Relative risk and odds ratios: both compare the rate in one group to the rate in another. Relative risk is easier to understand. Odds ratios are standard in many analyses but can be misinterpreted. Adding this to the thread rather than to the wiki, because I am not confident enough for the wiki.
Measurement error in home scales, with a worked standard deviation Writing it up because I had to work it out twice and would rather nobody else did.
What changes if the standard account of measurement error in home scales is wrong? I ask because I have been treating it as settled and I noticed this week that I could not say why.
Working through the consequences rather than the evidence, since others here are better placed on the evidence.
Adding the measurement that the opening post says would settle it.
Number needed to treat: how many people need to be treated to prevent one bad outcome or achieve one good outcome. More intuitive than relative risk reduction.
A partial answer, offered because a partial answer beats none.
Agreed on measurement error in home scales, with one qualification that I think matters. The reasoning holds for the case as described. Change the starting assumption and it does not, and the starting assumption is the part nobody states.
Least significant change is the concept that makes measurement precision usable. Below it, a difference between two readings is not distinguishable from noise.
The short version is the first sentence; the rest is why.
What I would want before treating measurement error in home scales as settled: the method, the sample, and whether anyone tried to find the opposite result. Two of the three are usually missing.
Relative risk and odds ratios: both compare the rate in one group to the rate in another. Relative risk is easier to understand. Odds ratios are standard in many analyses but can be misinterpreted.
Adding this to the thread rather than to the wiki, because I am not confident enough for the wiki.
Confirming post #5 from a second method, which matters more than confirming it from a second person.
Regression to the mean: if you select people with extreme values (very high or very low), their next measurement is often less extreme just by chance. This can look like a treatment effect when it is just statistics.
I would put this at better than even and not much better.
The most common statistical error in this community is not technical: it is treating a self-selected collection of reports as a sample from a population.
That is the practical version. The rigorous version is longer and says the same thing.
Two questions I would want answered before drawing anything from the measurement error in home scales data above: how were the cases selected, and what happened to the ones that dropped out.
Post #10 is right about the mechanism and I think understates the practical bit.
On measurement error in home scales the community has more anecdote than the confidence in this thread implies, and I include my own contribution in that.
Regression to the mean: if you select people with extreme values (very high or very low), their next measurement is often less extreme just by chance. This can look like a treatment effect when it is just statistics.
I am reporting what happened, not recommending it.
Relative risk and odds ratios: both compare the rate in one group to the rate in another. Relative risk is easier to understand. Odds ratios are standard in many analyses but can be misinterpreted.
Someone should write this up properly, and it should probably not be me.
Confirming post #14 from a second method, which matters more than confirming it from a second person.
Posting my measurement error in home scales numbers with the method attached so they can be discounted properly. Uncontrolled, unblinded, and collected by someone who wanted a particular answer.
Building on post #17 rather than restating it.
One caution on measurement error in home scales: everything above assumes the underlying documentation is what it claims to be. That assumption is doing real work and is rarely stated.
Time-to-event analysis handles differing follow-up in a way a simple proportion cannot, which is why event rates and Kaplan-Meier estimates can differ.
This is the version I would want a new member to read first.
Picking up post #21: that is the part I would want checked first.
Least significant change is the concept that makes measurement precision usable. Below it, a difference between two readings is not distinguishable from noise.
The most common statistical error in this community is not technical: it is treating a self-selected collection of reports as a sample from a population.
Adding a data point of agreement rather than a data point.
This follows post #25 rather than contradicting it.
Confidence intervals: rather than a single point estimate, a range of plausible values. A narrow interval means precise measurement; a wide interval means measurement is imprecise. Wider intervals (more uncertainty) are honest about limitation.
Posting it because the silence on this was starting to look like agreement.
Everything in post #25 holds. The case it does not cover is the one I have.
Confounding: a third variable explains an apparent association. In randomised data, randomisation balances confounders. In observational data, confounders can be adjusted for but unknown ones cannot.
Narrowing post #29, because the general version has more than one answer.
Confidence intervals: rather than a single point estimate, a range of plausible values. A narrow interval means precise measurement; a wide interval means measurement is imprecise. Wider intervals (more uncertainty) are honest about limitation.
It is a small point and it changes the answer, which is an awkward combination.