2

I am reading an health research paper in which the authors are calculating the percent of males that have a certain disease, and the confidence intervals on this percentage. The authors state that:

  • They studied 20808 males
  • 67.1% of males that they studied have the disease
  • A 95% Confidence Interval of (65.8% and 68.4%)

Using the information provided and this post (Confidence interval without std?), I tried to calculate this number myself:

2* sqrt((0.671 - 0.671^2)/20808) * 100 = 0.651 %

My number comes out as exactly half of the estimates provided by the authors i.e. 68.4 - 67.1 = 67.1 - 65.8 = 2 * 0.651.

I tried to research online to see that if there might be some other formulas that can be used which will result in my calculations being equivalent to the calculations of the authors, but I could not find anything.

  • Does anyone know that if sometimes the Confidence Intervals are "arbitrarily" multiplied by 2? (perhaps for a more "conservative" estimate)

Thank you!

RobPratt
  • 50,938
stats_noob
  • 4,107
  • 1
    I don't think you're computing it correctly. Please use MathJax (see this: https://math.meta.stackexchange.com/questions/5020/mathjax-basic-tutorial-and-quick-reference) to format your math (and if you're using a calculator, put the parentheses in the right places!). – Sean Roberson Sep 04 '22 at 18:09
  • 1
    Could you link to the paper? – J.G. Sep 04 '22 at 18:19
  • 1
    @SeanRoberson The OP's calculation is of $z\sqrt{p(1-p)/n}=0.00651$. What do you think the mistake is? – J.G. Sep 04 '22 at 18:21
  • 1
    Since the ends of the confidence interval are $\hat p\pm 1.96$ standard errors, the width of the confidence interval is double $1.96$ standard errors – Henry Sep 05 '22 at 00:19
  • Thank you everyone for your replies! – stats_noob Sep 05 '22 at 01:42
  • @ J.G. I have included a link to the paper over here : https://bmchealthservres.biomedcentral.com/track/pdf/10.1186/s12913-020-4967-3.pdf .... but I will soon be removing the paper for privacy reasons. thank you so much! – stats_noob Sep 06 '22 at 04:03
  • @ Sean Roberson: thank you for your reply! I will try to do this! – stats_noob Sep 06 '22 at 04:04
  • @ Henry: Thank you for your reply! Can you please elaborate on this? thanks! – stats_noob Sep 06 '22 at 04:04

0 Answers0