Guttman Split-half Reliability Research Articles

Developing and validating a nursing website evaluation questionnaire

The aim of this paper is to report on the development and validation of an evaluation questionnaire for nursing websites. Increasing numbers of the general public and health care professionals are using the Internet to access health and health care-related information. The quality of such websites, therefore, needs to be evaluated. A total of 24 websites in the United States and Taiwan, including websites in 13 leading schools and 11 hospitals, were selected as test websites. A total of 120 participants with Master's degrees in either nursing or mechanical engineering were invited to examine the 24 websites using a nursing website evaluation questionnaire. A 32-item questionnaire was developed, covering overall impression, download and switch speed, accessibility and convenience, web page content, service function and compatibility with common browsers. Participants were requested to connect to the specific websites and browse the contents before filling in the questionnaire. The results were arranged into 12 different sequences and prepared on a CD using Macromedia Dreamweaver. Data were analysed using Cronbach's alpha, Guttman split-half reliability and factor analysis. Internal consistency was good, with a Cronbach's alpha of 0.93 in Part I and a KR-20 (Kuder Richardson) of 0.85 in Part II. Four factors given in Part I were identified as content truth, accessibility and convenience, speed and connection, and overall impression. The nursing website evaluation questionnaire had good construct validity and internal consistency.

Comparison of internal reliability and validity of the McGill Pain Questionnaire and the Short Pain Inventory

We compared the psychometric properties of the McGill Pain Questionnaire (MPQ) with the 17-item Short Pain Inventory© (SPI) in 60 outpatients with osteoarthritic knee pain. Split-half reliability, Guttman split-half reliability, Cronbach alpha and the correlation between the first and second half of the test were higher in the SPI total pain disturbance than any of the summary or subscales of the MPQ: 0.94, 0.94. 0.88 and 0.90, versus 0.71, 0.70, 0.69 and 0.55 for the MPQ sensory present pain intensity, 0.59, 0.57, 0.65 and 0.43 for the affective and 0.62, 0.55, 0.49 and 0.45 for the evaluative MPQ scale, respectively. The parameters for the SPI total mood disturbance were superior to all MPQ-derived scales. Dividing into high and low ‘pain experienced right now’ identified screening samples. The SPI ‘pain right now’ was more discriminating than the comparative MPQ item on both the SPI and MPQ. Additionally, none of the summary scales of the MPQ could show significant internal discrimination whereas the SPI did achieve this. There were 23 significant correlations with the SPI severity compared with 15 with the MPQ. SPI sadness, anxiety, anger, total mood disturbance and total pain disturbance were significantly correlated with the MPQ severity rating. The MPQ variables fared less well than the SPI in the degree of association between various pain parameters and physical severity. Factor analyses revealed that the SPI accounts for the majority of the variance (50%) compared with 17% for Factor 2. This second factor is best indexed by the McGill ‘pain now’ and also with the SPI ‘pain severity now’ item. Since the SPI indexes the severity as accurately (0.79) as the McGill, the only difference between the two is the sensory MPQ variance. However, since the SPI (Factor 1) also indexes some of the common variance of the MPQ sensory variable, the SPI also gains in this respect. If it were the case that the outcome of an analgesic clinical trial was the sensory aspects of the pain (cutting, throbbing, rasping) then the McGill should be the obvious outcome measure. For the patient, the most important feature of pain surely must be the physical severity and the unpleasantness of the experience. The MPQ is a rather long procedure and the evaluative scale is of dubious value. The majority of the pain variance is captured by the SPI and, secondly, by the sensory aspect of the MPQ. The SPI measures the emotional aspects of pain well and the McGill assesses the physical or sensory aspects of pain better than any other available. Both have their place according to one’s research interests and the clinical relevance. For example, if an investigation involves opiates like morphine, the SPI may be the better placed instrument, since the induced euphoria may present as a patient who can still feel the pain but is no longer bothered by it. Algesimetry requires measuring both the physical and emotional sensations of the patient.

A validation of the concept of current perceived health and the Current Perceived Health-42 (CPH-42) questionnaire.

Perceived or self-rated general health in its trait form reliably predicts subsequent mortality among certain sections of the community. In an investigation of the predictive value of health status at the time of assessment, 574 cases drawn from five groups (police trainees, firemen, students, out-patients and in-patients) were used to validate what we have called current perceived health (CPH) and the CPH-42, a measure of CPH is described. Results indicate those rating CPH as poorer were more likely to consult a doctor (chi 2 = 45.9, df = 1, p < 0.0000) and use medication (chi 2 = 61.0, df = 1, p < 0.0000). In comparison, general perceived health ratings did not predict either consultation or medication use. The instrument has good psychometric properties, with high divergent validity from general mental health and general health, with a Guttman split-half reliability > 0.79. Discriminant analysis using the CPH-2 achieved high correct classification rates (up to 100%). Comparison of a scoring method based on canonical coefficients (83% correct classification) with that based on summative scores (74% correct classification) indicate the former method to be more accurate. Age, and among people 40 years or older, gender were significant independent predictors of current perceived health status, with older people rating their current health less positively than younger, and females over 40 years rating their current health less positively than their male counterparts.

Development and psychometric properties of the Nurses' Assertiveness Inventory (NAI)

The Nurses' Assertiveness Inventory (NAI) is an 84-item self-report inventory designed to measure levels of assertiveness in nurses. In Study I, the scale was administered to 238 nurses whose responses were analyzed to ascertain the scale's psychometric properties and concurrent validity. Results revealed a test-retest reliability coefficient of r = 0.92, an α-reliability coefficient = 0.93 and a Guttman split-half reliability = 0.84. The scale was found to have high internal consistency and temporal stability. The concurrent validity of the NAI was significant ( P ⩽ 0.001) and highly correlated with the Rathus Assertiveness Scale and Wolpe-Lazarus Scale. Study II examined the scale's concurrent validity with other conceptually related measures with 51 nurses. Results from Studies I and II indicated that the NAI possessed good discriminative, convergent and concurrent validity. These results and recommendations for future research are discussed.