The event-related potential (ERP) data reported here in 2005 by Parker, Nelson, and The Bucharest Early Intervention Project Core Group were collected in Bucharest, Romania, between April 2001 and November 2003. We recently discovered that during a 4-month period (July–October 2001) the channel for the right mastoid electrode (M2) was mistakenly exchanged with the channel for the right occipital electrode (O2) at the connection to the amplifier. This Erratum presents the key comparisons between the results as originally reported and as reanalyzed given the corrected data.
As described in the original paper, the data were collected from infants and toddlers residing in institutions (institutionalized group [IG]) or living with their families of origin in the local Bucharest community (never institutionalized group [NIG]). Because data were collected at different times for the two groups, the instrumentation problem affected data primarily from the NIG group. Analyses were completed on the new data set in the same manner as reported in the original paper. The reanalysis included 59 IG participants (30 younger IG, 29 older IG) and 29 NIG participants (15 younger NIG, 14 older NIG). For three of the ERP components (early negative [N170], early positive [P250], midlatency negative [Nc]), repeated measures analyses of variance (ANOVAs) were computed using peak amplitude and peak latency as the dependent measures; the positive slow wave (PSW) was analyzed using average amplitude as the dependent variable. For each ANOVA, the within-subject factors were electrode (Fz, Cz, Pz for midline electrode analyses) or region (frontal, central, parietal, temporal for lateral electrode analyses) and condition (caregiver's face, stranger's face). The between-subject factors were group (IG, NIG) and age (older, younger). For all but the P250, the amplitude of these components was larger in the NIG than the IG; this pattern was reversed for the P250. Typical effects of the Nc (i.e., amplitude greater to stranger vs. caregiver) were observed in both groups; in contrast, the IG group showed an atypical pattern in the PSW.
Our reanalysis of the new data set is reproduced in full below. As can be seen, the vast majority of the effects we originally reported have been replicated in full.
N170
Midline electrode sites
Amplitude
We originally reported main effects of group and electrode, with the main finding that the NIG evinced larger amplitude N170s than IG. This effect was replicated in full.
Latency
No significant effects for N170 latency at midline electrodes were reported originally. This reanalysis duplicates those nonsignificant results.
Lateral electrode sites
Amplitude
We originally reported main effects of group and region on the amplitude of the N170 at lateral electrodes, as well as a number of higher order interactions. The main effects of group and region, as well as the four-way interaction of Region × Hemisphere × Group × Age, were replicated in the current reanalysis. The Region × Hemisphere × Condition × Age interaction in the original report was marginally replicated in the reanalysis (p = .07).
Latency
A main effect of hemisphere was originally found, which did not replicate in the current reanalysis; also reported were a main effect of region and an interaction between region and group. The former was replicated in full and the latter was marginally replicated (p < .06). In addition, our reanalysis revealed a main effect of group that was not originally reported, F (1, 84) = 4.48, p < .05, in which the NIG showed significantly faster latency to the N170 than the IG.
P250
Midline electrode sites
Amplitude
We originally reported main effects of group and electrode for the P250 amplitude over midline leads, with the main finding being that the IG showed larger P250 amplitudes than the NIG. Both of these main effects were reproduced in the reanalysis.
Latency
For the P250 latency over midline leads, we originally reported a main effect of electrode, which was not replicated. We also reported an Electrode × Age interaction, which was reproduced in the reanalysis. In addition, the reanalysis revealed a Condition × Group interaction not originally reported, F (1, 84) = 4.51, p < .05.
Lateral electrode sites
Amplitude
We originally reported were main effects of both group and region for the P250 amplitude over lateral leads, which both fully replicated. In addition, a Region × Hemisphere × Group × Age interaction that we originally reported was partially replicated in the reanalysis as a Region × Hemisphere × Group interaction, F (3, 82) = 3.06, p < .05.
Latency
A main effect of hemisphere for the P250 latency that we originally reported was replicated; specifically, the latency to P250 was faster over the right than the left hemisphere. However, a main effect of region and a Region × Age interaction was not replicated.
NC
Midline electrode sites
Amplitude
For the Nc amplitude over midline leads, we originally reported main effects of electrode and condition, as well as an Electrode × Age × Group interaction, all of which were reproduced. In addition, our reanalysis revealed a new main effect of group, F (1, 81) = 4.69, p < .05, whereby the NIG exhibited larger Nc amplitude overall compared to the IG. A new Condition × Age interaction was also obtained, F (1, 84) = 4.12, p < .05.
Latency
For the Nc latency over midline leads, we originally reported a main effect of group that was replicated and an Electrode × Group interaction that marginally replicated (p < .07).
Lateral electrode sites
Amplitude
For the Nc amplitude over lateral leads, our original report was main effects of group, condition, and region, and a Region × Hemisphere interaction, all of which were replicated in the reanalysis. In addition, two new interactions emerged in the reanalysis: a Region × Hemisphere interaction, F (3, 82) = 11.58, p < .001, and a Region × Age × Group interaction, F (3, 82) = 3.6, p < .05.
Latency
We originally reported main effects of group and region, and a Region × Age × Group interaction, all of which were replicated in full. We did not replicate a Condition × Age interaction. In addition, the reanalysis revealed a new main effect of condition, which was further qualified by a Hemisphere × Condition × Group interaction, F (1, 84) = 4.63, p < .05.
PSW
Midline electrode sites
Amplitude
The main effects of electrode and group were replicated from the original report for the average amplitude of the PSW. However, the Group × Condition interaction was not obtained in the reanalysis.
Lateral electrode sites
Amplitude
We originally reported main effects of region, group, and hemisphere. The reanalysis replicated the region and group effects but failed to replicate a significant effect of hemisphere.
Summary
On the whole, our reanalysis of the data largely replicated our original findings while simultaneously revealing a number of new effects. It is important to stress that, although there were some differences across the analyses (old vs. new), the main finding of lower ERP amplitudes for certain components of the IG compared to the NIG was replicated. Accordingly, it continues to be the case that institutionalized children show markedly reduced brain activity compared to children residing with their biological families.
