Thursday, March 25, 2010

Reseach bytes 3-25-10: Is impulsivity a hierarchical multidimensionl construct?

I just read the following excellent article re: the possible factor structure of the construct of impulsivity.  The authors borrowed items from many established measures of impulsivity and then conducted exploratory factor analysis.  The sample size was decent (although probably only generalizable to young adults---they were college students).

Whenever I see an article that says it factor analyzed item level data, I always cringe as many folks are not aware of the methodological confound of finding item "difficulty or popularity" factors (which mask substantive construct factors), especially on rating scales.  Thus, the standard recommendation is to factor analyze a matrix of polychoric correlations...which they did (yes, I know, there are even some methodological questions about the use of polychoric correlations and other types of correlation metrics have been sugested).  But, I think this is a decent article worth reading.

I find the idea of a proposed hierarchical structure of impulsivity very intriguing. 

  • Kirby, K & Finch, J. (2010).  The hierarchical structure of self-reported impulsivity. Personality and Individual Differences  Volume 48, Issue 6, Pages 704-713 (click here to view)

The hierarchical structure of 95 self-reported impulsivity items, along with delay–discount rates for money, was examined. A large sample of college students participated in the study (N = 407). Items represented every previously proposed dimension of self-reported impulsivity. Exploratory PCA yielded at least seven interpretable components: Prepared/Careful, Impetuous, Divertible, Thrill and Risk Seeking, Happy-Go-Lucky, Impatiently Pleasure Seeking, and Reserved. Discount rates loaded on Impatiently Pleasure Seeking, and correlated with the impulsiveness and venturesomeness scales from the I7 (Eysenck, Pearson, Easting, & Allsopp, 1985). The hierarchical emergence of the components was explored, and we show how this hierarchical structure may help organize conflicting dimensions found in previous analyses. Finally, we argue that the discounting model (Ainslie, 1975) provides a qualitative framework for understanding the dimensions of impulsivity.

1. Introduction
2. Methods

2.1. Participants
2.2. Materials and procedure

2.2.1. Self-report impulsivity items
2.2.2. Discount rate measure
2.2.3. Reward selection

2.3. Analyses

2.3.1. Polychoric correlations
2.3.2. Principal components analyses (PCA)
2.3.3. Hierarchical component emergence
2.3.4. Number of dimensions

3. Results

3.1. Number of dimensions
3.2. Component emergence
3.3. Discount rates

4. Discussion

4.1. Relations to previous dimensions
4.2. The I7
4.3. The BIS-11
4.4. The problem of redundant items
4.5. Face-valid items
4.6. Reward attributes and the time scales of impulsivity
4.7. Reward attributes
4.8. Time scales
4.9. Limitations

5. Conclusion

Technorati Tags: , , , , , , , , , , ,

No comments: