. Results
Fig. 1 shows the adjusted mean estimates, and twice their standard errors (i.e., 95% confidence interval) of the percentage of time spent looking at the novel stimulus in each pair. Each pair of bars represents the two groups as defined by familiarization order. These means suffice to capture the major findings of the study, which were that: (a) children's object discrimination was intact (with the exception that they were unable to discriminate spoon from shoe when first familiarized to shoe; see below for elaboration) and (b) children in both groups showed intact discrimination of most expression pairs, except when first familiarized to fear (again, see below for elaboration). Table 1 provides F-statistics, p-values and estimated effect sizes for the main effects of care group, familiarization group and gender. Five of the six ANCOVAs (all except neutral-fear) yielded significant main effects of familiarization group, yet only two of the other 42 inter-subject main effects and their interactions were significant, and those only at p < .05. Since neither of these effects was hypothesized in advance, and since they were not supported by other similar effects, we would minimize their importance to our findings. The repeated measures effect deriving from the three trials of the emotion comparisons did not approach significant, nor did its interactions with the inter-subject effects.
Estimated percent of time infants spent looking at novel objects or faces. ...
Fig. 1.
Estimated percent of time infants spent looking at novel objects or faces. Sample sizes are reported in Table 1; bars represent 95% confidence intervals.
Figure options
Table 1.
Significance tests of main effects in ANCOVAs
Care group Stimulus order Gender
Shoe-Spoon (N = 128; d.f. = 1,119)
n 89, 39 65, 63 60, 68
F .32 23.84 5.54
p .57 <.001 .02
d .11 .95 .46
Key-Mug (N = 127; d.f. = 1,118)
n 89, 38 65, 62 59, 68
F .28 24.36 3.00
p .60 <.001 .09
d .10 .97 .34
Fear-Sad (N = 142; d.f. = 1,133)
n 100, 42 72, 70 67, 75
F .17 27.88 2.33
p .16 <.001 .13
d .17 .62 .18
Neutral-Fear (N = 142; d.f. = 1,133)
n 100, 42 72, 70 67, 75
F 1.22 3.72 .04
p .27 .06 .84
d .18 .21 .02
Happy-Sad (N = 142; d.f. = 1,133)
n 100, 42 72, 70 67, 75
F 1.23 7.08 3.17
p .27 .009 .08
d .12 .29 .19
Happy-Fear (N = 142; d.f. = 1,133)
n 100, 42 72, 70 67, 75
F 1.27 16.70 .17
p .26 <.001 .68
d .11 .40 .04
Notes. (1) Computations done with maximum likelihood estimation (see text). (2) d = effect size = (difference between means)/(standard deviation). (3) All means estimated after adjusting for age, although age is not statistically significant in any of the ANCOVAS, p > .25 for all six analyses.
Table options
2.1. Object comparisons
Both the Spoon-Shoe and Key-Mug dyads showed a highly significant effect between care group and familiarization group. As may be seen in Fig. 1, when first familiarized to the spoon, the percentage of time spent looking at the shoe (the novel stimulus) was greater than that spent looking at the spoon (the familiar stimulus; Shoe = 61% versus Spoon = 39%), but neither of the objects dominated after familiarization to the shoe (Spoon = 51% versus Shoe = 49%). Familiarization to either the key or the mug produced significant novelty preference, although it was somewhat stronger after familiarization to the mug (p < .0001). These findings were consistent across the two groups of children. There was no significant interaction between group and familiarization. In the Spoon-Shoe trial, there was a suggestion of a main effect of gender (p = .02); girls looked at the shoe more often than did boys (58% versus 52%). There were no other significant main effects or interactions involving group or gender. Age, the covariate, was not a significant effect, although the means we report in Fig. 1 are adjusted for age.
2.2. Emotion comparisons
The Fear-Sad, Happy-Sad, and Happy-Fear dyads all showed significant effects of familiarization. As may be seen in Fig. 1, familiarization to Sad in the Sad-Fear dyad elicited a novelty response to Fear (the novel stimulus; 59% versus 41%), but familiarization to Fear did not produce a difference in response (Fear = 49% versus Sad = 51%); these responses were significant (p < .0001). There was a modest (p = .018) three factor interaction, possibly arising from a smaller difference between boys and girls familiarized to Sad in the Institutionalized group than between boys and girls familiarized to Sad in the Community group.
In the Happy-Sad condition there was significantly smaller percentage of looking directed toward the familiar stimulus for Sad than for Happy (46% versus 49%, p = .009).
In the Happy-Fear comparison there was a greater novelty reaction to fear among those familiarized to Happy than among those familiarized to Fear (55% versus 52%, p < .0001). No other main effects or interactions were found to be significant.
2.3. Comment on results
The absence of significant differences between institutional children and community controls must be interpreted with reference to statistical power, that is, whether our study had enough subjects to detect an effect if it were truly present. Given our sample sizes of 89 and 39, respectively, we should have been able to detect (with power = .80) an effect size of about .53 standard deviations on each main-effect test. This effect would represent two nearly overlapping populations with very little to distinguish them. We therefore feel that it is quite likely that the institutional and community children are very similar in their responses on these measures.
