Ruminations and Flow: Why Do People with a More Harmonious Passion Experience Higher Well-Being?

Research shows that passion can be distinguished from other motivational constructs such as intrinsic and extrinsic motivation. From a theoretical perspective (Vallerand 2010), passionate activities differ from intrinsically motivated ones in that they are deeply valued by the individual and they are internalized into the person’s identity. This is not necessarily the case for intrinsically motivated activities. The concept of passion (both harmonious and obsessive) also differs from extrinsic forms of motivation because, contrary to extrinsic motivation (which means to engage in an activity to obtain something outside of the activity), passion implies a strong liking for the activity. Because passion entails both an identification to and a strong liking for an activity, it should have a more profound impact on people’s lives than motivation or goals. Empirical evidence confirms that when motivation and passion are assessed toward a given activity, harmonious and obsessive passion, respectively predict positive and negative affects above and beyond what is predicted by intrinsic and extrinsic motivation (Vallerand et al. 2003, Study 2).

The χ _M ² tests for differences between the estimated and observed correlation matrices, such that a non-significant p value supports the adequacy of the proposed model. The NC, which is the ratio of the chi-square statistic on its degrees of freedom, takes into account the sample size and is thus usually a better fit index than the χ _M ² . Values smaller than 2.0 for this index indicate a good fit (Tabachnick and Fidell 2007). The CFI, NFI, and NNFI are comparative or incremental fit indices that assess fit relative to other models (Kline 2005). Values greater than .95 on the CFI, the NFI and the NNFI are indicative of a good-fitting model (Tabachnick and Fidell 2007). The NNFI is adjusted for model complexity and can yield values greater than 1.00. The RMSEA is a parsimony-adjusted index which corrects for model complexity (Kline 2005). Browne and Cudeck (1993) suggest that RMSEAs less than .05 are indicative of a “close fit” and that values up to .08 represent reasonable errors of approximation. Finally, SRMR is a measure of the mean absolute correlation residual (i.e., the mean difference between observed and predicted covariances) and should be less than .10 to indicate a reasonable fit (Kline 2005).

Specifically, correlations between passion and the other variables are expected to differ from the relations found using SEM because SEM controls for the common variance typically found between the two types of passion (e.g., Philippe et al. 2010; Vallerand et al. 2003). By controlling for this positive correlation, relations between each type of passion and the dependent variable can be investigated independently from the effect of the other type of passion. The two types of passion are expected to correlate positively because as types of passion they share common elements. Indeed, both types of passion refer to a unique relationship with an activity that a person has internalized in his or her identity. In addition, whether more obsessively or harmoniously passionate, people equally report liking their activity, finding it important, and devoting time to it (for a similar analytical strategy, see Carbonneau et al. 2010; Mageau et al. 2005; Philippe et al. 2009a, c; Rousseau and Vallerand 2008). The correlation between harmonious and obsessive passion observed in the present study is similar to the one found in previous research (e.g., Philippe et al. 2010; Vallerand et al. 2003).

Because past research shows that men and women often react differently in the school context (Cai 2005; Else-Quest et al. 2010; Ewert 2010; Fan 2011; Tison et al. 2011), we replicated our final model while controlling for the impact of gender on all endogenous variables. Results showed that gender was not significantly linked to the model’s variables and that all significant relations were observed above and beyond any gender differences. For the sake of parsimony, the model is presented without controlling for gender.