AAP Grand Rounds doesn't often review articles from social science journals. The format of the article is a bit different, but the story the authors tell us is revealing.
Source: Julian MM, McCall RB. Social skills in children adopted from socially-emotionally depriving institutions. Adopt Q. 2016;19(1):44-62; doi:10.1080/10926755.2015.1088106. See AAP Grand Rounds commentary by Dr. Emily Todd (subscription required).
PICO Question: Among children adopted from institutions, does age of adoption, age of assessment, or gender impact social skills?
Question type: Descriptive
Study design: Survey
The psychologist authors of this study obtained names of parents adopting children from a single local adoption agency specializing in placement of Russian children. Then they obtained parent-completed survey information about their adopted child's social skills and behaviors using standardized rating tools. They ended up with 214 children and 127 adolescents with completed surveys. Most of the children were adopted from Russian "Baby Homes" for children under the age of 4 years, with the rest coming from similar settings in Eastern Europe. They found that children adopted before 18 months of age showed higher social skills than those adopted later. Girls had more problems than boys, especially those adopted at later ages.
Articles published in social science journals tend to have slight differences compared to medical science publications. In particular, the former tend to read more like stories (in my opinion), with the divisions between the Methods and Results sections blurred; often details about methods are found in the Results section, and vice versa. This isn't necessarily bad, just a little difficult for those of us accustomed to finding information in specific parts of an article.
This article also is a good time to review a couple key points when reading articles using surveys. First, the response rate is important. For any study, seeing a poor survey completion rate, or a large dropout rate, brings the issue of bias into play. Parents who chose not to complete the survey might have very different experiences than those who did, and thus we might miss a significant finding. In this study, the response rate was 42%, which isn't great from a statistical sampling viewpoint, but not unusual in survey studies.
Second, questionnaire responses may differ from real life situations. Parents may be subtly biased to portray their children either better or worse than their true behaviors. The only way to counteract this is to have some objective measure (e.g. psychological examination of each child by an objective observer) to complement the survey data. Of course, that would be a much different, and more difficult, study.