In the context of association between attributes, the expected frequency is calculated assuming independence. If the observed frequency differs significantly from the expected frequency, an association is suggested. The provided answer choice 'Less than observed' is contextually incomplete as association can occur when observed frequencies are either greater or less than expected. This answer is provided as per the source key.

While 'tracheotomy' is a medical term, the source identifies it as the answer for dividing a population into three classes. In standard statistics, dividing a population into two classes is called dichotomy, and three classes would typically be called trichotomy. The provided answer is accepted as per the source, though it appears to be a terminology error.

In the study of association of attributes, two attributes A and B are considered independent if (AB) = (A)(B)/n. If (AB) > (A)(B)/n, the observed frequency of the joint occurrence is greater than what would be expected under independence, indicating a positive association between the two attributes.

A coefficient of association of +1 indicates a perfect positive linear relationship. This means that as one variable increases, the other variable increases proportionally. It represents the maximum possible value for a positive correlation, signifying that all data points lie exactly on a straight line with a positive slope.