Helsing K J, Comstock G W
Int J Epidemiol. 1976 Jun;5(2):125-30. doi: 10.1093/ije/5.2.125.
In a questionnaire administered to a general population sample of 1673 adults, a 14-item scale of social desirability was divided into two groups of questions, almost evenly divided between desirable and undesirable behaviour, the percentage answering 'true' was higher when the question was asked early than when it was asked late. The difference was statistically significant for five of the 11, four of them relating to socially undesirable behaviour. The results indicate that responses can be dependent on the location of questions in a questionnaire, and that the effect of question location may not be the same for groups with different personal characteristics. For this reason, it seems unwise to alter the administration or design of a questionnaire in any way if the results are to be compared directly to those of other studies, unless the altered questionnaire is standardized against the original version.
在对1673名成年人的普通人群样本进行的问卷调查中,一个由14个项目组成的社会期望量表被分为两组问题,在期望行为和不期望行为之间几乎平均分配,当问题较早提出时回答“是”的百分比高于较晚提出时。在11个问题中的5个问题上,这种差异具有统计学意义,其中4个与社会不期望行为有关。结果表明,回答可能取决于问题在问卷中的位置,并且问题位置的影响对于具有不同个人特征的群体可能不一样。因此,如果要将结果与其他研究的结果直接比较,以任何方式改变问卷的管理或设计似乎都不明智,除非修改后的问卷与原始版本进行了标准化。