Investigating Importance Weighting of Satisfaction Scores from a Formative Model with Partial Least Squares Analysis

Chia-Huei Wu, Lung Hung Chen, Ying-Mei Tsai
2008 Social Indicators Research  
This study introduced a formative model to investigate the utility of importance weighting on satisfaction scores with partial least squares analysis. Based on the bottom-up theory of satisfaction evaluations, the measurement structure for weighted/unweighted domain satisfaction scores was modeled as a formative model, whereas the measurement structure for global satisfaction scores was modeled as a reflective model according to topdown theory. The purpose was to see if the predictive effect of
more » ... importance-weighted domain satisfaction scores is stronger than unweighted domain satisfaction scores in predicting global satisfaction scores. Three datasets in life, self, and job satisfaction were analyzed. In the life satisfaction dataset, 237 undergraduates at Central Taiwan University of Science and Technology voluntarily provided their responses. The mean age of respondents was 20.80 years (SD = 1.05). In the self-satisfaction dataset, 269 undergraduates at National Taiwan University provided their responses. The mean age of respondents was 19.78 years (SD = 1.44). Finally, in the job satisfaction dataset, 557 staff members in seven Taiwan provincial hospitals provided their responses. The mean age of respondents was 35.87 years (range from 21 to 65, SD = 8.60). Three measures of domain satisfaction, domain importance, and global satisfaction were collected in each dataset. Partial least squares analysis was used in model estimation. All the results revealed that unweighted domain satisfaction scores have a stronger predictive effect for global satisfaction measures than importance-weighted domain satisfaction scores, indicating that importance weighting on satisfaction scores did not have an empirical benefit. The procedure of importance weighting on satisfaction rating scores at the item level has been applied in quality of life (QOL) instruments, such as the Comprehensive Quality Of Life Scale (ComQol, Cummins 1997), the Quality of Life Index (QOL Index, Ferrans and Powers 1985), the Quality of Life Inventory (QOL Inventory, Frisch 1992), and the Quality of Life Profile-Adolescent version (QOLPAV, Raphael et al. 1996). The goal of importance weighting is to obtain individualized QOL scores by incorporating individuals' importance perceptions into their satisfaction judgments. However, Russell and Hubley (2005) , Wu (2007, 2008a, b) and Wu and Yao (2006a) recently indicated that this weighting procedure does not have a solid psychological basis. Based on Locke's (1969 Locke's ( , 1976 range-of-affect hypothesis, Wu's studies (2007, 2008a, b) and Yao (2006b, 2007a) showed that satisfaction evaluations are determined by the interaction effect between have-want discrepancy (the discrepancy between what the individual wants and what he/she perceives himself/herself as getting) and importance (the importance to the individual of what is wanted). Specifically, at a given level of discrepancy, an item with high personal importance can produce a wide affective reaction ranging from great satisfaction to great dissatisfaction, while an item with low personal importance can only produce a restricted affective reaction around the neutral point of the satisfaction-dissatisfaction dimension. Item satisfaction, therefore, already has incorporated within it information regarding the item's importance, which renders weighting item satisfaction with item importance redundant. In addition to quality of life research, the range-of-affect hypothesis was also supported in job satisfaction studies (McFarlin et al. 1995; McFarlin and Rice 1992; Mobley and Locke 1970; Rice et al. 1991a , b). Friedlander (1965 and Trauer and Mackinnon (2001) also indicated that items of extreme satisfaction/ dissatisfaction are more important than mild satisfaction/dissatisfaction items, supporting that the responses of satisfaction have a wider range on dissatisfaction-satisfaction scale for the high-importance item. Consistent with the implication derived from the range-ofaffect hypothesis, many studies have found that importance-weighted satisfaction scores, compared to unweighted satisfaction scores, did not have an additional contribution in predicting criterion variables, such as overall subjective well-being (e.g.also mentioned other drawbacks of importance weighting. First, the weighted scores are not easy to interpret. For example, a weighted score 4 would produced from a 4-point satisfaction score and 1-point importance score and a 1-point satisfaction score and 4-point importance score. It can be seen that the same weighted score represents two different situations. If we only used weighted scores to evaluate individual's QOL, we cannot thoroughly understand these two kinds of life conditions. Second, satisfaction and importance rating scores are usually non-ratio level data. The multiplicative nature of the weighted scores cannot hold the statistic position of its meaning. This is also an obstacle for interpreting the meaning of weighted scores. However, Hsieh (2004) did not think these findings and obstacles were sufficient to justify the abandonment of incorporating importance weighting into QOL instruments. He thought that incorporating importance weighting into QOL instruments is appropriate, but the critical issue is how to weight. He proposed a formative model as the theoretical basis to compute the importance-weighted satisfaction score. The perspective of a formative model is different from the perspective of a reflective model. In brief, in a reflective model,
doi:10.1007/s11205-008-9264-1 fatcat:ilidjzhmr5ckfhskk3avc7ch54