You say “these are very wide error bars”, in reference to the plot of Ct value versus vaccination status, hence casting doubt on the idea that the vaccinated and unvaccinated have similar Ct values.
But I think the boxplots shown represent the distribution of values, NOT the uncertainty in the estimated mean. With n=84 and n=127 for the two groups, the standard error in the difference in the two means should be around 0.5 (an exact figure isn’t obtainable from the displayed data, which shows quantiles, not standard deviations). Whether that’s “small” or not is of course a subjective judgement, but it’s a lot smaller than the size of the boxes.
However, a serious problem with this plot is that the unvaccinated are lumped together with “unknown” and “not fully vaccinated”. Since unknowns may well be vaccinated, and the partially-vaccinated have a large part of the immunity of the fully-vaccinated, the plot is basically useless.
You say “these are very wide error bars”, in reference to the plot of Ct value versus vaccination status, hence casting doubt on the idea that the vaccinated and unvaccinated have similar Ct values.
But I think the boxplots shown represent the distribution of values, NOT the uncertainty in the estimated mean. With n=84 and n=127 for the two groups, the standard error in the difference in the two means should be around 0.5 (an exact figure isn’t obtainable from the displayed data, which shows quantiles, not standard deviations). Whether that’s “small” or not is of course a subjective judgement, but it’s a lot smaller than the size of the boxes.
However, a serious problem with this plot is that the unvaccinated are lumped together with “unknown” and “not fully vaccinated”. Since unknowns may well be vaccinated, and the partially-vaccinated have a large part of the immunity of the fully-vaccinated, the plot is basically useless.
Another problem is that Ct values are a logarithmic function of viral load. Increasing the Ct value by 1 corresponds to halving the viral load.