Effect Size


Less is more. Stay pure. Stay poor.
You need to provide more information in order for use to provide pointed feedback. Since you could have a single value, before and after data, or possibly a time series. We just don't know.

My sample size is 70. I have a series of values for time to haemostasis. I do not have pre and post data. The samples is analysed as mean and standard deviation.
Let me know if more information is required.


Less is more. Stay pure. Stay poor.
Are the values for the same person or a sample of individuals?

An effect size is a measure of an effect (e.g., intervention, policy change, etc.). Do you have another variable in this dataset? Effect sizes can be measures for correlation, rate ratios or differences, hazards, etc. In these scenarios, there is a another variable. I am not familiar with an effect size of a univariate dataset. You can do a one-sample test, but I am not imagining that is what you are looking for. Why do you think you need an effect size? You can present a metric for time with a precision value for it. But not sure that would be an "effect size".

What is the background and context for your inquiry?
Research Question: How much time does it take to stop bleeding (Haemostasis) after transracial catheterisation.

Study product: Chitosan dressing used to stop bleeding from puncture site

Method : Chitosan dressing was applied on the puncture site to stop bleeding.

Sample Size: 70

I have data for each patient:
Patient 1 : time to stop bleeding = 4 min
Patient 2: time to stop bleeding = 7 min
Patient 3: ...
Patient 70: time to stop bleeding = 5 min

Mean time to stop bleeding = 5.43 +/- 1.36 minutes.

I want to conclude that this mean time to stop bleeding is lesser as compared to standard ( as per literature standard time tostados top bleeding is 10 minutes.

I need an effect size to write in the article.


Less is more. Stay pure. Stay poor.
Thanks, this helps. The big question would be the comparator is 10 mins, but that is an estimate without a dispersion metric. So it is 10 minutes give or take how many minutes? Also before conducting the experiment did you hypothesis a one-way or two-way confidence interval. Lastly, was this a superiority, non-inferiority, or equivalence analysis. Also, how comparable was your sample to the sample the 10 minute estimate came from. If it came from subjects with thrombotic issues or differences, your outcome could be related to sample differences, not just products differences.

If I was right this up, I would just present your value (given it is normally distributed) with 95% and 99% CI's and that is it. If you want you could plot your dot whisker plot for your values and add a reference line for 10 minutes along with reference shading to show where the 10 minutes estimates 95% and 99% CI's would land on the plot.



Less is more. Stay pure. Stay poor.
I guess my stance is that if the two products are not being directly compared (randomized to subjects within the same sample) it would be inappropriate to collect an estimate of their differences in time. You should report your time in the Results section and in the Discussion section bring in the information about the other product. If you never used the other product you don't have original data on it for your Results sections. You could do a one-sample t-test where you compare your value to a constant (e.g., 10 minutes), but using that constant would not include the variability in the original 10 minute estimate and that its value is conditional on the sample it came from.

If I was peer-reviewing your manuscript and you compared your value to a single constant from another study, I would tell you not to do that and to revise and resubmit without that component in the analyses. You don't need an effect estimate in this setting and it would be inappropriate to present one!
Noted with thanks.

I require assistance with another study as well.
Please let me know if you can help.
I will write in another post.