It is important to note that a significant reduction in the size of an association between two variables by taking out the influence of a third variable does not necessarily mean that that reduction indicates the degree of inaccuracy in the original relationship between the two variables. An alternative possibility that should be seriously considered is that one of the two variables is an intervening or mediating variable which is influenced by the third variable and which affects the other variable (Schore, 2000).
The absence of a significant association between two measures, on the other hand, does not necessarily signify lack of causal connection between those two variables. It is possible that the relationship is suppressed or hidden by the influence of one or more other variables.
Even though Kogan and Carter (1996) conduct a very useful and interesting experiment, a problem with the still face situation, however, is that so much changes upon presentation of the still face that it is difficult to account for its effects on infant behaviour. For example, parental behaviour in a range of modalities (e.g., facial expression, looking behaviour, vocalizations, touching, head movements) decreases, the overall level of stimulation decreases, the change in parental behaviour is abrupt and unanticipated, parental behaviour is no longer responsive to that of the infant, and the parent shows a pattern of behaviour that the infant may not have seen before. Therefore, instead of supporting the arousal model, the effects of the still face could be due to expectancy violations, decreases in a single or multiple modalities, a loss of social control, the suddenness of the change, and so on. More likely, probably more than one of these processes is operating. D’Entremont and Muir, (1997) have conducted a series of studies aimed at exploring which aspects of the situation may be most influential in affecting infant behaviour. These studies provide data supporting the arousal regulation, expectancy violation, and non-contingency explanations. For example, supporting arousal regulation was the finding that if adults (mothers or experimenters) provided physical stimulation during the still face sessions (even if the adult’s hand was not visible), the infants showed more positive affect than if no physical contact was provided (D’Entremont and Muir 1997). The moderating effect of adult touch, however, was only significant for active, not passive, touch. Thus, infants who continued to receive stimulation in the form of active touch during the still face period probably experienced levels of arousal high enough that they did not become bored and distressed.