Saturday, February 28, 2009

Minnesota Shared Parenting Report - Part IX

Assessing Social Science Research

Jeffrey L. Edleson, PhD
University of Minnesota

Dr Edleson examined four aspects of the research used in creating the studies such as those found in this report. His examination does not make an analysis of the information collected but instead assesses the methodology in which the data was collected.

What is the study about?
Divorce literature often varies on whether it is the divorce itself or parental conflict that is impacting child emotional health. Although battered mother’s mental health affects the children's mental health the intervening variable of father’s violence is often missing in studies.

How was the study conducted?
Dr Edleson asks participants to examine who was studied such as all families, high conflict, joint custody or sole custody. Where they were found is also an important element. If the study subjects were found through courts, family services or from the community could impact the results of the study.
How the subjects were selected, be it randomly or through other means, can affect the results. He noted that it is important to know how representative of the general population they are. He cautioned that different sources lead to different outcomes.
What research design was used can affect the results. Dr Edleson questioned whether the design of the study aimed to rule out other factors that may have actually caused the changes seen. Is there random assignment to differing conditions, comparisons or other controls? It is important to have controls to rule out other possible explanations.
How did participants provide information on measures by supplying records, answering questionnaires or participating in interviews by phone or in person. There should be room for unintended consequences in the study mechanisms.

What was found?
Dr Edleson first notes the importance of how the data were analyzed. Whether such things as an attempt by the authors to create a control for multiple factors was made or are there variations present in the study must be considered. For example, Joan Kelly (2007) argues that there is a continuum of need for child visitation arrangements.
Great variation among child experiences was noted. One area studied was DV research where differences within the group were noted. On average children exposed to DV showed more problems than those not exposed. However, within the exposed group, many children showed no greater problems than comparison children up to 50% in some samples. Clearly there should be a varied approach.

What do the results mean?
What the authors conclude as well as does the data support their conclusions must be examined. Within that it should be questioned whether there were alternative explanations to their conclusions and did the authors consider these alternatives.

Dr Edelson concludes by cautioning that no one study is definitive. That a valid study must be based on a representative sample and be capable of being replicated to be generalized. Finally, two stressed cautions; BE CAUTIOUS WITH CAUSAL CLAIMS (Ramsey & Kelly, 2006) and DO NOT EXPECT BLACK LETTER TRUTH FROM SOCIAL SCIENCE (Ramsey & Kelly, 2006)

No comments: