Instead, a representative sample is used. The larger the sample the better, obviously, because the larger the sample the less wiggle room there is going to be between the sample looked at and the entire body.
Perhaps another explanation for the margin of error in this study is noted in the abstract of the document, whereby it is acknowledged that they questionnaire looked at forensic psychologists that were retained or secured by the court as well as situations where one of the child custody parties in question retained the forensic psychologist. It is not immediately clear whether that figured into the margin of error because that would certainly affect the motives and even the outcome of the study in general but whether/how much of an impact was had was not immediately made clear.
Conclusion
There can and should also be concerns about whether the body is homogenous or not and that perhaps drives possible reasoning for using separate pools of respondents so that the margin of error can be driven as low as possible. In the end, however, a larger sample usually leads to more reliabile results and a lower margin of error but a larger sample can create issues like it did in this study.
References
Arch, M., Jarne, a., Pero, M., & Guardia, J. (2011). CHILD CUSTODY ASSESSMENT:
A FIELD SURVEY of SPANISH FORENSIC PSYCHOLOGISTS' PRACTICES. European Journal of Psychology Applied to Legal Context, 3(2), 107-128.
