100 likes | 377 Views
A Reliability Generalization of the Life Satisfaction Index. K. A. Wallace & J. C. Caruso University of Montana. Presented at the Annual Meeting of The Gerontological Society of America, November 2002. Purpose.
E N D
A Reliability Generalization of the Life Satisfaction Index K. A. Wallace & J. C. Caruso University of Montana Presented at the Annual Meeting of The Gerontological Society of America, November 2002.
Purpose • To examine score reliability for a widely used measure of life satisfaction, the Life Satisfaction Index (LSI; Neugarten et al., 1961) • Average score reliability • Variation in score reliability as a function of sample characteristics (e.g., gender; mean age; scale length; etc.)
Developing Ratings of Life Satisfaction • Crafted to assess well-being using a subjective evaluation of one’s own present and/or past life • Part of the Kansas City Study of Adult Life (Neugarten et al., 1961) • Thematic analysis of measures of adjustment and morale • 5 components of well-being: • Zest versus apathy • Resolution and fortitude • Congruence between desired and achieved goals • Self-concept • Positive mood tone
The Scales • Life Satisfaction Index A (LSIA) • Life Satisfaction Index B (LSIB) • Life Satisfaction Index Z (LSIZ; Wood et al., 1969) • 18-item version (Adams, 1969) • 8-item version (LSIW; James et al., 1986) • 30-item version (Maynard, 1993)
Sample size • Scale length • Mean age of sample • Standard deviation of age • Gender • Mean LSI • Standard deviation LSI • Language of administration • Type of sample Reliability Generalization • Meta-analytic technique • Examines average score reliability (e.g., Vacha-Haase, 1998) • Examines relationships between study characteristics and score reliability
Method – Data Collection • PsycINFO literature search • Life satisfaction index and LSI • 157 possible articles • 59.87% no mention of reliability • 9.56% indicated LSI reliable test, no data • 6.37% cited reliability from previous work • 3.18% reported reliability in unusable form • .64% not empirical • 1.27% could not be obtained • 19.11% (30) provided usable reliability information • Total of 34 samples used
Descriptive Statistics for Sample Characteristics (N=34) Variable M SD Range Sample size 235.53 328.46 20-1571 Scale length 17.09 4.21 8-30 Mean age 61.79 17.17 20.2-83.3 Std dev age 6.89 3.38 3-15.7 Proportion female .63 .28 0-1.00 Mean LSI 15.56 13.01 3.08-65.5 Std dev LSI 3.40 2.34 .52-11.83 Lang of admin .78 .42 0-1.00 Sample type .26 .51 0-1.00
Results • Average Score Reliability • Mean = .79 (SD = .10) • Median = .79 • Range of .56 (.42 to .98) • Bivariate correlations • Score reliability was not significantly related to: scale length, mean age, standard deviation of age, proportion female, sample size, mean LSI, or standard deviation LSI • t tests • No difference in score reliability as function of language of administration or sample type
Discussion • Adequate average score reliability for the LSI • File drawer problem? • No relationship between score reliability and sample characteristics • Preliminary evidence for adequate reliability of scores generated with LSI across various sample characteristics • Limitations?
Future Research & Implications • Replication • Conceptualization of reliability as property of scores (e.g., Wilkinson & APA Task Force on Statistical Inference, 1999) • Inclusion of more detailed demographic and reliability information