This post has been de-listed
It is no longer included in search results and normal feeds (front page, hot posts, subreddit posts, etc). It remains visible only via the author's post history.
Hello. I'm from Germany (please excuse my english) and I have an assignment due in 9 days about a retrospective cohort study. Normally, I wouldn't want to bother you, but my supervisor has been ignoring me for 3 weeks straight, so I thought I could get some help here. I'm also terrible at statistics.
I have a cohort of patients who have had a spinal implant. I have divided my cohort into two groups: one group that received an spinal implant for the first time, and another group that had it replaced because the old implant was loose (=revision surgery).
I want to investigate whether the mean value for age, BMI and two other numerical variables is higher in the group that received revision surgery than the group that received the implant for the first time. Is the t-test appropriate for this? Would it be professional to perform a t-test for each individual variable (=4 T-tests), or is this too much?
Thank you so much.
Subreddit
Post Details
- Posted
- 4 months ago
- Reddit URL
- View post on reddit.com
- External URL
- reddit.com/r/AskStatisti...