There is a variety of reasons why we should look at the results of Ainsworth et al.'s (1978) longitudinal study with caution. First of all, the sample of participants (26 infant-mother pairs) was small compared to the number of variables. A sample of 26 infants is simply too small to be divided up into 3 groups and 7 subgroups. It seems that the researchers easily could have overanalyzed trivial differences among the infants and, as a result, too many subgroups were created. In order to confirm that the 7 subgroups indeed exist, it would have been helpful if a second group of participants, possibly larger in size, had yielded the same results. This is especially important because the initial measures were not derived on a priori theoretical grounds, but rather after an examination of the results.