I have a spreadsheet in Excel used for a meta-analysis.
I want to create a funnel plot to check for bias. Creating the scatterplot is easy, but I don't know how to do the confidence intervals in Excel (the lines representing e.g. a triangle).
The articles I have read on this topic are not very clear on explaining exactly what these lines are, or how you add them to a plot (especially in Excel).
Attached is a picture of the scatterplot. The confidence lines are missing, and any advice on how to add and interpret them is appreciated.
On another note, I have used hedges unbiased g, and there are a small amount of huge effect sizes (such as g = 7). Should these be treated as outlier and excluded? Or would it be more prudent to convert the effect sizes to r?
Dear Espen,
Our free set of Excel workbooks might be of use for you. See www.meta-essentials.com
The workbooks generate all the standard (and more...) plots and statistics for you.
Good luck!
Henk
I'm not sure how you would do this automatically, but if you have a formula you can use to calculate where the lines should be, then you can add the values for the lines in a separate series of cells, and then add the lines to the plot as a new data series. Here's a tutorial: http://www.officetooltips.com/excel/tips/how_to_add_a_vertical_line_to_the_chart.html
Dear Espen,
Our free set of Excel workbooks might be of use for you. See www.meta-essentials.com
The workbooks generate all the standard (and more...) plots and statistics for you.
Good luck!
Henk
another alternative would be RevMan 5 from Cochrane Collaboration Group
Best regards
Hey Espen,
I know this isn't 100% what you're looking for but my R package MAVIS is a Shiny application that lets you use the metafor package (and a few others) though your web browser. Check out this online demo here http://kylehamilton.net/shiny/MAVIS/
Right now the only effect sizes it takes is either d or r but there are effect size calculators there that might be helpful.
Kyle
Thank you for all these answers. I will look into the different options and get back to you what worked out.
Dear Espen,
May I ask which tool you eventually chose to use? And why?
Best,
Henk
Dear Henk
I have not yet done this, as the project is temporarily paused while we are figuring out some details of an experiment. Once I have an answer I will post here again :)
Dear all, sorry that I have not yet got back to you about this as I am swamped with other projects. This is, however, on my to-do list, and I will post again when I get around to trying these out either later this year or early next year.
- Badges
- Science topic