Studying Affect Among the Chinese: The Circular Way

Abstract

Past research on Chinese emotion has been plagued by the lack of a measurement map for affective feelings. In this article, I developed a fine-grained circumplex model of affective feelings that is arbitrarily defined by 12 segments. I created twelve 4-item self-report scales to measure affect felt during a clearly remembered moment (N = 395), and I cross-validated them twice with affect felt during the current moment (Ns = 269, 302). The structural and psychometric properties of these scales were strongly supported. The CIRCUM-extension procedure (M. Browne, personal communication, June 12, 1999) that places external correlates into the affective space showed that 28 of 28 mood states and 6 of 13 traits were significantly related to affect. External correlates did not clarify which of the affect dimensions were basic. The newly developed scales will serve as a useful tool in assessing affect among Chinese people and as a platform on which to extend the nomological net of affect.

Acknowledgments

The research reported in this paper was supported by the RGC General Research Fund (Project No. 644508) and the RGC Direct Allocation Grant (DAG05/06.HSS10). I thank Virginia Unkefer and Victoria Caplan for their help in preparing this article. I profusely thank Jim Russell for inspiring the development of the research agenda pursued in this article.

Notes

¹All participants specified the moment on which they based to report their affect.

²Affect ratings are always contaminated by the presence of a general factor (Bentler, 1969). Ipsatization removes individual differences in grand means and variances. This procedure is recommended for assessing circumplexity of data (Acton & Revelle, 2004; see Di Blas, 2007; Yik & Russell, 2003). To ipsatize the “satisfied” item, for instance, I deducted an individual's grand mean of all 113 items from that individual's satisfied rating; this difference is divided by the standard deviation of the 113 ratings for the same individual. I ipsatized each rating by 113 items.

³In the principal component analysis of the 113 ipsative items, the first 4 eigenvalues were 31.40 (27.79% of the total variance), 11.34 (10.04%), 5.60 (4.96%), and 3.53 (3.13%).

⁴I ipsatized each affect segment by the 12 scales.

^a Mehrabian and Russell (1974).

^b Russell, Weiss, and Mendelsohn (1989).

^c Barrett and Russell (1998).

^d Thayer (1996);

^e Larsen and Diener (1992).

^f Watson and Tellegen (1985).

^g Costa and McCrae (1992).

^h Yik and Bond (1993).

^a Mehrabian and Russell (1974).

^b Russell, Weiss, and Mendelsohn (1989).

^c Barrett and Russell (1998).

^d Thayer (1996).

^e Larsen and Diener (1992).

^f Watson and Tellegen (1985).

^g Costa and McCrae (1992).

^h Yik and Bond (1993).