Scaling the Creative Self: An Item Response Theory Analysis of the Short Scale of Creative Self

ABSTRACT

Self-report scales have become the most widely used instruments to capture people’s self-perception of creativity. Previous studies, however, provided only a limited insight into the psychometric properties of such measures. This paper reports an extensive item response theory (IRT) analysis of the Short Scale of Creative Self (SSCS): one of the most frequently used scales of creative self-concept. Based on samples from 14 studies (overall N > 26,000), we report IRT parameters of creative self-efficacy and creative personal identity scales’ items. We examined whether the scores obtained in the SSCS depend on the length of the response scale (5-versus-7-point Likert scales) and whether latent scores are comparable across different data collection methods (online, paper-and-pencil, phone), age, and gender. The results confirmed the two-factor structure of the SSCS, good psychometric properties of its items, as well as invariance regarding response scales, age, gender, and method of data collection. At the same time, the items (and consequently—scales) were easy in the psychometrical sense, thus providing much more reliable scores among individuals who scored low or medium in creative self-concept. Longer (7-point) and shorter (5-point) Likert scales performed similarly, with some psychometric arguments favoring fewer points on the scale. Gender differences were negligible (Cohen’s d between 0.00 and 0.01). We discuss potential ways of further improvement and development of the SSCS.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10400419.2022.2123139

Notes

1. Our unsystematic overview suggests that most of the recent studies that use SSCS are conducted online: resembling the general tendency in psychology nowadays. While online testing provides multiple advantages, like access to more diversified samples or the possibility to randomize items, it does not go without costs, be it careless responding or lack of control over the testing situation. In this article, we compare psychometric properties of SSCS across data collection methods without any strong hypothesis regarding potential influences of the data collection on possible scores and parameters of SSCS.

2. In this analysis, participants’ responses to each item were divided by the potential maximum (i.e., 5 in the case of the 5-point scale and 7 in the case of the 7-point scale). We appreciate the recommendation by the anonymous reviewer to proceed with the analysis this way.

3. As Silvia and Rodriguez (2020) explain, TAM infit and outfit estimates are not directly comparable to the same estimates obtained in other packages, given that TAM uses individual posterior ability distributions to obtain fit indices, while other methods usually rely on individual ability estimates.

4. We have also evaluated differential item functioning (DIF) regarding gender, data collection method, and age in lordif package. Lordif uses a logistic ordinal regression approach to examine whether people from different groups (e.g., men and women or representing different age cohort) yet with the same level of latent traits (CSE and CPI) vary in their predicted scores in each SSCS item. Having large samples, we used a very stringent cutoff of R² = .005 (even more rigorous than recommended by Silvia and Rodriguez (2020)) to flag items with potential DIF. This criterion, however, resulted in no items flagged, regardless of the categorical variable analyzed (gender, data collection method, age) and the SSCS factor (CSE, CPI) (see OSF archive for more details).

5. Additional measurement invariance analyses taking into consideration response scale length as well as testing for gender and age invariance separately for 5-point and 7-point response scales are presented in SOM (see SOM Tables 4–7). Comparing differences in CSE and CPI across specific age cohorts is out of scope of the present article. However, for interested readers, we present such an analysis in SOM (see Figure 2 and Tables 8–9 with the description). More detailed examination and discussion on age-related changes in CSE and CPI can be found elsewhere (Karwowski, 2015, 2016).

Additional information

Funding

Results reported in this article were obtained thanks to grants received from the National Science Centre Poland, among others UMO-2016/22/E/HS6/00118 and 2016/23/B/HS6/03898. Izabela Lebuda was supported by funding from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No 896518.