Definitely a risk of selection bias, but it's due to the poor randomization, not the allocation concealment
-
-
Replying to @GidMK @MaartenvSmeden
Allocation concealment is about what happens after randomization. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2596474/ …pic.twitter.com/fFKkuvLGjT
1 reply 0 retweets 0 likes -
Replying to @statsepi @MaartenvSmeden
I'm confused now. Surely that's blinding? Isn't allocation concealment the process of ensuring researchers/patients can't influence the method of group selection?
1 reply 0 retweets 0 likes -
Replying to @GidMK @MaartenvSmeden
And that they can't change it after, or drop people from study if they don't like the result.
2 replies 0 retweets 1 like -
Replying to @statsepi @MaartenvSmeden
Sure, but my that wouldn't change my dice roll scenario. Once the roll has been made, you have to use the result
1 reply 0 retweets 0 likes -
Replying to @GidMK @MaartenvSmeden
Darren Dahly, PhD Retweeted Darren Dahly, PhD
Darren Dahly, PhD added,
Darren Dahly, PhD @statsepiReplying to @statsepi @GidMK @MaartenvSmedenIt's about creating secure systems (i.e. not envelopes in your office) that make it virtually impossible to change the allocation (e.g. using an off-site randomization service). Blinding is the next step in maintaining all of this. You're 100% right in that it all goes together.1 reply 0 retweets 1 like -
Replying to @statsepi @MaartenvSmeden
Haha fine. It's a condition of the dice roll, recorded under observation by witnesses Even so, the randomization itself is a potential source of bias!
1 reply 0 retweets 0 likes -
Replying to @GidMK @MaartenvSmeden
Now you'll have to explain, because I'm not getting it :)
1 reply 0 retweets 0 likes -
Replying to @statsepi @MaartenvSmeden
Well, the researchers don't know the result prior to the roll, they can't change it after, but dice can still be biased. Even a slight tendency towards a specific number - undetectable to the researchers - would skew the randomization
1 reply 0 retweets 1 like -
Replying to @GidMK @MaartenvSmeden
But would it skew it systematically disproportionate to baseline risk? That said, given it's theoretically possible, it's much much easier to have a truly random allocation than it is to have strong allocation concealment procedures - but people often focus on the former.
1 reply 0 retweets 2 likes
I think it potentially could, yes. You just can't really know without analyzing the data yourself. But yeh you're definitely right about which one is easier, it takes five minutes and no effort at all to generate a random sequence but oodles off effort to conceal allocation
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.