First the link to this week’s complete list as HTML and as PDF.
At first glance Luby et al. sounds very relevant. They go on in length about maternal support and maternal nurturance without once defining or operationalising these terms in the main text. Hidden in methods we find they are based on one single 8 minute instance in a highly artificial situation. Even accepting that single observation as representative and assuming deviations to average out, does it mean anything? Low p-values only tell us there is a high probability of there being an influence, they say nothing whatever about its size and relevance. No correlation is given and looking at figure 1 we find the predictive value of the blue line to be essentially nil and its slope to be solely determined by one or two outliers to the far left and right – without them it would tend to about zero in both subfigures. In all a meaningless non-result if there ever was one.
Williams is typical for the abysmal standard of editing the J. Arch. Sci. has sunk to. In equation one the sign of the exponent is wrong, which is totally obvious when seeing equations 1 and 2 side by side. On top of that the form of that equation is entirely out of the question. It sums a value with a dimension and a pure number, it raises a value with the dimension time to an arbitrary power, and it carries a meaningless constant. The only reasonable and dimensionally sensible form that equation could have is (τ/(t+δ))^α where δ and τ are times, α is positive, and all constants are open to a sensible and meaningful interpretation. And why does it start with the arbitrary and meaningless value of 128 at t=0 instead of something slightly less than 1 or 100 %?