4 Comments
⭠ Return to thread

Holden did a correlation check. He found a negative correlation. However, his methodology was incomplete. I used this approach to show him that one has to apply time shifts. The figure I refer to is the correlation coefficient versus the time delay based on my approach.

For a time delay of zero, the negative correlation was found that Holden referred to. However, the correlation coefficient was maximized for a 5 months delay.

This was merely an exercise to show Holden how to perform this type of analysis and not to jump to hasty conclusions.

Expand full comment

All are using the time of the 2nd dose right? So there would be a negative time shift if a person dies after 1st dose, but before 2nd?

Expand full comment

I used the second dose. For all doses, a positive correlation coefficient is found for certain time shifts. However, because monthly data was used, the significance of the correlation coefficient for the boosters was not sufficiently high enough. As indicated, I would have preferred weekly data, plus a different method. Unfortunately I do not have a lot of time.

Expand full comment

Thank you!

Expand full comment