also called two sample t-test. And simply put in the effect size, Alpha, and power.