An Item Respons Theory Analysis of the Parental Distress Subscale

Purpose: Several studies have been done assessing the factor structure of the Parental Distress Subscale. If we plan to use this subscale as an assessment tool to help identify those in need of services, more information is needed on the ways in which the items are better able to discriminate at various levels of stress and distress. This would allow for the refinement of the measure and potentially the development of a shorter measure to be use in clinical practice. The purpose of this study is to use Item Response Theory (IRT) to examine the discriminability of the items associated with the parenting demands stress and general distress subscales, which have been identified as two of the Parental Distress subscales (McKelvey et al. 2009).

Methods: Using data from the Fragile Family Child Wellbeing Study, an IRT analysis was conducted to assess the discriminability of the items associated with the parenting demands stress and general distress subscales. The sample consisted of 2,655 mothers who had complete data at wave three. An IRT was conducted using MPLUS. Two separate IRT analyses were conducted—one for each subscale. For the purpose of the IRT analysis the items were dichotomized, where 1= strongly agree and agree, and 0 = not sure, disagree, and strongly disagree.

Results: As for the general distress subscale, the results of the 2 parameter logistic IRT model was significant x2(113, N = 2,655) =175.71, p < .001. Examination of the item characteristic curves, range of item difficulty (.89 to 2.57) and item discrimination (.05 to.17) provided evidence that the seven items making up the general distress subscale all have distinguishable relationships to the latent trait being measured. The questions which asked the respondents to indicate do they feel alone and without friends and when you go to a party, you usually expect to have a bad time discriminated at high levels of general distress. It should be noted that the latter was the hardest question. Whereas, the question that asked the respondents how often they have feelings that they cannot handle things very well discriminated well at low levels of general distress. For the parenting demands stress subscale, the results of the 2 parameter logistic IRT model was significant x2(21, N = 2,655) =48.09, p < .001. Examination of the item characteristic curves, range of item difficulty (.05 to 2.75) and item discrimination (.83 to.17) provided evidence that the five items making up the parenting demands stress subscale all have distinguishable relationships to the latent trait being measured. The question that asked respondents do they find themselves giving up more of their lives to meet child's needs than expected discriminated well at low levels of parenting stress. Whereas, the question that asked the respondents having a child has caused more problems than expected in relationship with men discriminated at the high levels of parenting stress.

Implications for Clinical Practice: These subscales are important tools for social work practice.