When you say some messages increased support by X%, I worry that people might interpret it as *percentage points*.
i.e. a 10 percent increase from 15% is 16.5%, a 10 percentage point increase from 15% is 25%, which are obviously big things!
I generally find that when we're talking changes from 1 percent to another, people generally assume percentage point changes, rather than percent. Might be good to clarify this in the post!
Thanks for the feedback, and we're really glad to hear you appreciate the work.
These were all percentage point increases - as you said, people usually assume percentage point changes. We've definitely taken it on board though, and will be more explicit about which type of percentage we are talking about in the future!
Yes, these are quite exciting and promising results! We should keep in mind the caveat that social desireability bias will be at play - the tendency to show greater support for the issue in order to 'look good' to the researchers. So the effects might be somewhat inflated by that, but it never-the-less gives us an idea about which messages work, and an upper range for how effective they might potentially be out in the wild.
Amazing work! Can't wait for the full report.
When you say some messages increased support by X%, I worry that people might interpret it as *percentage points*.
i.e. a 10 percent increase from 15% is 16.5%, a 10 percentage point increase from 15% is 25%, which are obviously big things!
I generally find that when we're talking changes from 1 percent to another, people generally assume percentage point changes, rather than percent. Might be good to clarify this in the post!
Thanks for the feedback, and we're really glad to hear you appreciate the work.
These were all percentage point increases - as you said, people usually assume percentage point changes. We've definitely taken it on board though, and will be more explicit about which type of percentage we are talking about in the future!
Oh wow, if these are percentage point changes, that makes these findings quite dramatic!?
Yes, these are quite exciting and promising results! We should keep in mind the caveat that social desireability bias will be at play - the tendency to show greater support for the issue in order to 'look good' to the researchers. So the effects might be somewhat inflated by that, but it never-the-less gives us an idea about which messages work, and an upper range for how effective they might potentially be out in the wild.
Could you post a link to the study, please? Thanks for your important work.
The report isn't publicly available yet, but we are hoping to publish all of our research to date by the end of the year.