Ensuring content validity: an illustration of the process.

Assessing content validity is one of the most critical steps in instrument development. Neither statistical elegance nor sophisticated measurement strategies are substitutes for validity. Systematic in-depth approaches to the assessment of content validity are needed, including a priori and a posteriori procedures. In this manuscript the a priori approach of specifying an instrument's content domain is addressed along with the a posteriori procedure of having a panel of judges assess the validation of the items. This article focuses on the process used to assess the content validity. Data from the psychometric testing of the Postpartum Depression Screening Scale (PDSS) are used to illustrate this process. The PDSS is a Likert-type self-report composed of seven dimensions: Sleeping/Eating Disturbances, Anxiety/Insecurity, Emotional Lability, Cognitive Impairment, Loss of Self, Guilt/Shame, and Contemplating Harming Oneself. Items on the PDSS were generated from a series of qualitative research studies on postpartum depression. Content validity was assessed using two approaches to review the scale: (a) a panel of five content experts with professional expertise in postpartum depression, and (b) a focus group of 15 nurses knowledgeable in the field.

Comparative analysis of the performance of the Postpartum Depression Screening Scale with two other depression instruments.

BACKGROUND: Postpartum depression affects approximately 13% of mothers but up to 50% of all cases of this tragic illness can go undetected. OBJECTIVE: The purpose of this study was to compare the performance of a newly created instrument, the Postpartum Depression Screening Scale (PDSS), with the Edinburgh Postnatal Depression Scale (EPDS) and a general depression scale, the Beck Depression Inventory-II (BDI-II). METHOD: In this methodological design a total of 150 new mothers completed these three instruments in random order, followed immediately by a DSM-IV diagnostic interview. Using the LABROC I program, the areas under each of the instrument's Receiver Operator Characteristic (ROC) curves were compared to determine if they were significantly different. RESULTS: Eighteen (12%) of the women were diagnosed with major postpartum depression, 28 women (19%) with minor postpartum depression, and 104 women (69%) with no depression. Compared to the EPDS, the PDSS had a significantly larger area under the ROC curve when screening for major or minor postpartum depression. When using the published recommended cut-off scores for major depression for the three instruments, the PDSS achieved the highest combination of sensitivity, 94%, and specificity, 98%. When detecting women with major or minor postpartum depression, the PDSS again yielded the highest combination of sensitivity (91%) and specificity (72%) of the three instruments. The PDSS identified 17 (94%) of the women diagnosed with major postpartum depression, the EPDS identified 14 of these women (78%), and the BDI-II identified 10 of the 18 women (56%). CONCLUSION: If mothers identified as "most depressed" are substantially determined by the instrument used, the implications for both research and clinical practice are significant. Researchers and clinicians need to be aware of the differential sensitivity of depression instruments which, while supposedly measuring the same construct, are focused on different components of this mood disorder.

Item response theory in affective instrument development: an illustration.

The benefits of item response theory (IRT) analysis in obtaining empirical support for construct validity make it an essential step in the instrument development process. IRT analysis can result in finer construct interpretations that lead to more thorough descriptions of low- and high-scoring respondents. A critical function of IRT is its ability to determine the adequacy with which the attitude continuum underlying each dimension is assessed by the respective items in an instrument. Many nurse researchers, however, are not reaping the benefits of IRT in the development of affective instruments. The purpose of this article is to familiarize nurse researchers with this valuable approach through a description of the Facets computer program. Facets uses a one parameter (i.e., item difficulty) Rasch measurement model. Data from a survey of 525 new mothers that assessed the psychometric properties of the Postpartum Depression Screening Scale are used to illustrate the Facets program. It is hoped that IRT will gain increased prominence in affective instrument development as more nurse researchers become aware of computer programs such as Facets to assist in analysis.

Further validation of the Postpartum Depression Screening Scale.

BACKGROUND: Up to 50% of all cases of postpartum depression go undetected. The Edinburgh Postnatal Depression Scale (EPDS) has been the only instrument available that was specifically designed to screen for this mood disorder. None of the items on the EPDS, however, are written in the context of new motherhood. OBJECTIVE: The purpose of this study was to further assess the construct validity of the newly designed Postpartum Depression Screening Scale (PDSS) along with its sensitivity, specificity, and predictive values. METHOD: A total sample of 150 mothers within 12 weeks postpartum participated in the study. Each mother completed in random order three questionnaires: The PDSS, EPDS, and The Beck Depression Inventory-II (BDI-II). Immediately after completing these three questionnaires, each woman was interviewed by a nurse psychotherapist using the Structural Clinical Interview for DSM-IV Axis 1 Disorders. RESULTS: Twelve percent (n = 18) of the mothers were diagnosed with major postpartum depression, 19% (n = 28) with minor postpartum depression, and 69% (n = 104) with no depression. The PDSS was strongly correlated with both the BDI-II (r = 0.81) and the EPDS (r = 0.79). The ability of the PDSS to explain variance in diagnostic classification of postpartum depression above that explained by the BDI-II and EPDS (i.e., incremental validity) was assessed using hierarchical regression. After explaining variance in group classification by the other two depression instruments, the PDSS explained an additional 9% of the variance in depression diagnosis. Using receiver operating characteristic (ROC) curves, a PDSS cut-off score of 80 (sensitivity 94% and specificity 98%) is recommended for major postpartum depression and a cut-off score of 60 (sensitivity 91% and specificity 72%) for major or minor depression. CONCLUSION: Based on the results of this psychometric testing, the PDSS is considered ready for use in routine screening of mothers.

Postpartum Depression Screening Scale: development and psychometric testing.

BACKGROUND: Approximately 400,000 mothers in the United States experience postpartum depression each year. However, only a small proportion of these women are identified as depressed by health care professionals. OBJECTIVES: To improve detection of this postpartum mood disorder, the purpose of this study was to assess the psychometric properties of a newly devised instrument, the Postpartum Depression Screening Scale (PDSS), a 35-item Likert-type self-report instrument. METHODS: Content validity was supported through the literature and the judgments rendered by a panel of five content experts and a focus group. The PDSS was administered to 525 new mothers. RESULTS: Confirmatory factor analysis provided empirical support for the existence of the hypothesized seven dimensions. A Tucker-Lewis goodness-of-fit index of 0.87 and a root mean square residual of 0.05 were judged supportive of model fit. Item response theory techniques provided further construct validity support for finer interpretations of the respective seven dimensions. Analysis of the Likert 5-point response categories further supported meaningful score interpretations. Alpha internal consistency reliabilities ranged from 0.83 (sleeping/eating disturbances) to 0.94 (loss of self). CONCLUSIONS: Empirically, all of the reliability and validity analyses supported the score interpretations posited for the PDSS. Currently, the sensitivity, specificity, and positive predictive value of the PDSS are being determined.

Self-concept: a cross-cultural study.

Self-concept differences between 190 American and 98 Danish 5th and 6th grade suburban students were examined in light of known societal and educational differences. Danish males and American females reported significantly higher self-confidence in personal attributes; American students reported higher achieving in school scores than Danish students. Results are discussed in the contex of cultural differences; further areas of investigation are offered.