turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- SAS Programming
- /
- Base SAS Programming
- /
- Calculating Confidence intervals for observed/expe...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

07-31-2017 02:42 AM

I have a simplified dataset: runner name, run trial (there are 10 trials), observed time for each trial, expected time for each trial.

I sum up my observed and expected times using proc means and output a dataset.I then use the data statement to calculated a ratio for observed / expected for each runner. how do i calculate a confidence interval to see if there is a difference in o/e between my runners?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to sharonlee

07-31-2017 03:13 AM

I'm not sure if this is what you are efter, but you can request confidence limits for the mean by specifying the CLM option in the PROC MEANS statement.

An example is in the PROC MEANS documentation here

http://support.sas.com/documentation/cdl/en/proc/61895/HTML/default/viewer.htm#a002473542.htm

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to draycut

07-31-2017 05:03 PM

Hi, Thanks for responding. It's not the mean value I want per runner. I was the ratio for each runner.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to sharonlee

08-01-2017 01:28 PM

You should post some sample data, which will make it easier for people to help you.

You might consider the **difference** in the runner's time, rather than the ratio. If you have several runners and you want to compare their mean times and determine if the mean times are significantly different, you can use PROC GLM to run an ANOVA and a multiple comparison of means. There are several examples in the PROC GLM doc, including one that treats repeated observations.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Rick_SAS

09-14-2017 02:55 PM

My sample data look like this:

runner_name | run trial | observed time (minutes) | expected time (minutes) |

name1 | 1 | 10 | 15 |

name1 | 2 | 20 | 32 |

name2 | 1 | 25 | 45 |

name2 | 2 | 30 | 7 |

name3 | 1 | 45 | 5 |

name3 | 2 | 3 | 8 |

name3 | 3 | 51 | 9 |

name3 | 4 | 87 | 1 |

name3 | 5 | 12 | 12 |

I sum up my observed and expected times using proc means and output a dataset.I then use the data statement to calculated a ratio for observed / expected for each runner.

I want to see if my overall observed/expected ratio is significantly different between runners. I'd like to generate 95% confidence intervals. Can I do this in SAS? This is not a binomial proportion so i can't use proc freq to calculate CIs.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

Posted in reply to sharonlee

09-14-2017 03:35 PM

Given the irregular values in your table of expected run times, I presume your runners are not all running the same set of distances, ... or are otherwise not subjected to the same set of running conditions. So I can see why you are considering O/E instead of O-E. But maybe if O is a predictor var, you can still use O-E as the dependent var.