If i understand correctly some dialysis patients will not receive the intervention (control group) and some will (intervention group). Then you measure something.

I would suggest initially a comparison of the mean differences post-pre between the intervention and control group.

you thus do an unpaired two sided t test.

power/ sample size calculations for this are given in theattached extract..

see example 6.3 page 182

James is right, but a more powerful test would take the baseline values into account as covariates. One way of doing this is via a linear mixed effects model, where the coefficient of time-group interaction term is the formal test of the intervention.

Given an expected medium effect size (0.5), alpha=0.05, power=80%, an expected correlation between baseline and end readings of 0.6,  then 67 patients would be required in each group.

I also i have a related question. I would like to conduct a nurse-led health education programme among a set of pregnant women in a setting ane evaluate its effectiveness on improving knowledge and ultimately anemia among this set of women. I am not using a control group. I will obtain their baseline data and then measure after the intervention and then 4 weeks later among the same set of women. Please, how best can i calculate my sample size?

If you can maintain the whole group you could do paired t tests (differences between base line and post intervention value for each individual for variables studied). Then the power calculations would be as in attached.

http://www.sample-size.net/sample-size-study-paired-t-test/

Thank you for the prompt response and solution provided.

James Leigh , what is the name and author of the book you attached earlier?

Thanks in advance

Statistical Methods in Medical Research 2nd ed

Armitage P and Berry G Oxford,Blackwell 1987

I would like to conduct a pre post experimental design to the determine the effect of an intervention in treatment of hypertension and diabetes at a secondary health care facility. How do I calculate the sample size for this study. Thank you

Thank you @James Leigh... its helpfully. It doesn't matter for experimental study or quasy..

Thank you James Leigh for your recommendation. May I divided my population into two groups (control and intervention group). So each group contain 33 respondents ?

yes. then you could use a two sample t-test to test significance of mean effect measure in intervention group vs mean effect measure in control group.

the groups do not have to be of equal size but t-tests are more powerful if they are

see attached for calculation

http://sample-size.net/means-effect-size/

James Leigh Hello, I read your answer and I have another question related to this. I am carrying out a study with a quasi-experimental method applying a between-group design. I have four groups, so I would like to know how can I calculate the sample size. And which analysis should I apply to compare the groups in the end as in cause-effect of each group.

Thank you very much

The attached gives power/sample size calculations for one way ANOVA which is probably what you would use.

https://www.statskingdom.com/180Anova1way.html

How can I calculate sample size for Quasi experimental study for single group? James Leigh kindly guide