Sie sind auf Seite 1von 4
Psychological Reports, 2007, 100, 375-378, © Psychological Reports 2007 THE HOPELESSNESS SCALE: A FACTOR ANALYSIS* MAURIZIO POMPILI Department of Psychiatry, Sant’ Andrea Hospital University of Rome "La Sapienza” McLean Hospital, Harvard Medical School ROBERTO TATARELLI Department of Psychiatry, Sant’ Andre Hospital University of Rome “La Sapienza” JAMES R. ROGERS DAVID LESTER University of Akron The Richard Stockton College of New Jersey Summary.—A. confirmatory factor analysis of the Beck Hopelessness Scale in a sample of 340 Italian students did not support the 3-factor model reported for previ- ‘ous samples of psychiatric patients. A follow-up principal axis factor analysis yielded two interpretable correlated factors, suggesting that the structure of the scale may dif- fer across clinical and nonclinical groups and as a function of nationality. Beck, Weissman, Lester, and Trexler (1974) described a 20-item true- false scale to measure hopelessness, the cognitive component of depression. In the original study, a factor-analysis of the scale was carried out using the responses of 294 people who had attempted suicide. Three factors account- ing for 53.5% of the total variance were identified and labeled as feelings about the future (e.g., | look forward to the future with hope and enthusi- asm), loss of motivation (e.g., I never get what I want so it’s foolish to want anything), and future expectations (e.g., The future seems vague and uncer- tain to me). Dyce (1996) found a similar 3-factor solution in a sample of psy- chiatric outpatients of unspecified diagnosis. Although these factors have been used to create subscales for use in re- search, the structure has not been investigated using confirmatory procedures nor have subsequent factor analyses of the Hopelessness Scale with nonclin- ical and non-American samples consistently replicated this 3-factor solution. For example, Tanaka, Sakamoto, Ono, Fujihara, and Kitamura (1998) found a 2-factor structure in a community sample of Japanese subjects, and neither factor matched any of the three factors reported by Beck, et al. or by Dyce. Similarly, Steed (2001), for a sample of Australian university students, did "Address correspondence to M. Pompili, M.D., Department of Psychiatry, Sant'Andrea Hospital, Via di Grottarossa, 1035, 00189 Roma, Italy or e-mail (maurizio.pompili@uniromal it ‘or mpompili@mclean.harvard.edu). DOI 10.2466/PRO.100.2.375-378 376 M. POMPILI, ET AL. not replicate the 3-factor structure. Since the three subscales continue to be used in research, however, it is important to ascertain their structural valid- ity. The purpose of this research was to explore Beck, ef a/.’s (1974) 3-fac- tor structure of the Hopelessness Scale in an Italian sample using confirma- tory factor analysis. An Italian version of Beck, ef al.’s 20-item scale was pre- pared by translating the scale into Italian, back-translating it into English, and correcting the final Italian version. This version was administered to 142 male and 198 female undergraduate students at the University of Rome, Italy. The mean age for this sample was 22.6 yr. (SD=3.4). The internal con- sistency reliability for the total scale with these subjects was .73 and Cron- bach alphas for the three subscales were .50, .48, and .50, respectively. First, we used confirmatory factor analysis to test the 3-factor model in our data. Confirmatory factor analysis is a hypotheses-testing approach ap- plied to measurement models to provide a test of @ priori structures (Hatch- er, 1994; Rogers, Abbey-Hines, & Rando, 1997). In confirmatory factor anal- ysis, the measurement model or structure of a scale such as the Hopelessness Scale is identified based on prior research or theory, and the concordance or “fit” of the data to the model is evaluated statistically. For this study, we evaluated Beck, et al.’s (1974) 3-factor, uncorrelated model to assess its fit with the data. For comparison and because a logical argument could be made that the three factors of the scale should be intercorrelated, the rela- tive fit of the correlated model was also investigated. Based on recommenda- tions in the literature (Bollen & Long, 1993), five “goodness of fit” indices were employed to evaluate the models; x’, the Normed Fit Index (NEI), the Tucker-Lewis Fit Index (TFI), the Parsimonious Normed Fit Index, (PNFI), and the Root Mean Square Error of Approximation (RMSEA). TABLE 1 Common Factor ANaLYsts: 3-Facton Monst of Horetessness ScaLe 3-factor Model x? af P NFL TLI PNET RMSEA Uncorrelated 564.92 BS <.001.-—S1.SSS2SSCAS~—«ODT(,089-106) Correlated 34939132, <.001_—— 707560070 (.061-.079) Note—Statistically significant x? values indicate poor fit to the data. Cut-off values for inter- Bicting good for for the NFY and the TLI are 90. The recommended. cutoff value for the °NFI is .80. RMSEA of $05 is recommended for interpreting a good fit and values between .05 and .08 are interpreted as adequate fit. Table 1 lists the confirmatory factor analysis results which indicate that the correlated model provided a better fit to the data than the uncorrelated model. While the correlated model provided a better fit, neither model pro- vided an adequate fit to the data based on established cut-off values. Since the confirmatory factor analysis procedure did not support either HOPELESSNESS SCALE: FACTOR ANALYSIS 377 the correlated or uncorrelated three-factor model, the responses were next subjected to a principal axis factor analysis. Based on a combination of Kai- ser’s criterion (ie., eigenvalues 21.00) and scree plot analysis, six factors were initially identified. However, using a factor loading cutoff of 2.40, only two factors met the interpretative criteria of at least three significant load- ings for retention (Hatcher, 1994). Because the correlated model provided a stronger fit to the data than the uncorrelated model in the confirmatory pro- cedures, the results of this factor analysis were subjected to a promax (ob- lique) rotation, allowing the two retained factors to correlate. Table 2 pre- sents the results of the factor analysis along with the means and standard deviations for the individual items and item-total correlations (r). The corre- lation between the two retained factors was .52. TABLE 2 Rorarep Factor Loapincs, [rem MEANS AND STANDARD Deviations, aNp Irem-Totat CORRELATIONS Tem Factor 1 __ Factor 2 Item M SD Ta 1 26 AT 4 35 35 2 10 36 08 27 3S 3 -.05 24 2 33 B 4 10 37 a 45 25 5 04 33 9 40 23 6 32 AZ 07 26 32 7 3 AT AL 3l 47 8 19 09 35 48 21 9 = 13 01 82 38 12 10 01 35 21 Al 27 u JSF 08 09 28 45 12 204 40 23 42 36 13, 08 OL 22 Al 05 14 08 Al 36 48 36 15 02 64 27 44 51 16 7 -08 .08 27 AB 7 57 05 .08 27 39 18 -H 8 48 50 a@ 19 56 10 06 24 Bl 20 9 ~13 04 2 Al % variance 18.9 55 Note.—Loadings 240 are indicated in bold type. Neither of the factors which emerged in this analysis provided a clear match to those identified by Beck, et al. and by Dyce. Factor 1 in the pres- ent sample appears to tap a more self-referent negative expectation of the fu- ture, e.g., I never get what I want so it’s foolish to want anything (positively scored item), while Factor 2 seems to reflect a more generalized negative expectation, e.g., I have great faith in the future (negatively scored item). 378 M, POMPILI, ET AL. These results suggest that the factor structure of the Beck Hopelessness Scale may differ in nonclinical and clinical groups (who almost certainly dif- fer in their mean Hopelessness score) and may also differ in samples from other nations. Thus, researchers should be cautious in interpreting the te- sults of the Beck Hopelessness Scale in the absence of evidence for its valid- ity in their samples. REFERENCES Beck, A. T., Weissman, A., LesTER, D., & Trexer, Le (1974) The measurement of pessimism. Journal of Consulting & Clinical Psychology, 42, 861-865. Bouy, K Ay &e Lone 5 (1993) Testing siructural equation models. Newbury Park, CA: age. Dyce, J. A. (1996) Factor structure ‘of the Beck Hopelessness Scale. Journal of Clinical Psychol- ogy, 52, 555-558. Harcuen, L. (1994) A step-by-step approach to. using the SAS system for factor analysis and sen 1998 on mcdeling, Cary, NC: SAS Insta. Rocers, J. R., Aspey-Hines, J., & Ranpo, R. A. (1997) Confirmatory factor analysis of the gen- ‘der role conflict scale: a cross-validation of Good, et al., 1995. Measurement and Evalua- tion in Counseling and Development, 30, 137-145. Sreep, L. (2001) Further validity and reliability evidence for Beck Hopelessness Scale scores in 1 20 ak sample, Educational and Psychological Measurement, 61, 303-316. Tanaka, E., Sakamoto, S., ONO, Y., Fuyara, S., & KrtaMURA; T. (1998) Hopelessness ina i inunity population. Journal of Social Psychology, 138, 381-590. Accepted February 1, 2007.

Das könnte Ihnen auch gefallen