A Cognitive Interpretation is not at Odds with Equal Odds: A Latent Variable Investigation of Creative Thinking

Abstract

A fundamental question in creativity research is whether chance models of creativity align with the idea that cognitive ability can explain individual differences in creative cognition. Using two datasets (N_Dataset1 = 462 and N_Dataset2 = 331) we extended previous work on the equal odds baseline (i.e., a chance model of creativity) by using a latent variable analytic approach to model latent residual factors rectified for fluency. We compared three measurement models: the EOB latent variable model, a residualized model, and a ratio score model, examining their reliability and associations with cognitive abilities (i.e., fluid intelligence and working memory capacity). We found that when a chance model of creativity is deemed to be an appropriate fit for a given dataset, as evidenced by Dataset 2 in our study, a cognitive interpretation of originality factor η is warranted (i.e., around 21% of variance can be explained by working memory capacity). Our work, thus, refines discussions about the compatibility of chance models with cognitive explanations of creative cognition.

Keywords

divergent thinking residuals equal odds structural equation modeling creative cognition

Ideas or products of high creative quality must be original and effective (Amabile & Pratt, 2016; Runco & Jaeger, 2012). Chance models of creativity suggest that generating more ideas increases the likelihood of generating a high-quality idea, or, as Osborn famously stated, “quantity breeds quality” (Osborn, 1963; p. 131). Chance models of creativity are ubiquitous in creativity research and have been applied to scientific productivity (Feist, 1997; Forthmann et al., 2020a; Simonton, 2004; Sinatra et al., 2016), brainstorming research (Briggs & Reinig, 2010; Nijstad et al., 2010; Osborn, 1963; Rietzschel et al., 2007), creativity in music (Hass & Weisberg, 2015; Kozbelt, 2005, 2008), and divergent thinking (Forthmann et al., 2021b; Mouchiroud & Lubart, 2001). Previously, researchers have argued that chance models are not akin to any sort of psychological cognitive interpretations (Rietzschel et al., 2007); in other words, high-quality ideas should not be understood as a result of careful reasoning but are rather attributable to the random generation of a large number of ideas.

However, while this may be true for a special case of chance models (Forthmann et al., 2021c), recent developments in research on chance models of creativity allow for explicit modeling of individual differences in creative thinking (Caviggioli & Forthmann, 2022; Forthmann et al., 2021c), which, in turn, may allow researchers to evaluate the relationship between creative cognition and other cognitive abilities at the between-person level. Thus, the objective of this paper is to compare three different latent variable modeling approaches to separate originality (i.e., the primary component of creativity; Diedrich et al., 2015; Pichot et al., 2024; Zhou et al., 2017) from the number of responses (i.e., fluency) to a given prompt; the three models are the equal odds baseline (EOB) model, a residualized model, and a ratio score model. Furthermore, we aim to evaluate the association of a (residual) factor of originality in each model with general cognitive functioning, as operationalized by fluid intelligence (Gf) and working memory capacity (WMC).

In the following, we first outline the EOB model, which is the focal chance model of creativity in the current work. In this section, we also introduce a latent variable extension of the EOB that allows us to model and evaluate reliable individual differences in creative thinking while explicitly accounting for the complex relationship between quantity and quality. This model is then compared to two different modeling approaches that also account for the fluency contamination in originality (Forthmann et al., 2020b; Hocevar, 1979). We then outline the relationship between cognitive abilities and creative thinking. The three models are then applied to two distinct datasets to relate individual differences in originality (as captured by the different models) to latent variables reflecting cognitive abilities.

Chance Models of Creativity and Their Cognitive Psychological Interpretation

Simonton's equal odds baseline model (Simonton, 2004, 2010)—i.e., the chance model of creativity employed in the current work—proposes that the number of creative hits H_p of creator p is a linear function of the total number of productions T_p of that creator

H_{p} = ρ T_{p},

(1)

with hit ratio ρ. Importantly, this very basic model does not capture any individual differences in hit ratios. Data adhering to this model display a level of between-person hit ratio variation that can be solely explained by sampling variation. This emphasizes a pure chance process, because between-person variation in cognitive abilities as an explanation for empirical variation in hit ratios is not incorporated in this model.

This very basic model has been coined “strict EOB” in the literature (Forthmann et al., 2021c), and it is most likely this extreme special case of the EOB that was mentioned by Rietzschel et al. (i.e., “the baseline approach in its strictest form,” 2007; p. 934) when discussing a potential incompatibility between chance models of creativity and cognitive capacity explanations of creative ideational behavior (e.g., as expressed while working on divergent thinking tasks).

To see why a cognitive interpretation is not at odds with the EOB, one must consider the most recent version of the model (Simonton, 2004, 2010) that further incorporates a random error term $u_{p}$ to explicitly model individual differences in hit ratios

H_{p} = ρ T_{p} + u_{p} .

(2)

This random error term may include true individual differences in general cognitive abilities that are relevant for the given task or even across different task types. It may further incorporate sources of random variation. Thus, if Var(u_p) > 0, individual differences in u_p could potentially be explained by individual differences in cognitive abilities. Clearly, reliable individual differences in u_p are necessary for a cognitive interpretation within the EOB framework. Finally, it should be noted that the EOB implies a linear relationship between H and T, and the EOB further requires that the ratio score H/T is uncorrelated with T (Simonton, 1988, 2003, 2004).

However, when dealing with empirical data, it is important to consider sampling variation as an additional source of variance. It is critical to determine if Var(u_p) exceeds the expected amount of variance attributable to sampling error. This can be empirically tested when H_p is operationalized by a simple count of “hits” (i.e., the number of original responses). However, in many cases, H_p represents the sum of citations received for a researcher's scholarly papers or the sum score of subjective ratings in a divergent thinking task. In these situations, testing for residual variance beyond sampling error is not straightforward. Hence, for the purpose of the current work, we propose a latent variable extension of the EOB (Caviggioli & Forthmann, 2022) to explore for reliable individual differences in the residual term.

A Latent Variable Extension of the Equal Odds Baseline That Allows for Item-Specific Quantity and Quality Scores

The EOB, represented by Equation 2, is a basic linear regression model. The intercept is fixed at zero and the regression slope, represented by the average hit ratio ρ, is fixed to the ratio of average quality and average quantity. Hence, it is straightforward to estimate the EOB model as a structural equation model, and previous work has explored this option empirically (Forthmann et al., 2020a; Forthmann et al., 2021a; Forthmann et al., 2021c). Latent variable extensions of the EOB have also been proposed by modeling multiple quality indicators for the same set of inventors’ patents (Caviggioli & Forthmann, 2022). For divergent thinking tasks, we propose a latent variable extension that considers item-specific quantity and quality scores, T_i and H_i, respectively. T_i and H_i refer to fluency and summative originality, respectively, in the context of divergent thinking research.

First, consider summative originality H_ip of item i and person p as a linear function of fluency score T_ip:

H_{i p} = ρ_{i} T_{i p} + u_{i p},

(3)

with regression slope ρ _i and disturbance (i.e., error term) u_ip.

Next, the error term can be split into a latent originality factor η _p (weighted by factor loading $λ_{i}$ ) and unsystematic sources of error $ε_{i p}$ :

H_{i p} = ρ_{i} T_{i p} + λ_{i} η_{p} + ε_{i p} .

(4)

This model can be fitted to various originality scorings for divergent thinking, as long as single responses are scored in the range of non-negative real numbers and aggregated by summation. An advantage of this model is the availability of reliability estimates based on structural equation modeling, such as Raykov's reliability (Raykov, 1997). This way, the amount of reliable individual differences in original thinking can be quantified on an intuitive metric.

Models Competing with the EOB Extension

Building on the latent variable extension of the EOB model, which addresses the issue of fluency contamination of originality (e.g., Forthmann et al., 2020b; Hocevar, 1979), two additional models also arguably address this fluency-originality contamination and can be compared against the extended EOB approach: a residualized model and a ratio score model.

While the EOB model is based on the assumption that “fluency breeds quality,” the residualized model offers an alternative perspective, aiming to isolate originality from fluency (“originality without fluency”). In this model, a latent originality factor η represents pure originality, with fluency being partialled out through residualization:

H_{i p} = β_{0, 1} + β_{i} T_{i} + λ_{i} η_{0} + ε_{i} .

(5)

The ratio score model takes a different approach by incorporating a weighted relationship between originality and fluency that reflects the efficiency of an ideational process. In this model, originality is weighted by fluency (i.e., average originality scores). Here, the latent factor η is not an expression of pure or isolated originality; rather, it reflects a relationship between the number of original ideas relative to the number of ideas generated:

H_{i} / T_{i} = λ_{i} η_{H / T} + ε_{i} .

(6)

Figure 1 summarizes the above-described competing models and their interpretation of the latent factor η.

Figure 1.

Competing chance models for creativity.

Fluency, Originality, and Fluid Cognitive Abilities

Creative ideation describes the process of generating divergent ideas (Barbot, 2018; Fink & Benedek, 2014). Divergent thinking tasks—the most used operationalization of creative ideation (Kaufman et al., 2008)—require test takers to be both fluent (quantity of responses; e.g., Runco, 2020) and original (quality of responses; e.g., Carroll, 1993) regarding their responses to an open-ended task or problem. Both aspects of creative ideation have been subsumed under the larger first-order factor of retrieval from long-term memory in contemporary intelligence structure models (e.g., Carroll, 1993; Schneider & McGrew, 2018), but this classification has recently been questioned (Weiss et al., 2024). However, this possible reclassification of structural aspects does not necessarily change the fact that both aspects are inherently interwoven with fluid abilities that are essential for a wide range of intelligence-related performances, and that both may also contribute to generating original ideas (cf. Gerwig et al., 2021). In the following, we outline the status quo of fluid abilities in order to establish how they might be related to a latent originality factor η that is derived from the above-described models.

Fluid abilities play a critical role in contemporary models of intelligence (e.g., Marrs, 2011; Schneider & McGrew, 2018). In fact, fluid intelligence (Gf) is often considered the most crucial first-order factor of intelligence (Carroll, 1993), and it has even been equated with general intelligence (e.g., Gustafsson, 1984; Kan et al., 2011; Undheim & Gustafsson, 1987). The term “Gf” refers to one's capacity to solve new and unfamiliar problems in not yet experienced situations (e.g., Cattell, 1971). However, performance in prototypical Gf tests (Wilhelm, 2005) is known to depend on more refined cognitive operations, like maintaining, manipulating, and storing single units of information (Wilhelm & Schroeders, 2019). These cognitive operations align with contemporary definitions of working memory, such as the binding hypothesis of working memory (e.g.,Wilhelm et al., 2013). Working memory is understood as a cognitive system that supports non-automatized mental processes by retaining information while, at the same time, also allowing for the processing of new inputs (Conway et al., 2008). With that, working memory can be understood as the fundamental basis of more complex cognitions, like reasoning (e.g., Kane et al., 2005; Kyllonen & Christal, 1990; Oberauer et al., 2005; Wilhelm et al., 2013) or retrieving information from long-term memory (cf. Goecke et al., 2024; Rosen & Engle, 1997). The limited capacity of working memory is often termed “working memory capacity” (WMC; cf. Benchmark 12.1 in Oberauer et al., 2018).

Both Gf and WMC can be considered important for creative ideation. Prototypical creativity tests, or divergent thinking tests, often require a combination of two processes: (a) retrieving existing information about a given problem from long-term memory and (b) restructuring this information and combining it to something novel. This latter aspect in particular may require fluid abilities, although the former aspect also requires fluid abilities like reasoning ability or WMC (Goecke et al., 2024). For example, in tasks like the Alternate Uses Task (Wallach & Kogan, 1965), where individuals are required to come up with creative uses for a brick, individuals must retrieve information in terms of existing knowledge regarding an everyday object from long-term memory (Gilhooly et al., 2007) and combine this information with novel thoughts (Paulus & Brown, 2007).

In this process, working memory might act as a bottleneck. That is, individuals with higher WMC may have a more opportunity to simultaneously consider information that they draw from long-term memory, and, hence, these individuals might have more opportunity to generate new ideas or novel thoughts. Thus, creative ideation, which involves activation of stored knowledge and processing this information into novel outcomes, strains working memory. The ability to retain one piece of information while processing another is a key function of working memory in this context. As Gf is highly dependent on WMC (Kane et al., 2005; Oberauer et al., 2005; Süß et al., 2002), the role of Gf for creative ideation can be understood similarly.

Distinguishing between fluency and originality is theoretically straightforward, but empirically and psychometrically it is much more difficult (e.g., Acar, 2023; Forthmann et al., 2020b; Hocevar, 1979; Weiss et al., 2024). In particular, measuring originality is very challenging, as the methods available to control for fluency, i.e., residualization and ratio scores (Forthmann et al., 2020b; Hocevar, 1979; Hocevar & Michael, 1979), are theoretically (e.g., Cronbach, 1941; Morley, 1930) and empirically (e.g., Arndt et al., 1991; Hocevar, 1979; Hocevar & Michael, 1979) associated with decreases in reliability. Hence, the individual association of originality with fluid abilities like WMC is difficult to grasp. Although previous research has indeed investigated the relationship between these constructs (e.g., Gong et al., 2023), the available studies did not control for a possible fluency-originality contamination (Lu et al., 2022).

As stated above, theoretical and empirical evidence suggests that complex cognitions play a crucial role for creative ideation (cf. Sowden et al., 2015). However, fewer studies have investigated the relationship between WMC and creative ideation, and the available results are conflicting (Benedek et al., 2014; De Dreu et al., 2012, Gerver et al., 2023; Gong et al., 2023; Lee & Therriault, 2013; Lu et al., 2022; Orzechowski et al., 2023; Weiss et al., 2021; Weiss et al., 2024). While meta-analytical studies tend to report a small positive but significant relationship (r = .09, 95% CI [.07, .10], k = 176, m = 29; Gerver et al., 2023; r = .08, 95% CI [.05, .12], k = 75, m = 28; Gong et al., 2023), individual studies often report higher relations between divergent thinking and working memory (e.g., β = .29, Benedek et al., 2014; β = .53 for general cognitive abilities marked by WMC and originality, Weiss et al., 2024). The diversity in these findings might be attributable to different operationalizations in the assessments (e.g., fluency vs. originality, Reiter-Palmon et al., 2019; updating vs. binding, Wilhelm et al., 2013), different scorings, and different data treatments (Gerver et al., 2023; Weiss et al., 2024).

Given the confounding of fluency and originality in divergent thinking tasks, it is difficult to disentangle the degree to which WMC or cognitive abilities in general might affect the quality of ideas (i.e., originality in this context). Given the theoretical importance of WMC in creative ideation and the diverse findings, it is relevant to further investigate this relationship based on the models outlined above.

Aims of the Current Study

In the current study, we extend previous work on the EOB through a latent variable analytic approach and compare competing modeling approaches for the fluency-originality contamination. We compare three different models: an EOB model, a residualized model, and a ratio score model with respect to the interpretation of the latent originality factor η. The three models are evaluated in terms of their fit to two distinct datasets, the reliability of η, and its validity in light of its correlation with cognitive abilities (Dataset 1: Gf, Dataset 2: WMC).

First, we examined whether the EOB model could be fitted to divergent thinking data. Notably, however, previous correlational findings related to the EOB model have been rather heterogeneous (Acar et al., 2023; Forthmann et al., 2020b): Several studies have shown a positive correlation between fluency and ratio scores (e.g., Kleibeuker et al., 2013; Mouchiroud & Lubart, 2001; Plucker et al., 2014), some have shown a negative correlation between fluency and ratio scores (e.g., Plucker et al., 2011; Silvia et al., 2014), and others have shown correlations between fluency and ratio scores that were negligibly different from zero (e.g., Forthmann et al., 2021b). Thus, we did not necessarily expect a good fit of the EOB to the data.

However, in the event that the EOB model did display a reasonable fit to the data, we would then expect that all three models would display an equally well-fitting result. For this situation, we further expected the reliability of η to be rather homogeneous across the models. Second, assuming that η reliably displays an interpretation of originality, we assumed that it would meaningfully correlate with individual differences in cognitive abilities (Gf and WMC). Overall, more heterogeneous findings were anticipated in the event that the EOB did not demonstrate a satisfactory fit to the data. The aim of investigating these research questions empirically was to gain insight into the fundamental question of whether chance models of creativity align with a cognitive ability explanation of individual differences in creative cognition (cf. Rietzschel et al., 2007).

Method

Transparency and Openness

We report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures in the study, and we follow JARS (Appelbaum et al., 2018). We provide all data and materials necessary to reproduce the analyses in an online repository: https://osf.io/tz9ge/. Data were analyzed using R, version 4.4.1 (R Core Team, 2024) and the following packages: lavaan (version 0.6-18; Rosseel, 2012), psych (version 2.4.3; Revelle, 2023), semTools (version 0.5-6; Jorgensen et al., 2022), and tidyverse (version 2.0.0; Wickham et al., 2019). This study's design and its analysis were not pre-registered.

Participants and Procedure

In this work, we make use of secondary data analysis of two openly available datasets.

Dataset 1

The first sample included adolescents from lower secondary and higher secondary schools (public and private; from rural areas and cities) from Austria (Neubauer et al., 2018). The data were provided by Neubauer and colleagues (2018; https://osf.io/v8e5x/). The measures analyzed in the manuscript at hand were part of a larger data collection regarding the self-other knowledge asymmetry. As described by Neubauer and colleagues, the study was approved by the Ethics Committee of the University of Graz (Austria) and the school council of the Austrian province Styria. All students gave informed consent. The final sample consisted of students recruited at 13 schools—after excluding participants with missing values as reported by Neubauer et al.—and included 462 students (55.4% female). Their ages ranged from 13 to 20 years.

Dataset 2

The second sample consisted of adult participants between 18 and 45 years. The study for the second dataset was promoted by multiple ways (e.g., social media, mailing lists, flyers, and advertisements). The measures analyzed in the manuscript at hand were part of a larger multivariate data collection. The computerized cognitive test batteries were administered in group testing sessions in a laboratory (see Goecke et al., 2024, and Weiss et al., 2024, for a more comprehensive description). All participants provided written consent, and the study was approved by the local ethics committee of Ulm University. The final sample we used for statistical analysis consisted of N = 331 participants. These participants were 65.6% female and ranged in age from 18–42 years (M_age = 25.35 years, SD_age = 5.41).

Measures

Dataset 1

Divergent Thinking. Dataset 1 included three items of the Alternate Uses Task (cf. Jauk et al., 2014) as a measure of creative potential (Runco & Acar, 2012). These items are displayed in Table 1. The items were administered using a “be creative” instruction and were all rated for response fluency (count of appropriate answers) and originality (based on the ratings of three human raters). The original authors (Neubauer et al., 2018) reported good inter-rater reliabilities with Cohen's κ = .74 for the originality ratings (M = 1.53, SD = 0.32).

Table 1.

Indicators of Divergent Thinking for Both Datasets.

Dataset	Task	Time Condition	Item
Dataset 1	AUT	2 min	Umbrella
	AUT	2 min	Plastic Bottle
	AUT	2 min	Shoe
Dataset 2	AUT	1 min	Wooden Lath
	AUT	3 min	Towel
	AUT	1 min	Book
	AUT	3 min	Knife

Gf. Additionally, Dataset 1 contained a measure of general intelligence using the Intelligenz-Struktur-Analyse test (ISA; Fay et al., 2001). This test consisted of three subtests that measured spatial (figure completion), verbal (finding commonalities), and numerical (number series) abilities, which can arguably be understood as measures of Gf. Neubauer and colleagues (2018) reported acceptable to good reliability for all three subtests. Scores of the single subtests were subsequently used as indicators for latent variable modeling.

Dataset 2

Divergent Thinking. In the second dataset, divergent thinking was also assessed using an Alternate Uses Task (cf. Weiss et al., 2024). In contrast to Dataset 1, four items were applied for this study. The items are presented in Table 1. Again, the items were administered using a “be creative” instruction, and three trained human raters coded the open-ended answers for fluency and originality. The originality of each produced idea was rated regarding its rareness and uniqueness on a scale from one (“not creative”) to five (“very creative”). The relative agreement between the human raters (ICC3k = 1) was perfect for fluency scores and sufficient for originality scores (ICC3k ranging from .75 to .89; cf. Cicchetti, 2001).

Working Memory Capacity. WMC was measured with three tasks aligned with the binding hypothesis of working memory (cf. Wilhelm et al., 2013). The first task was a spatial-figural updating paradigm, where colored rectangles were displayed and participants were tasked with recalling the final position of each rectangle within a 3 × 3 grid matrix. The second task was a numerical updating paradigm, and presented numbers were continually refreshed on the screen such that participants had to recall the last number for each potential position in every trial. The third task involved a location-letter binding paradigm, where participants were presented with a sequence of letter-position pairings to memorize. For a more comprehensive description of these tasks, see Goecke and colleagues (2024) or Wilhelm et al. (2013). Scoring for all WMC tasks adhered to a partial credit scoring procedure (Conway et al., 2005). Accuracy in each task was determined by calculating the proportion of correct responses relative to the total queries in a trial. These task scores, reflecting mean performance across trials, were subsequently used as indicators in the latent variable analysis.

Analytical Approach

Structural equation modeling (SEM) was carried out with the R package lavaan (Rosseel, 2012). We used full information maximum likelihood estimation under the assumption of missing at random to combine handling of missing data and parameter estimation in a single step (Enders, 2001; Schafer & Graham, 2002). However, participants with missing values on all observed variables were excluded in each of the analyses. A maximum likelihood estimator with robust standard errors (MLR) was used to address potential deviations from multivariate normality.

The following fit statistics were considered to indicate a good model fit: CFI (comparative fit index) ≥ .95, RMSEA (root mean square error of approximation) ≤ .06, and SRMR (standardized root mean square residual) ≤ .08 (Hu & Bentler, 1999; West et al., 2012). For acceptable model fit, these boundaries were used: CFI ≥ .90, RMSEA ≤ .08, and SRMR ≤ .10 (Bentler, 1990; Browne & Cudeck, 1992).

In order to examine the EOB model, item-specific originality scores were regressed on item-specific fluency scores (see Figure 2). The intercepts in these regressions were fixed to a value of zero, and the regression coefficients were constrained to the ratio of average originality and average fluency (cf. Caviggioli & Forthmann, 2022; Forthmann etal., 2021a). In the EOB model, item-specific originality scores were additionally predicted by a latent residual originality factor. For purposes of model identification, the loading of the first item on this originality factor was fixed at a value of one. Next, the residualized model was estimated in a manner analogous to that of the EOB model, with the exception that the intercepts and slopes in the item-specific regressions of originality on fluency were freely estimated (cf. Figure 2); the EOB model represents a special case of the residualized score model. Finally, the ratio score model was estimated as a simple unidimensional factor model with the item-specific ratio scores of originality being used as observed indicators (see Figure 2). Here also, the loading of the first item was fixed to a value of one for model identification purposes. We note that the ratio score model is not a special case of the other two models, and vice versa.

Figure 2.

Schematic presentation of the EOB model, the residualized model, and the ratio score model for dataset 1. Notes. Upper chart: EOB model; middle chart: residualized model; lower chart: ratio score model.

Reliability estimates were based on Raykov's reliability estimates (Raykov, 2001), which we implemented into our syntax and adapted for all evaluated models. For example, the reliability of variation in a residual originality composite score (i.e., after statistically controlling for the fluency contamination effect) can be estimated by the following formula (cf. Raykov, 1997):

R e l (D_{O}) = \frac{{(\sum_{i = 1}^{k} λ_{i})}^{2} \cdot V a r (η)}{{(\sum_{i = 1}^{k} λ_{i})}^{2} \cdot V a r (η) + \sum_{i = 1}^{k} V a r (ε_{i})},

with D_O being the composite of residual originality scores.

Results

Descriptive statistics and bivariate manifest correlations can be found in the OSF repository (https://osf.io/tz9ge/). First, we examined the three different modeling approaches that all account for fluency contamination in originality. The path models illustrating the EOB model, the residualized model, and the ratio score model (see Figure 1) are schematically displayed for Dataset 1 in Figure 2. For Dataset 2, these models are expanded by one item, thus including four items in total (see Table 1). In all models, a latent variable η is estimated displaying different interpretations of originality. As described in the introduction, in the EOB model, η displays the remaining originality after accounting for what can be expected by fluency; in the residualized model, η represents the pure originality, free from fluency; and in the ratio score model, η represents originality that is weighted by fluency.

In Table 2 we display the fit indices for the competing models (illustrated for Dataset 1 in Figure 2). Furthermore, we present a reliability coefficient of the latent originality factor η that is based on the adaptation of Raykov's reliability estimate (Raykov, 1997). As displayed in Table 2, the EOB model does not fit the data satisfactorily in Dataset 1; however, the fit of the EOB model is acceptable in Dataset 2. The residualized model and the ratio score model fitted the data well in both datasets. In Dataset 1, the reliability of η in the residualized model (.65) was lower than in the ratio score model (.69). Given the misfit of the EOB model in Dataset 1, the reliability of η was overestimated (.70). The reliability estimates of η in Dataset 2 show less heterogeneous results across the three models compared to the results in Dataset 1.

Table 2.
Measurement Models for Dataset 1 and Dataset 2.

Dataset N Model χ² (df) CFI RMSEA SRMR ρ_η

1 459 EOB 189.13 (9) .924 .209 .107 .696

Residualized 39.30 (6) .986 .110 .028 .650

Ratio Score 0 1 0 0 .692

2 330 EOB 74.26 (18) .974 .097 .058 .707

Residualized 54.43 (14) .982 .094 .052 .700

Ratio Score 3.56 (2) .991 .049 .021 .676

Notes. ρ_η = reliability coefficient. The ratio score model is saturated in Dataset 1 due to the number of indicators (see Figure 2).

Second, we investigated the relationship between the latent originality factor η and Gf and WMC in all three models for each dataset, respectively. This is schematically shown in Figure 3. In Dataset 1, the relationship between η in the ratio score model and Gf is higher than the relationship between originality and fluid abilities in the models that disentangle fluency and originality (EOB model and residualized model). The same pattern can be found in Dataset 2, for the relationship between originality and WMC. In sum, the relationship between η and WMC is somewhat lower than between η and fluid abilities, but both relationships are statistically meaningful.

Figure 3.
Latent correlations between η and gf in dataset 1 and between η and WMC in dataset 2 for each of the proposed models (standard errors in parenthesis).

Discussion

In this manuscript, we examined the compatibility of a cognitive psychological interpretation of creativity with chance models of creativity, which had been previously questioned (e.g., Rietzschel et al., 2007). Using latent variable modeling, we conducted a comparative analysis between an EOB model (i.e., a chance model of creativity) and two alternative models, a residualized model and a ratio score model. Our findings suggest that, contingent upon the EOB model's capacity to align with the data, the overall results are likely to exhibit a high degree of comparability. It is precisely in this situation that we posited the existence of a reliable latent originality factor, denoted as η, which shares a substantial amount of variance with Gf or WMC. Consequently, our approach diverges from previous work that presumed a fundamental incompatibility between chance models of creativity and psychological cognitive interpretations.

Regarding our first research question, we found that the EOB model could not be fitted to the Dataset 1. Conversely, the other two models (residualized model and ratio score model) fitted this dataset well. The misfit of the EOB model in Dataset 1 can be traced back to manifest positive correlations between T and H/T (see the OSF repository: https://osf.io/tz9ge/), which are not compatible with the underlying assumptions of the EOB model. In this context, the EOB model's partialling out of fluency is less pronounced than in the residualized model. Consequently, this resulted in an overestimation of the reliability η in the EOB model relative to the residualized model in Dataset 1. The reliability estimates of η in the residualized model and the ratio score model in Dataset 1 exhibit slight discrepancies from one another, which can be attributed to the partialling out of construct-relevant variance in the residualized model, which is not partialled out in the ratio score model. As a result, the ratio score model displayed higher reliability. In Dataset 2, all three models fitted the data comparably well (with the RMSEA as a notable exception), leading to less heterogeneous patterns of results across the three models (see, for example, the reliability estimates for η).

Regarding the second research question, we analyzed whether the different interpretations of originality factor η correlated meaningfully with Gf and WMC and, thus, with basic factors of cognitive ability. We found that η was significantly positively correlated with both constructs in all models. In fact, all analyses yielded moderate to large correlations. However, the EOB model did not fit Dataset 1 well, and the correlations with Gf should be better interpreted based on the other two models. In the residualized model, η (i.e., the pure originality factor) correlated strongly with Gf, but this correlation was less strong as compared to the ratio score model. This difference in effect sizes can be attributed to the fact that more construct-relevant variance is removed from η in the residualized model when the EOB does not fit well due to positive correlations between fluency and manifest ratio scores. In this situation, a cognitive interpretation of the EOB model is not straightforward. In Dataset 2, where the EOB model fitted reasonably well, we found that WMC was similarly related to η in all three models. In sum, Dataset 2 shows that about 21% of variance in η is shared by WMC, implying that a cognitive interpretation is not at odds with equal odds.

Are Chance Models of Creativity Compatible with a Cognitive Interpretation of Individual Differences in Creative Cognition?

The EOB model can be estimated using structural equation modeling (Forthmann et al., 2021a; Forthmann et al., 2021c), and in this paper we proposed a latent variable extension of the EOB for its application to divergent thinking tasks. The poor fit in Dataset 1 and the mostly good fit of the EOB model in Dataset 2 calls into question the generalizability of the EOB model across different item sets and datasets. Nevertheless, Dataset 2 suggests that a cognitive interpretation of η in the EOB model is possible and meaningful.

Hence, our theoretical arguments and empirical findings support the notion that when a chance model of creativity is deemed to be an appropriate fit for a given dataset, as evidenced by Dataset 2 in our study, a cognitive interpretation of η is warranted. Conversely, when a chance model is deemed to be an inappropriate fit for a given dataset, as evidenced by Dataset 1 in our study, a cognitive interpretation of η is not straightforwardly applicable—at least not in the context of the EOB. Of course, a cognitive interpretation of individual differences in η further requires that there is substantial variance of the latent variable. In other words, the EOB model should not reduce to a strict EOB model in which residual variation can be solely explained by sampling error (Forthmann et al., 2021c). Thus, our work contributes to and refines related discussions that have previously viewed chance models of creativity as being incompatible with such a cognitive interpretation (Rietzschel et al., 2007).

Furthermore, we found the ratio score model provided the best model fit across both datasets. Notably, the lower reliability in the residualized model suggests that attempting to isolate pure originality might introduce measurement instability. This calls into question the appropriateness of removing fluency (residualized model) or removing what can be explained by fluency (EOB model), in turn raising the question of whether originality is more appropriately measured when fluency is accounted for but not entirely removed of (e.g., originality weighted by fluency, as specified in the ratio score model). Nonetheless, the deterioration of model fit for the residualized and EOB models was small, indicating that removing fluency-originality contamination is possible, reliable, and interpretable.

(Pure) Originality and Cognitive Abilities

First, and importantly, the correlations between the specified originality factors and cognitive abilities (Gf and WMC) were of similar magnitude across models and datasets. This implies that cognitive abilities are important for being original, regardless of whether we consider originality as being free of fluency or whether originality is weighted by fluency. Based on three competing models and two datasets, we showed that the impact of cognitive abilities on different interpretations of originality might be even larger than previously expected (Benedek et al., 2014, De Dreu et al., 2012; Weiss et al., 2021). However, the stronger relationship between η (in the ratio score model) and fluid abilities emphasizes the importance of considering fluency when examining cognitive correlates of originality.

In this sense, our findings suggest that the ratio score can be interpreted as an indicator of the efficiency of the ideational process. As such, by failing to account for the role of fluency in this process, the disentangling models (EOB and residualized model) may underestimate the cognitive connections of originality. However, prior work did aim to model individual differences in hit rates by adding the residual term to the EOB (Simonton, 2004, 2010), and differences in results between the EOB and the ratio score model (when the EOB model fits a given dataset reasonably well) could be due to technical reasons that are not yet fully understood. Thus, future methodological work might be needed on the latent variable extension of the EOB.

Finally, we note that the less strong but meaningful relationship between WMC and originality (in comparison to the relationship between Gf and originality) suggests that while originality is linked to cognitive abilities, fluid reasoning might be a more critical factor than working memory in the creative process. On the other hand, the correlation between originality and Gf might be inflated because crystallized affordances are suspected to be higher in verbal and numerical Gf tests than in visual WMC tests (cf. Wilhelm & Kyllonen, 2021).

Limitations

Our study compared three different measurement model approaches for addressing the fluency-originality contamination in two datasets that included different Alternate Uses Tasks. We used different datasets and stimuli to ensure generalizability of our results. However, we note that the generalizability is limited to the Alternate Uses Task, and the misfit of the EOB model in Dataset 1 suggests that the model may not generalize across different sets of task prompts. Therefore, future studies should further generalize to other stimuli, especially to other divergent thinking tasks, because limiting a construct to one task (as here, where alternate uses was the only marker for creative cognition) is always questionable (e.g., Campbell & Fiske, 1959; Saretzki et al., 2024).

Furthermore, the interpretation of the latent originality factor η varies between the models (e.g., pure originality without fluency in the residualized model). Despite this varying interpretation of η, the range of correlations of η with cognitive abilities between the models was quite similar (e.g., Dataset 2 ranging from .45 to .53). This implies that the influence of cognitive abilities does not vary largely for different interpretations of originality. Future studies should, therefore, investigate further covariates where the differences in interpretation of η between the models lead to greater variance in the correlation with the covariate. Such covariates could include personality as well as covariates that have a focus on originality, such as creative achievements (e.g., Diedrich et al., 2018).

Conclusion

Our results indicate that while the EOB model did not fit reasonably well to all datasets, the ratio score model demonstrated the most robust correlations with cognitive abilities, perhaps underscoring the pivotal role of fluency in conjunction with originality, which collectively indicate the efficiency of an ideational process. However, the fact that a residualized model also fitted the data well in both datasets leads to the conclusion that originality may be meaningfully displayed free of fluency (depending on the reliability of the latent factor). Thus, depending on the research question one wishes to investigate, one might even choose a residualized model. We argue that theoretical deliberations should always guide the decision of whether construct-relevant variance needs to be partialled out.

Overall, this research advances the field's understanding of how fluency and originality in creative thinking can be modeled within a chance model of creativity (and related models) and demonstrates that cognitive abilities significantly relate to individual differences in creative performance. Future research should aim to refine these models further and explore their applicability across diverse creative tasks and populations, while also considering practical implications for assessing and fostering creativity in educational and organizational settings. For example, a timely question would be to evaluate whether individual differences in creative cognition persist and can be modeled and explained in accordance with the EOB, particularly in light of the potential impact of homogenization on creative performance in the age of artificial intelligence (Anderson et al., 2024).

Dataset	N	Model	χ² (df)	CFI	RMSEA	SRMR	ρ_η
1	459	EOB	189.13 (9)	.924	.209	.107	.696
Residualized	39.30 (6)	.986	.110	.028	.650
Ratio Score	0	1	0	0	.692
2	330	EOB	74.26 (18)	.974	.097	.058	.707
Residualized	54.43 (14)	.982	.094	.052	.700
Ratio Score	3.56 (2)	.991	.049	.021	.676

Footnotes

Acknowledgements

We thank Amina Rajakumar, Jasmin Thelen, and Tim Trautwein for their support in data collection for Dataset 2. We are thankful for the possibility to analyze the dataset published by Neubauer and colleagues (2018), and we thank Aljoscha Neubauer, Anna Pribil, Alexandra Wallner, and Gabriela Hofer for allowing a publication of the results. We thank Celeste Brennecka for proofreading of our manuscript.

Data Availability

All data-analytic code and the according data to reproduce the current results is available via the Open Science Framework:

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The research project has been funded by Ulm University (Graduate and Professional Training Center Ulm, 029/126/P/IIP).

Ethical Approval and Informed Consent Statements

This work relies on secondary data analysis of openly available datasets and an ethical approval was thus not required. The primary studies from which the data were taken, however, were approved by local ethics committees (cf. Neubauer et al., 2018; Weiss et al., 2024). In both primary studies participants either gave written informed consent to participate (Weiss et al., 2024) or fulfilled the requirements for consent in accordance with country-specific laws (Neubauer et al., 2018).

ORCID iD

Boris Forthmann

Author Biographies

Boris Forthmann is a post-doctoral researcher in Psychology. His area of research is creativity assessment focusing on creative thinking and scientific productivity.

Selina Weiss is a professor of Psychological Assessment. Her main research interest is in the area of creativity assessment with a focus on the assessment of creative abilities and methodological developments in the area of assessment, item selection and person selection.

Benjamin Goecke is a postdoctoral researcher. His recent research efforts focus on understanding the structure and impact of individual differences in cognitive abilities, including creativity, as well as refining psychometric methods and assessment techniques.

References

1.
Acar S. (2023). Does the task structure impact the fluency confound in divergent thinking? An investigation with TTCT-figural. Creativity Research Journal, 35(1), 1–14. https://doi.org/10.1080/10400419.2022.2044656

2.
Acar S. Ogurlu U. Zorychta A. (2023). Exploration of discriminant validity in divergent thinking tasks: A meta-analysis. Psychology of Aesthetics, Creativity, and the Arts, 17(6), 705–724. https://doi.org/10.1037/aca0000469

3.
Amabile T. M. Pratt M. G. (2016). The dynamic componential model of creativity and innovation in organizations: Making progress, making meaning. Research in Organizational Behavior, 36, 157–183. https://doi.org/10.1016/j.riob.2016.10.001

4.
Anderson B. R. Shah J. H. Kreminski M. (2024, June). Homogenization effects of large language models on human creative ideation. In Proceedings of the 16th conference on creativity & cognition (pp. 413–425). Association for Computing Machinery. https://doi.org/10.1145/3635636.3656204

5.
Appelbaum M. Cooper H. Kline R. B. Mayo-Wilson E. Nezu A. M. Rao S. M. (2018). Journal article reporting standards for quantitative research in psychology: The APA publications and communications board task force report. American Psychologist, 73(1), 3–25. https://doi.org/10.1037/amp0000191

6.
Arndt S. Cohen G. Alliger R. J. Swayze I. I. V. W. Andreasen N. C. (1991). Problems with ratio and proportion measures of imaged cerebral structures. Psychiatry Research: Neuroimaging, 40(1), 79–89. https://doi.org/10.1016/0925-4927(91)90031-K

7.
Barbot B. (2018). The dynamics of creative ideation: Introducing a new assessment paradigm. Frontiers in Psychology, 9, 2529. https://doi.org/10.3389/fpsyg.2018.02529

8.
Benedek M. Jauk E. Sommer M. Arendasy M. Neubauer A. C. (2014). Intelligence, creativity, and cognitive control: The common and differential involvement of executive functions in intelligence and creativity. Intelligence, 46(1), 73–83. https://doi.org/10.1016/j.intell.2014.05.007

9.
Bentler P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238–246. https://doi.org/10.1037/0033-2909.107.2.238

10.
Briggs R. O. Reinig B. A. (2010). Bounded ideation theory. Journal of Management Information Systems, 27(1), 123–144. https://doi.org/10.2753/MIS0742-1222270106

11.
Browne M. W. Cudeck R. (1992). Alternative ways of assessing model fit. Sociological Methods & Research, 21(2), 230–258. https://doi.org/10.1177/0049124192021002005

12.
Campbell D. T. Fiske D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56(2), 81–105. https://doi.org/10.1037/h0046016

13.
Carroll J. B. (1993). Human Cognitive Abilities: A Survey of Factor-Analytic Studies. Cambridge University Press. https://doi.org/10.1017/CBO9780511571312.

14.
Cattell R. B. (1971). Abilities: Their structure, growth, and action. Houghton Mifflin.

15.
Caviggioli F. Forthmann B. (2022). Reach for the stars: Disentangling quantity and quality of inventors’ productivity in a multifaceted latent variable model. Scientometrics, 127(12), 7015–7040. https://doi.org/10.1007/s11192-022-04328-3

16.
Cicchetti D. V. (2001). Methodological commentary the precision of reliability and validity estimates Re-visited: Distinguishing between clinical and statistical significance of sample size requirements. Journal of Clinical and Experimental Neuropsychology, 23(5), 695–700. https://doi.org/10.1076/jcen.23.5.695.1249

17.
Conway A. R. A. Jarrold C. Kane M. J. Miyake A. Towse J. (2008). Variation in working memory: An introduction. In Conway A. R. A. Jarrold C. Kane M. Miyake A. Towse J. (Eds.), Variation in working memory (1st ed., pp. 3–18). Oxford University Press.

18.
Conway A. R. A. Kane M. J. Bunting M. F. Hambrick D. Z. Wilhelm O. Engle R. W. (2005). Working memory span tasks: A methodological review and user's guide. Psychonomic Bulletin & Review, 12(5), 769–786. https://doi.org/10.3758/BF03196772

19.
Cronbach L. J. (1941). The reliability of ratio scores. Educational and Psychological Measurement, 1(1), 269–277. https://doi.org/10.1177/001316444100100121

20.
De Dreu C. K. Nijstad B. A. Baas M. Wolsink I. Roskes M. (2012). Working memory benefits creative insight, musical improvisation, and original ideation through maintained task-focused attention. Personality and Social Psychology Bulletin, 38(5), 656–669. https://doi.org/10.1177/0146167211435795

21.
Diedrich J. Benedek M. Jauk E. Neubauer A. C. (2015). Are creative ideas novel and useful? Psychology of Aesthetics, Creativity, and the Arts, 9(1), 35–40. https://doi.org/10.1037/a0038688

22.
Diedrich J. Jauk E. Silvia P. J. Gredlein J. M. Neubauer A. C. Benedek M. (2018). Assessment of real-life creativity: The inventory of creative activities and achievements (ICAA). Psychology of Aesthetics, Creativity, and the Arts, 12(3), 304–316. https://doi.org/10.1037/aca0000137

23.
Enders C. K. (2001). The impact of nonnormality on full information maximum-likelihood estimation for structural equation models with missing data. Psychological Methods, 6(4), 352–370. https://doi.org/10.1037/1082-989X.6.4.352

24.
Fay E. Trost G. Gittler G. (2001). Intelligenz-Struktur-Analyse (ISA). Swets Test Services.

25.
Feist G. J. (1997). Quantity, quality, and depth of research as influences on scientific eminence: Is quantity most important? Creativity Research Journal, 10(4), 325–335. https://doi.org/10.1207/s15326934crj1004_4

26.
Fink A. Benedek M. (2014). EEG Alpha power and creative ideation. Neuroscience & Biobehavioral Reviews, 44, 111–123. https://doi.org/10.1016/j.neubiorev.2012.12.002

27.
Forthmann B. Leveling M. Dong Y. Dumas D. (2020a). Investigating the quantity–quality relationship in scientific creativity: An empirical examination of expected residual variance and the tilted funnel hypothesis. Scientometrics, 124(3), 2497–2518. https://doi.org/10.1007/s11192-020-03571-w

28.
Forthmann B. Szardenings C. Dumas D. (2021a). On the conceptual overlap between the fluency contamination ef-fect in divergent thinking scores and the chance view on scientific creativity. The Journal of Creative Behavior, 55(1), 268–275. https://doi.org/10.1002/jocb.445

29.
Forthmann B. Szardenings C. Dumas D. (2021b). Testing equal odds in creativity research. Psychology of Aesthetics, Creativity, and the Arts, 15(2), 324–339. https://doi.org/10.1037/aca0000294

30.
Forthmann B. Szardenings C. Dumas D. Feist G. J. (2021c). Strict equal odds: A useful reference to study the relationship between quality and quantity. Creativity Research Journal, 33(2), 96–105. https://doi.org/10.1080/10400419.2020.1827605

31.
Forthmann B. Szardenings C. Holling H. (2020b). Understanding the confounding effect of fluency in divergent thinking scores: Revisiting average scores to quantify artifactual correlation. Psychology of Aesthetics, Creativity, and the Arts, 14(1), 94–112. https://doi.org/10.1037/aca0000196

32.
Gerver C. R. Griffin J. W. Dennis N. A. Beaty R. E. (2023). Memory and creativity: A meta-analytic examination of the relationship between memory systems and creative cognition. Psychonomic Bulletin & Review, 30(6), 2116–2154. https://doi.org/10.3758/s13423-023-02303-4

33.
Gerwig A. Miroshnik K. Forthmann B. Benedek M. Karwowski M. Holling H. (2021). The relationship between intelligence and divergent thinking—A meta-analytic update. Journal of Intelligence, 9(2), 23. https://doi.org/10.3390/jintelligence9020023

34.
Gilhooly K. J. Fioratou E. Anthony S. H. Wynn V. (2007). Divergent thinking: Strategies and executive involvement in generating novel uses for familiar objects. British Journal of Psychology, 98(4), 611–625. https://doi.org/10.1111/j.2044-8295.2007.tb00467.x

35.
Goecke B. Weiss S. Wilhelm O. (2024). Driving factors of individual differences in broad retrieval ability: Gr is more than the sum of its parts. Journal of Experimental Psychology: Learning, Memory, and Cognition, Advance online publication. https://doi.org/10.1037/xlm0001336

36.
Gong Z. Miao K. Liu X. Luo M. Yu Y. Chen Z. (2023). A positive association between working memory capacity and human creativity: A meta-analytic evidence. Journal of Intelligence, 11(1), 15. https://doi.org/10.3390/jintelligence11010015

37.
Gustafsson J.-E. (1984). A unifying model for the structure of intellectual abilities. Intelligence, 8(3), 179–203. https://doi.org/10.1016/0160-2896(84)90008-4

38.
Hass R. W. Weisberg R. W. (2015). Revisiting the 10-year rule for composers from the great American songbook: On the validity of two measures of creative production. Psychology of Aesthetics, Creativity, and the Arts, 9(4), 471–479. https://doi.org/10.1037/aca0000021

39.
Hocevar D. (1979). Ideational fluency as a confounding factor in the measurement of originality. Journal of Educational Psychology, 71(2), 191–196. https://doi.org/10.1037/0022-0663.71.2.191

40.
Hocevar D. Michael W. B. (1979). The effects of scoring formulas on the discriminant validity of tests of divergent thinking. Educational and Psychological Measurement, 39(4), 917–921. https://doi.org/10.1177/001316447903900427

41.
Hu L. Bentler P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

42.
Jauk E. Benedek M. Neubauer A. C. (2014). The road to creative achievement: A latent variable model of ability and personality predictors. European Journal of Personality, 28(1), 105–195. https://doi.org/10.1002/per.1941

43.
Jorgensen T. D. Pornprasertmanit S. Schoemann A. M. Rosseel Y. (2022). semTools: Useful tools for structural equation modeling. R package version 0.5-6. https://CRAN.R-project.org/package=semTools

44.
Kan K.-J. Kievit R. A. Dolan C. Der Maas H. V. (2011). On the interpretation of the CHC factor gc. Intelligence, 39(5), 292–302. https://doi.org/10.1016/j.intell.2011.05.003

45.
Kane M. J. Hambrick D. Z. Conway A. R. A. (2005). Working memory capacity and fluid intelligence are strongly related constructs: Comment on ackerman, beier, and boyle (2005). Psychological Bulletin, 131(1), 66–71. https://doi.org/10.1037/0033-2909.131.1.66

46.
Kaufman J. C. Plucker J. A. Baer J. (2008). Essentials of creativity assessment. Wiley.

47.
Kleibeuker S. W. De Dreu C. K. W. Crone E. A. (2013). The development of creative cognition across adolescence: Distinct trajectories for insight and divergent thinking. Developmental Science, 16(1), 2–12. https://doi.org/10.1111/j.1467-7687.2012.01176.x

48.
Kozbelt A. (2005). Factors affecting aesthetic success and improvement in creativity: A case study of the musical genres of mozart. Psychology of Music, 33(3), 235–255. https://doi.org/10.1177/0305735605053732

49.
Kozbelt A. (2008). Longitudinal hit ratios of classical composers: Reconciling „darwinian“ and expertise acquisition perspectives on lifespan creativity. Psychology of Aesthetics, Creativity, and the Arts, 2(4), 221–235. https://doi.org/10.1037/a0012860

50.
Kyllonen P. C. Christal R. E. (1990). Reasoning ability is (little more than) working-memory capacity?!. Intelligence, 14(4), 389–433. https://doi.org/10.1016/S0160-2896(05)80012-1

51.
Lee C. S. Therriault D. J. (2013). The cognitive underpinnings of creative thought: A latent variable analysis exploring the roles of intelligence and working memory in three creative thinking processes. Intelligence, 41(5), 306–320. https://doi.org/10.1016/j.intell.2013.04.008

52.
Lu R. Zhang Y. Bao N. Su M. Zhang X. Shi J. (2022). Visuospatial, rather than verbal working memory capacity plays a key role in verbal and figural creativity. Thinking & Reasoning, 28(1), 29–60. https://doi.org/10.1080/13546783.2021.1911848

53.
Marrs H. (2011). Catell-Horn-Carroll theory of intelligence. In Goldstein S. Naglieri J. A. (Hrsg.), Encyclopedia of child behavior and development (pp. S. 314–S. 315). Springer US. https://doi.org/10.1007/978-0-387-79061-9_472

54.
Morley C. A. (1930). The reliability of the achievement quotient. Journal of Educational Psychology, 21(5), 351–360. https://doi.org/10.1037/h0072259

55.
Mouchiroud C. Lubart T. (2001). Children’s original thinking: An empirical examination of alternative measures derived from divergent thinking tasks. The Journal of Genetic Psychology, 162(4), 382–401. https://doi.org/10.1080/00221320109597491

56.
Neubauer A. C. Pribil A. Wallner A. Hofer G. (2018). The self–other knowledge asymmetry in cognitive intelligence, emotional intelligence, and creativity. Heliyon, 4(12), e01061. https://doi.org/10.1016/j.heliyon.2018.e01061

57.
Nijstad B. A. De Dreu C. K. W. Rietzschel E. F. Baas M. (2010). The dual pathway to creativity model: Creative ideation as a function of flexibility and persistence. European Review of Social Psychology, 21(1), 34–77. https://doi.org/10.1080/10463281003765323

58.
Oberauer K. Lewandowsky S. Awh E. Brown G. D. A. Conway A. Cowan N. Donkin C. Farrell S. Hitch G. J. Hurlstone M. J. Ma W. J. Morey C. C. Nee D. E. Schweppe J. Vergauwe E. Ward G. (2018). Benchmarks for models of short-term and working memory. Psychological Bulletin, 144(9), 885–958. https://doi.org/10.1037/bul0000153

59.
Oberauer K. Schulze R. Wilhelm O. Süß H.-M. (2005). Working memory and intelligence–their correlation and their relation: Comment on ackerman, beier, and boyle (2005). Psychological Bulletin, 131(1), 61–65. https://doi.org/10.1037/0033-2909.131.1.61

60.
Orzechowski J. Gruszka A. Michalik K. (2023). The impact of working memory on divergent thinking flexibility. Thinking & Reasoning, 29(4), 643–662. https://doi.org/10.1080/13546783.2022.2109730

61.
Osborn A. F. (1963). Applied imagination: Principles and procedures of creative problem-solving. Scribner.

62.
Paulus P. B. Brown V. R. (2007). Toward more creative and innovative group idea generation: A cognitive-social-motivational perspective of brainstorming. Social and Personality Psychology Compass, 1(1), 248–265. https://doi.org/10.1111/j.1751-9004.2007.00006.x

63.
Pichot N. Forthmann B. Bonetto E. Arciszewski T. Bonnardel N. Jaubert S. Pavani J. B. (2024). Aiming at creativity and ending up with a range from low-hanging fruits to foolishness: A reflective model of creativity. The Journal of Creative Behavior, 58(3), 444–459. https://doi.org/10.1002/jocb.667

64.
Plucker J. A. Qian M. Schmalensee S. L. (2014). Is what you see what you really get? Comparison of scoring techniques in the assessment of real-world divergent thinking. Creativity Research Journal, 26(2), 135–143. https://doi.org/10.1080/10400419.2014.901023

65.
Plucker J. A. Qian M. Wang S. (2011). Is originality in the eye of the beholder? Comparison of scoring techniques in the assessment of divergent thinking. The Journal of Creative Behavior, 45(1), 1–22. https://doi.org/10.1002/j.2162-6057.2011.tb01081.x

66.
Raykov T. (1997). Estimation of composite reliability for congeneric measures. Applied Psychological Measurement, 21(2), 173–184. https://doi.org/10.1177/01466216970212006

67.
Raykov T. (2001). Estimation of congeneric scale reliability using covariance structure analysis with nonlinear constraints. British Journal of Mathematical and Statistical Psychology, 54(2), 315–323. https://doi.org/10.1348/000711001159582

68.
R Core Team (2024). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

69.
Reiter-Palmon, R., Forthmann, B., & Barbot, B. (2019). Scoring divergent thinking tests: A review and systematic framework. Psychology of Aesthetics, Creativity, and the Arts, 13(2), 144–152. https://doi.org/10.1037/aca0000227

70.
Revelle W. (2023). psych: Procedures for psychological, psychometric, and personality research. Northwestern University, Evanston, Illinois. R package version 2.4.3. https://CRAN.R-project.org/package=psych

71.
Rietzschel E. F. Nijstad B. A. Stroebe W. (2007). Relative accessibility of domain knowledge and creativity: The effects of knowledge activation on the quantity and originality of generated ideas. Journal of Experimental Social Psychology, 43(6), 933–946. https://doi.org/10.1016/j.jesp.2006.10.014

72.
Rosen V. M. Engle R. W. (1997). The role of working memory capacity in retrieval. Journal of Experimental Psychology: General, 126(3), 211–227. https://doi.org/10.1037/0096-3445.126.3.211

73.
Rosseel Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

74.
Runco M. A. (2020). Divergent thinking. In Encyclopedia of creativity (pp. S. 356–S. 361). Elsevier. https://doi.org/10.1016/B978-0-12-809324-5.23824-8

75.
Runco M. A. Acar S. (2012). Divergent thinking as an indicator of creative potential. Creativity Research Journal, 24(1), 66–75. https://doi.org/10.1080/10400419.2012.652929

76.
Runco M. A. Jaeger G. J. (2012). The standard definition of creativity. Creativity Research Journal, 24(1), 92–96. https://doi.org/10.1080/10400419.2012.650092

77.
Saretzki, J., Forthmann, B., & Benedek, M. (2024). A systematic quantitative review of divergent thinking assessments. Psychology of Aesthetics, Creativity, and the Arts. Advance online publication. https://doi.org/10.1037/aca0000691

78.
Schafer J. L. Graham J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177. https://doi.org/10.1037/1082-989X.7.2.147

79.
Schneider W. J. McGrew K. S. (2018). The Cattell–Horn–Carroll theory of cognitive abilities. In Flanagan D. P. McDonough E. M. (Hrsg.), Contemporary intellectual assessment: Theories, tests, and issues (pp. S. 73–S. 163). The Guilford Press.

80.
Silvia P. J. Beaty R. E. Nusbaum E. C. Eddington K. M. Kwapil T. R. (2014). Creative motivation: Creative achievement predicts cardiac autonomic markers of effort during divergent thinking. Biological Psychology, 102, 30–37. https://doi.org/10.1016/j.biopsycho.2014.07.010

81.
Simonton D. K. (1988). Scientific genius: A psychology of science. Cambridge University Press.

82.
Simonton D. K. (2003). Scientific creativity as constrained stochastic behavior: The integration of product, person, and process perspectives. Psychological Bulletin, 129(4), 475–494. https://doi.org/10.1037/0033-2909.129.4.475

83.
Simonton D. K. (2004). Creativity in Science: Chance, Logic, Genius, and Zeitgeist (1. Aufl.). Cambridge University Press. https://doi.org/10.1017/CBO9781139165358.

84.
Simonton D. K. (2010). Creative thought as blind-variation and selective-retention: Combinatorial models of exceptional creativity. Physics of Life Reviews, 7(2), 156–179. https://doi.org/10.1016/j.plrev.2010.02.002

85.
Sinatra R. Wang D. Deville P. Song C. Barabási A.-L. (2016). Quantifying the evolution of individual scientific impact. Science, 354(6312), aaf5239. https://doi.org/10.1126/science.aaf5239

86.
Sowden P. T. Pringle A. Gabora L. (2015). The shifting sands of creative thinking: Connections to dual-process theory. Thinking & Reasoning, 21(1), 40–60. https://doi.org/10.1080/13546783.2014.885464

87.
Süß H. M. Oberauer K. Wittmann W. W. Wilhelm O. Schulze R. (2002). Working-memory capacity explains reasoning ability—and a little bit more. Intelligence, 30(3), 261–288. https://doi.org/10.1016/S0160-2896(01)00100-3

88.
Undheim J. O. Gustafsson J.-E. (1987). The hierarchical organization of cognitive abilities: Restoring general intelligence through the use of linear structural relations (LISREL). Multivariate Behavioral Research, 22(2), 149–171. https://doi.org/10.1207/s15327906mbr2202_2

89.
Wallach, M. A., & Kogan, N. (1965). Modes of thinking in young children: A study of the creativity-intelligence distinction. Holt, Rinehart & Winston.

90.
Weiss S. Goecke B. Wilhelm O. (2024). How much retrieval ability is in originality? The Journal of Creative Behavior, 58(3), 370–387. https://doi.org/10.1002/jocb.659

91.
Weiss S. Steger D. Kaur Y. Hildebrandt A. Schroeders U. Wilhelm O. (2021). On the trail of creativity: Dimensionality of divergent thinking and its relation with cognitive abilities, personality, and insight. European Journal of Personality, 35(3), 291–314. https://doi.org/10.1002/per.2288

92.
West S. G. Taylor A. B. Wu W. (2012). Model fit and model selection in structural equation modeling. In Hoyle R. H. (Hrsg.), Handbook of structural equation modeling (pp. S. 209–S. 231). Guilford.

93.
Wickham H. Averick M. Bryan J. Chang W. McGowan L. D. A. François R. Yutani H. (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686. https://doi.org/10.21105/joss.01686

94.
Wilhelm O. (2005). Measuring reasoning ability. In Wilhelm O. Engle R. W. (Eds.), Handbook of understanding and measuring intelligence (pp. 373–392). Sage Publications, Inc. https://doi.org/10.4135/9781452233529.n21

95.
Wilhelm O. Hildebrandt A. Oberauer K. (2013). What is working memory capacity, and how can we measure it? Frontiers in Psychology, 4, 433. https://doi.org/10.3389/fpsyg.2013.00433

96.
Wilhelm O. Kyllonen P. (2021). To predict the future, consider the past: Revisiting carroll (1993) as a guide to the future of intelligence research. Intelligence, 89, 101585. https://doi.org/10.1016/j.intell.2021.101585

97.
Wilhelm O. Schroeders U. (2019). Intelligence. In Sternberg R. Funke J. (Eds.), The psychology of human thought: An Introduction (pp. 257–277). Heidelberg University Publishing.

98.
Zhou J. Wang X. M. Song L. J. Wu J. (2017). Is it new? Personal and contextual influences on perceptions of novelty and creativity. Journal of Applied Psychology, 102(2), 180–202. https://doi.org/10.1037/apl0000166