INDEX


[1]. ρ = 1 − 6ξd2
n(n2 − 1)

Resemblance equals one minus six times the sum of the differences (in rank) squared, over the number (of cases) times the number (of cases) squared, minus one.

[2]. ρ = 1 − 6ξd2
n(n2 − 1) Formula explained, opposite page.

[3]. Thompson has recently shown that a coefficient of zero does not necessarily mean absence of relationship between two factors. There might be a strong influence making for negative correlation, and at the same time an equally strong influence making for positive correlation, which might, by just counterbalancing each other, produce a spurious effect of no connection at all, namely, a coefficient of 0.00.

[4]. A curious case of negative correlation between cancellation and other tests has been reported by McCall (see references).

[5]. “Insanity” and “dementia” seem to be synonymous, as used by Hart and Spearman. But in American texts “dementia” is limited to mean intellectual deterioration.

[6]. Because of the probabilities in die-casting, every single value for red would have the same median value among the throws of yellow, which turn up in connection with it, if enough throws are made. This is not what happens in measuring mental traits. For any single value, high or low, in one function, the median of repeated measures in the other function is very different, for most traits, from the median for other values.