is creativity without intelligence possible? a necessary ......1.1. a look at the...

Intelligence xxx (2016) xxx–xxx

INTELL-01129; No of Pages 13

Contents lists available at ScienceDirect

Intelligence

Is creativity without intelligence possible? A Necessary Condition Analysis

Maciej Karwowski a,⁎, Jan Dul b, Jacek Gralewski a, Emanuel Jauk c, Dorota M. Jankowska a, Aleksandra Gajda a,Michael H. Chruszczewski d, Mathias Benedek c

a The Maria Grzegorzewska University, Warsaw, Polandb Erasmus University, The Netherlandsc University of Graz, Austriad University of Warsaw, Poland

⁎ Corresponding author at: The Maria Grzegorzewska02-353 Warsaw, Poland.

E-mail address: [email protected] (M. Karwow

http://dx.doi.org/10.1016/j.intell.2016.04.0060160-2896/© 2016 Elsevier Inc. All rights reserved.

Please cite this article as: Karwowski, M., ethttp://dx.doi.org/10.1016/j.intell.2016.04.00

a b s t r a c t
a r t i c l e i n f o
Article history:Received 22 January 2016Received in revised form 7 April 2016Accepted 26 April 2016Available online xxxx

This article extends the previous studies on the relationship between intelligence and creativity by providing anewmethodology and an empirical test of the hypothesis that intelligence is a necessary condition for creativity.Unlike the classic threshold hypothesis, which assumes the existence of a curvilinear relationship between intel-ligence and creativity, theNecessary Condition Analysis (Dul, 2016) focuses on and quantifies the overall shape ofthe relationship between intelligence and creativity. In eight studies (totalN=12,255), using differentmeasuresof intelligence and creativity, we observed a consistent pattern that supports the necessary-but-not-sufficient re-lationship between these two constructs.We conclude that although evidence concerning the threshold hypoth-esis on the creativity–intelligence relationship is mixed, the “necessary condition hypothesis” is clearlycorroborated by the results of appropriate tests.

© 2016 Elsevier Inc. All rights reserved.

Keywords:IntelligenceCreativityThreshold hypothesisNecessary condition hypothesisNecessary Condition Analysis

1. Introduction

It is hardly possible tomake an impact in anydomain of human func-tioning, especially in the creative domain (Simonton, 2013), without asubstantial amount of intelligence (Cox, 1926; Plucker, Esping,Kaufman, & Avitia, 2015; Simonton, 2014). It is equally obvious, howev-er, that intelligence alone is not a sufficient condition for creativeachievement (Feist & Barron, 2003; Plucker, 1999). To achieve creativeaccomplishments, one has to start by making a decision to engage increative activity as such (Sternberg, 2002), possess high and adequatecreative self-efficacy (Beghetto, 2006; Karwowski, 2011), be open(Feist, 1998; Jauk, Benedek, & Neubauer, 2014), invest time in training(Ericsson, 2014; Simonton, 2014), and – last but not least – function ina supportive environment (Dul & Ceylan, 2011; Karwowski & Lebuda,2013).

Creativity scholars usually agree in defining creativity as a humancapacity to produce ideas and products that are both novel and usefulor appropriate (Amabile, 1996; Sternberg & Lubart, 1999). Even if some-times other creativity criteria are added, including the expectation thata creative product will be surprising (Simonton, 2012) or characterizedby an esthetic value and authenticity (Kharkhurin, 2014), the combina-tion of originality and value/usefulness is most often seen as essentialcriteria for a product to be considered creative (Runco & Jaeger, 2012).

University, 40 Szczesliwicka St.,

ski).

al., Is creativity without intel6

Although different taxonomies of creativity have been proposedover the decades (see Glăveanu, 2010, 2014 for a discussion), the crucialdistinction from the perspective we take in this article is that betweencreative potential (Runco, 2003) and creative achievement (Carson,Peterson, & Higgins, 2005; Eysenck, 1995; Robertson, Smeets,Lubinski, & Benbow, 2010;Wai, Lubinski, & Benbow, 2005). Creative po-tential is usually treated as a synonym of creative ability and measuredusing so-called creativity tests — mainly divergent thinking tasks(Runco, 1991). On the other hand, there are convincing arguments(see Weisberg, 2006 for a discussion) that creative ability is more thandivergent thinking alone (Baer, 1993; Guilford, 1967; Runco, 1991) orvividness of imagination (Jankowska & Karwowski, 2015). It also in-volves deductive and inductive thinking (Dunbar, 1997; Vartanian,Martindale, & Kwiatkowski, 2003;Weisberg, 2006) aswell as the abilityto use specific problem solving strategies (Finke, Ward, & Smith, 1992)to generate novel and appropriate solutions and outcomes. Creativeachievement refers to observable and socially recognized accomplish-ments in one or more domains (Simonton, 1994). Previous studieshave demonstrated that creative ability predicts creative achievement(Feist & Barron, 2003; Plucker, 1999; Runco, Millar, Acar, & Cramond,2010) and that creative activity (i.e., time devoted to creative trainingand creative behavior) mediates this relationship (Jauk et al., 2014).Both, classic (i.e., Feist, 1998) and more recent works (S. Kaufmanet al., 2016) have also shown that not only cognitive characteristics,but also personality traits predict creative potential, activity andachievement. Openness to experience forms the most consistent per-sonality predictor of creativity: it is important not only for everyday

ligence possible? A Necessary Condition Analysis, Intelligence (2016),

http://dx.doi.org/10.1016/j.intell.2016.04.006

mailto:[email protected]

Journal logo


http://www.sciencedirect.com/science/journal/01602896


2 M. Karwowski et al. / Intelligence xxx (2016) xxx–xxx

creative engagement (Conner & Silvia, 2015; Silvia et al., 2014), but alsofor creative activity (Jauk et al., 2014), creative self-beliefs (Karwowski& Lebuda, 2015) and creative achievement (Silvia, Nusbaum, Berg,Martin, & O'Connor, 2009). Importantly, recentworks have demonstrat-ed that two distinct aspects of openness trait – Openness and Intellect(DeYoung, 2015) influence creativity differently: while Openness pre-dicts creative achievement in the arts, Intellect predicts creativeachievement in science-related domains (S. Kaufman, 2013; S.Kaufman et al., 2016). Another study (Nusbaum & Silvia, 2011a) alsofound that while Openness predicts overall creative achievement, Intel-lect is more closely related to fluid intelligence.

Although the relationship between intelligence and creativity formsone of the classic problems researchers have examined over the decades(Guilford, 1967; Torrance, 1962), awidely accepted answer to this ques-tion is yet to come (Batey & Furnham, 2006; Silvia, 2015). The authors ofearly theories (Guilford, 1967) perceived intelligence as a necessary-but-not-sufficient condition for creativity, operationally defined interms of the so-called “threshold hypothesis” (TH; see Jauk, Benedek,Dunst, & Neubauer, 2013; Karwowski & Gralewski, 2013; Preckel,Holling, & Wiese, 2006; Runco & Albert, 1986). The TH assumes thatthe correlation between intelligence and creative ability depends onthe level of intelligence and expects a positive relationship only in thegroups of individuals whose intelligence level is below an IQ of 120.Above this hypothetical threshold, the correlation is expected to weak-en and/or to become statistically insignificant (Guilford, 1967). Hence,the TH assumes, on average, a curvilinear inverted J-shaped relationshipbetween intelligence and creativity. However, classic works that defineintelligence as a necessary-but-not-sufficient condition for creativityusually exemplified this relationship with the use of a characteristicscatterplot that takes the shape of a triangle (Guilford, 1967; Runco,2007) (Fig. 1).

Such a distribution of these two variables of interests shows that in-dividuals with high intelligence (X axis) attain almost any range of cre-ativity (Y axis) scores, including low levels of creativity, while thosewith low intelligence are generally not so creative. Even more impor-tantly, almost nobody is characterized by low intelligence and high cre-ativity, so the upper-left corner of this distribution chart is usually

Fig. 1. Schematic illustration of the relationship between intelligence and creativity. Solidblack line denotes a linear relationship; dotted black line denotes a relationship assessedusing segmented regression analysis; solid gray line denotes a necessary conditionrelationship (CR-FDH), while broken stepped black line denotes CE-FDH NecessaryCondition Analysis (see text for details and further explanations).

Please cite this article as: Karwowski, M., et al., Is creativity without intelhttp://dx.doi.org/10.1016/j.intell.2016.04.006

empty. Although this visual pattern resembles exactly what the neces-sary condition hypothesis proposes, serious doubts exist aboutwhetherprevious studies of these relationships tested it appropriately. Goingfurther, is it really possible to analyze the hypothesized “necessary-but-not-sufficient” relationship between creativity and intelligence re-ally in a correlational or regression (even polynomial; see Jauk et al.,2013; Karwowski & Gralewski, 2013) manner? In this paper we arguethat such hypotheses require alternative analytical approaches that gobeyond correlation or regression models. Consequently, in this articlewe propose that this classic problem may be better resolved by usingmore appropriate analytical methods, developed specifically to testthe necessary-but-not-sufficient conditions. We present details of thisapproach in the last part of the introduction. Before, we start by brieflysummarizing the long history of intelligence–creativity relationship andrecent findings.

1.1. A look at the intelligence–creativity relationship

Although creativity researchers over the years have postulated thatintelligence and creativity are independent psychological phenomena(e.g., Torrance, 1972; Wallach & Kogan, 1965), a recent movement inthe creativity literature leads to the conclusion that these constructsare “pretty similar after all” (Silvia, 2015). Using advanced statisticalmethods, especially those that make it possible to control for measure-ment error (Benedek, Jauk, Sommer, Arendasy & Neubauer, 2014;Nusbaum& Silvia, 2011b), and a differentiatedmeasurement of creativ-ity, not limited to divergent thinking tasks (Silvia & Beaty, 2012), recentstudies have shown that correlations between intelligence and creativ-ity may be much higher than creativity scholars are used to believing.These findings provide arguments that the true (latent) correlation be-tween intelligence and creativity is .40 to .50 (Nusbaum& Silvia, 2011b)or even higher (Jauk et al., 2014). Although a correlation at this levelsupports the discriminant validity of both these constructs (Brown,2015), it also shows that they are in fact more closely related than cre-ativity researchers would like to admit.

Unlike creativity researchers, intelligence researchers usually con-sider creative ability simply as part of intelligence (Carroll, 1993). Bothclassic (Jäger, 1984; Guilford, 1967) and contemporary models of intel-ligence (Carroll, 1993; McGrew, 2009) place creativity within the broadrange of subcomponents of intelligence. In the Carroll–Horn–Cattell in-telligence model (CHC; Carroll, 1993; Keith & Reynolds, 2010; McGrew,2009), long-term storage and retrieval ability (Glr) is responsible notonly for storing and consolidating new information in long-termmem-ory, but also for the fluent retrieval of the stored information: a psycho-logical mechanism crucial for creativity (Nusbaum & Silvia, 2011b) aswell as for several other types of fluency: ideational, associational,expressional, verbal, or figural. Also, figural flexibility, sensitivity toproblems, and originality are theorized to be subcomponents of intelli-gence (Carroll, 1993; McGrew, 2009).

The TH, postulating stronger average associations observed betweenintelligence and creativity among less intelligent individuals thanamong more intelligent ones, which creativity scholars perceive as anargument for the conceptual distinctiveness of these characteristics, isdifferently explained by intelligence researchers. Intelligence literaturetends to present such a finding as coherent with the Spearman Law ofDiminishing Returns (SLODR) (Spearman, 1927) — that is, the convic-tion that a lower g saturation of cognitive tests comes alongwith the in-creasing level of ability (Karwowski & Gralewski, 2013; Preckel et al.,2006). The pattern observed in the case of TH – lower correlations be-tween g and creativity at the higher levels of g – is exactly what theSLODR postulates.

If the theoretical positions held by creativity and intelligence re-searchers regarding the relationship between intelligence and creativitydiffer so radically, can research results provide a more clear-cut conclu-sion about this relationship? To put it shortly — not really. On the onehand, recent cognitive works show that intelligence and creativity are



3M. Karwowski et al. / Intelligence xxx (2016) xxx–xxx

driven by similar cognitive processes (Benedek, Franz, Heene, &Neubauer, 2012; Benedek, Jauk et al., 2014). However, neuroscientificfindings provide arguments for their distinctiveness (Jung et al.,2009). Correlational data show weak (Gralewski & Karwowski, 2012;Karwowski & Gralewski, 2013) to substantial relationships betweenthese two constructs (Jauk et al., 2013). It is very likely, then, that thisrelationship is moderated by several variables, starting from samplecharacteristics, via applied intelligence tests, all the way to creativityoperationalization and measurement.

Therefore, we believe that, for the sake of clarification, theoristsand researchers should pay more attention to the possible modera-tors of the intelligence–creativity relationship. They should also ex-plore this relationship especially to clarify whether creativityshould be defined and measured as a trait (Eysenck, 1995) or as acreative achievement (Carson et al., 2005). The relationship betweenintelligence and creative ability – e.g., fluency, the flexibility or orig-inality of thinking (Jauk et al., 2013; Karwowski & Gralewski, 2013),the ability to think metaphorically (Beaty & Silvia, 2013; Silvia &Beaty, 2012), or rich imagination (Jankowska & Karwowski, 2015)– can be easily explained in terms of CHC assumptions, and the THmay be perceived as consistent with the SLODR. Consequently, theobserved high overall correlation between creative ability and intel-ligence may not only simply mean that creative ability and intelli-gence share a substantial portion of variance, but it may alsoindicate that creative ability is nothing more than an aspect of intel-ligence, and correlation coefficients should be read similarly to thefactor loadings, showing how g-loaded specific aspects of creativeability are.

On the other hand, the relationship between intelligence and crea-tive achievement constitutes a much more convincing argument thatintelligence is a necessary-but-not-sufficient condition of creativity un-derstood as an accomplishment (Eysenck, 1995; Simonton, 1997, 2014).Previous literature rarely controlled these two aspects of creativity, –i.e., potential and achievement – while testing for their relationshipwith intelligence (see Jauk et al., 2014 or Plucker, 1999, for exceptions).In this study, we test the necessary condition hypothesis for creative po-tential and creative achievement separately in order to sketch a morecomplex and complete picture of the possible relationships.

1.2. Existing tests of the threshold hypothesis

Several studies provide arguments to believe that the curvilinear re-lationship between intelligence and creative ability indeed exists (Cho,Nijenhuis, van Vianen, Kim, & Lee, 2010; Fuchs-Beauchamp, Karnes, &Johnson, 1993; Gralewski, Weremczuk, & Karwowski, 2012; Jauk et al.,2013; Karwowski & Gralewski, 2013), while other investigationsfound linear relationships (Preckel et al., 2006; Sligh, Conners, &Roskos-Ewoldsen, 2005). A meta-analysis of the link between intelli-gence and creative ability (Kim, 2005) estimated the overall linkbetween these two constructs at r = .17, but, importantly, the correla-tion between intelligence and creative ability obtained among individ-uals with an IQ below 120 was almost exactly the same as thatobtained among individuals above this point (r = .24 and .20, respec-tively). However, even meta-analytical findings should be read withcaution, because 65 correlation coefficients above and only 14 coeffi-cients below IQ = 120 were meta-analyzed, which means publicationbias may have distorted the estimated pattern.

Importantly, the conclusion about the TH (i.e., the support or rejec-tion of this hypothesis) strongly depends on theway of operationalizingthe main variables and on several analytical decisions made byresearchers. For example, while Sligh et al. (2005) confirmed the THby analyzing the relations between crystallized intelligence (gc)and creative ability, they rejected the TH in the case of the relationsbetween fluid intelligence (gf) and creative ability. Recent studies(Cho et al., 2010; Gralewski et al., 2012; Karwowski & Gralewski,2013) show that relations between fluid intelligence (gf) and creative


ability may actually be consistent with the TH, but the exact decisionis also strongly influenced by data analysis methods. A great majorityof studies devoted to the TH compare correlations between intelligenceand creativity obtained above and below the hypothetical threshold(Cho et al., 2010; Fuchs-Beauchamp et al., 1993; Gralewski et al.,2012; Runco & Albert, 1986; Sligh et al., 2005). Though common, suchan analytical strategy has at least two serious limitations, which maybias the conclusions. First, researchers usually test the threshold valueof 120 IQ points (Cho et al., 2010; Fuchs-Beauchamp et al., 1993),even though initial hypotheses (Guilford, 1967) only generally indicatean above-average level of intelligence and even though recent studiesreveal that the decision on whether to support or reject the THmay de-pend on the exact threshold value tested (Karwowski & Gralewski,2013). Additionally, the arbitrary threshold of IQ = 120 creates theproblem of range restriction above this point (Runco & Albert, 1986;Sligh et al., 2005). Restricted intelligence variance in the high IQ group(IQ ≥ 120) artificially weakens the correlation, which may lead to ahasty confirmation of the TH. Controlling for restriction range abovethe threshold usually changes the conclusions (Karwowski &Gralewski, 2013; Sligh et al., 2005). Second, the decision as to whetherto accept or reject the TH when analyzing correlations coefficients de-pends on a specific decision rule applied. Karwowski and Gralewski(2013) reconstructed at least three different decision rules that areusually used by researchers when testing the TH. The most liberal(called A) confirms the TH when researchers find a significant andpositive correlation between IQ and creativity below the thresholdand a nonsignificant correlation above the threshold (see Slighet al., 2005). The second rule (called B), which is the most conserva-tive, confirms the THwhen there is a statistically significant and pos-itive correlation between IQ and creativity below the threshold and anonsignificant correlation above the threshold, but with the concur-rent assumption that these coefficients differ from each other (Choet al., 2010; Jauk et al., 2013; Karwowski & Gralewski, 2013;Mourgues et al., 2015). The third rule (called C) confirms the THwhen the relationship between intelligence and creativity in thegroup below the threshold is statistically significant and significantlystronger than above the threshold (Kim, 2005; Preckel et al., 2006).Consequently, this third rule only tests whether the correlation be-tween creative ability and intelligence above the threshold weakens.While these three rules may lead to completely different conclu-sions, they are relatively rarely applied together (Gralewski et al.,2012; Jauk et al., 2013; Karwowski & Gralewski, 2013), with rule A,the most liberal but also statistically weakest, being applied mostoften.

1.3. How to properly test the necessary condition relationships

Recent studies on the relationship between intelligence and cre-ativity successfully applied a new analytical approach to test theTH: segmented regression (Jauk et al., 2013; Mourgues et al.,2015). Segmented regression analysis allows for an estimation ofthe threshold based on data, making assumptions about the exactpoint unnecessary. The first study that used this method (Jauket al., 2013) revealed different thresholds for different aspects of cre-ative ability and no curvilinear relationship (though a significant lin-ear relationship) between intelligence and creative achievement (cf.Park, Lubinski, & Benbow, 2007, 2008;Wai et al., 2005). In the case offluency, the threshold was found at about one standard deviationbelow the IQ mean (IQ = 86), at the mean IQ level for the originalityassessed according to the top-two method (Silvia et al., 2008) (IQ =104), and exactly at the hypothesized 120 IQ points for the averagerated originality (IQ = 120). Using the same method, Mourgueset al. (2015) recently found the threshold for flexibility at IQ =106, but theywere unable to demonstrate any thresholds for fluency,originality, elaboration, and the composite divergent thinking score.




Therefore, further work is required to test the robustness of the seg-mented regression method for testing the TH.

Importantly, though, none of the methods used to test the TH –i.e., none of those based either on correlation coefficient comparisons(Runco & Albert, 1986), polynomial regressions (Gralewski et al.,2012; Karwowski & Gralewski, 2013; Runco et al., 2010; Sligh et al.,2005), or segmented regression (Jauk et al., 2013; Mourgues et al.,2015) – seem optimal for a proper test of the hypothesized “neces-sary-but-not-sufficient” relationship. Their main disadvantage lies inthe fact that all these methods are “average-based,” meaning that theregression line is the best estimate across and between the differentvalue points. Yet, it largely ignores the general pattern of the relation-ship, which is postulated by the necessary condition hypothesis asFig. 1 shows. Consequently, although regression-based methods are fitfor testing the general implication (“if X, then Y,” on average), they arenot fit for testing the “necessity assumption” (“there is no Y withoutX”) for virtually any observation (Dul, 2016). If intelligence is indeed anecessary condition for creativity, the scatterplot of their relationshipshould have an empty upper-left corner, which means there should beno individuals (or at least there should be very few)with high creativityand low intelligence. However, formal statistical testing of whethersuch a distribution is indeed consistent with this expectation requiresan analytical approach different than the one that is based on correla-tional analyses. Such possibilities have recently been offered by theNec-essary Condition Analysis (NCA: Dul, 2016; Dul, Hak, Goertz, & Voss,2010). The NCA focuses on single determinants that are necessary butnot automatically sufficient for an outcome. In this paper, we applythe NCA together with classic methods of testing the TH. First, however,we briefly sketch the main assumptions of the NCA.

Although the logic of necessary conditions goes back to DavidHume's philosophy of science (1777), it has been largely ignored inmodern social sciences. Since Francis Galton's (1886) discovery of corre-lation, the focus has been on regression and average trends and on suf-ficiency logic: predicting the outcome from single or multiplepredictors. Nevertheless, necessary (but not sufficient) conditionswide-ly exist in real life. For example, the human immunodeficiency virus(HIV) is a necessary-but-not-sufficient condition for AIDS (Madsen,Hodge, & Ottman, 2011): without HIV AIDS will not develop, but withHIV AIDS may or may not develop. Similarly, a student will not be ad-mitted to a top business school's Ph.D. program if his or her GMAT testscore is low, but a high GMAT score does not guarantee admission. Ifthe single necessary condition is present, it allows the outcome to beproduced but does not produce it; if the single necessary condition isabsent, it prevents the outcome from taking place and produces guaran-teed failure. Consequently, the absence of the single necessary conditionalmost perfectly explains the absence of the outcome. This dichotomousinterpretation of the necessary condition – i.e., the assumption that thecondition and the outcome can be either absent or present – can bemodified into a continuous interpretation, according to which a certainlevel of condition X is necessary-but-not-sufficient for a certain level ofoutcome Y. For example, a certain level of sociability in a sales person isnecessary-but-not-sufficient for a certain sales performance (Hogan &Hogan, 2007). Or: a certain level of intelligence is necessary-but-not-sufficient for a certain level of creativity.

Necessary Condition Analysis (Dul, 2016) is a novel approach toidentifying necessary-but-not-sufficient conditions in datasets. TheNCA draws a ceiling above the observations in the space of observa-tions. The NCA puts a “blanket” on the data in the three-dimensionalspace and a line in the two-dimensional space. The ceiling Y =f(X) separates the area with cases and the area without cases. Theceiling approach is in contrast to traditional regression, where aline (2D) or surface (3D) is drawn through the middle of the data.The necessary condition X for Y is represented by the inequationY ≤ f(X): all cases are on or below the ceiling. In practice, exceptions(outliers, errors, etc.) may be present, such that the “empty zone”above the ceiling is not entirely empty. One recommended technique


to draw the ceiling line is called Ceiling Envelopment–Free DisposalHull (CE-FDH; Dul, 2016), a nondecreasing linear step functionthrough the upper-left observations of a scatterplot (see Fig. 1). An-other recommended technique is Ceiling Regression–Free DisposalHull (CR-FDH), which smoothes this step function by drawing an or-dinary least squares trend line through the upper-left observations(see solid gray line in Fig. 1). Because CR-FDH is a smoothing func-tion, some observations are above the ceiling line. The effect size(d) of a necessary condition is the area of the “empty” zone abovethe ceiling line divided by the area of the “scope,” which is the totalarea where observations would be possible given the minimumand maximum values of X and Y. Thus, the larger effect size, thelower the ceiling line and the larger the constraint that X puts on Y.The effect size of a necessary condition can take the values between0 and 1. Dul (2016) suggests that effect sizes of 0 b d b 0.1 can be con-sidered a “small effect,” 0.1 ≤ d b 0.3 a “medium effect,” 0.3 ≤ d b 0.5 a“large effect,” and d ≥ 0.5 a “very large effect.” Thus, rather than fo-cusing on the “full zone” of observations in the scatterplot, as in cor-relational analysis, the NCA focuses on the empty zone and draws theceiling line.

1.4. Overview of the present studies

In this article, we consider data from eight independent studies andanalyze the intelligence–creativity relationship using typical analyticalapproaches presented in the literature, namely segmented regressions,correlation coefficient comparisons, as well as NCA.

We used different operationalizations and measurements of intelli-gence and creativity across the studies as Raven Matrices (Raven,2000; Studies: 1–3) or the Baddeley Grammatical Reasoning Test(Baddeley, 1968; Hunt, 2010; Studies 4 and 6). In the case of creativity,we mainly focused on creative ability (Studies 1–6), although we alsoreplicated findings concerning creative activity (Study 7) and creativeachievement (Study 8). We defined and measured creative abilitybroadly across studies by using either a complex figural creativity test,namely the Test of Creative Thinking–Drawing Production (TCT-DP;Urban & Jellen, 1996) (Studies 1–2), a new test that measures creativeimagination: the Test of Creative Imagery Abilities (TCIA; Jankowska &Karwowski, 2015) (Study 3), and classic divergent thinking tasks scoredfor fluency (Studies 4–6, 8) and originality (Studies 5–6, 8). Such abroadmeasurement of creative abilitymakes it possible not only to rep-licate the obtained effect, but also to test the generalizability of these re-lationships and to examine creative ability as a moderator.

In sum, the studies presented belowcovermore than 12,000 individ-uals — from children beginning their elementary school education(Study 2), through children in late elementary school (Study 3), middleschool students (Study 4), and high school students (Studies 1 and 5), toadults (Studies 6–8) (Table 1). As all the studies focused on the same re-search problem (i.e., as they all tested the relationship between intelli-gence and creativity), we describe them together. Table 1 covers allthe crucial characteristics of the studies, and Table 2 summarizes maindata on the reliability and validity of our measures.

2. Method

2.1. Participants

More than 12,000 participants (N=12,255 in total) took part in theeight studies presented below. They were elementary school students(Studies 2 and 3, n = 6184; 50%), middle school students (Studies 1and 4, n = 3730; 30%), high school students (Studies 1 and 5, n =880; 7%), as well as adults (Studies 6–8, n = 1461; 12%). The total agerange across the studies was 6 to 77 years, with a mean age of M =15.74, SD= 1.82. An overview of each of the studies follows below.



Table 2Summary of reliability and validity of measures applied in this investigation.

Study Instrument'sname

Reliability (this investigation) Validity

Intelligence measures1–3 RPM α = .87 (Study 1)

α = .91 (Study 2)α = .91 (Study 3)

Strong correlations with SAT (r = .48, Frey & Detterman, 2004), measures of Gc (r = .60, Bates &Shieles, 2003) and WISC (rs N .60, Dunkel, 2013).

4, 6 BGRT α = .93 (Study 4)α = .73 (Study 6)

Substantial correlations with Wonderlic Personnel Test (ranging from r = .44,Chamorro-Premuzic & Furnham, 2008, to r = .65, Furnham & Chamorro-Premuzic, 2006).Substantial correlations with RPM (r = .39, Furnham & Chamorro-Premuzic, 2006)

5 APIS-Z α = .71 Substantial correlations with RPM (r = .60) (Matczak et al., 2006).7 ICAR α = .88 Good CFA and IRT models fit, strong correlations (rs N .50) with SAT, ACT and GRE (Condon &

Revelle, 2014).8 INSBAT Rasch-scaled tests, so internal consistency is

given. Target reliability set to α = .60.Good CFA model fit (Arendasy et al., 2004). 50–88% of item parameter variance is explained bythe construction rationale. Scales significantly predict school achievement (.20 to .37).

Creativity measures1–2 TCT-DP α = .74 (Study 1)

α = .71 (Study 2)Correlation with Alternative Uses Task (r = .28) (Mullineaux & Dilalla, 2009), object-baseddrawing (r = .28) (Lubart, Pacteau, Jacquet, & Caroff, 2010), creative activity in art (r = .19) andin science (r = .18) (Gralewski & Karwowski, 2013)

3 TCIA α = .91 Correlations with vividness of Visual Imagery Questionnaire (rs between .31 and .42), FranckDrawing Completion Test (rs between .20 and .48), generating imaginary animals (originality:r = .45, transformativeness: r = .32), TCT-DP (rs between .20 and .32), divergent thinking(originality, r = .26) (Jankowska & Karwowski, 2015).Good fit of the theoretical model of TCIA (CFI = .988, RMSEA = 0.019) (Jankowska & Karwowski,2015).

4, 6, 8 DT tasks α = .86 (Study 4, fluency)α = .91 (Study 6, fluency),α = .77 (Study 6, originality)ICC = .80 (Study 8, fluency)ICC = .69 (Study 8, originality)

Correlation with the VKT test: r = 0,34 (Rindermann & Neubauer, 2004), Creative ExpertPerformance: r = 0,23 (An, Song, & Carr, 2016), Franck Drawing Completion Test (creativeimagination): flexibility: r = .28, originality: r = .37 (Dziedziewicz, Gajda, & Karwowski, 2014).

5 Word UniquenessTest

– Correlations with openness to experience: r = .39) (Chruszczewski, 2010)

7–8 ICAA α = .80 (Study 7)α = .80 (Study 8)

Creative activities predict creative achievement (.67; Jauk et al., 2014).Jazz musicians show more creative musical activities and higher creative achievements thanclassical or folk musicians (Benedek, Borovnjak et al., 2014).

Table 1Study and sample characteristics.

Study N Age M (SD) Percentage of females Type of school Measure of intelligence Measure of creativity Country

Study 1 921 16.2 (3.1) 51% Middle and high school RPM TCT-DP PolandStudy 2 540 9.36 (1.68) 54% Elementary school RPM TCT-DP PolandStudy 3 5644 12.82 (0.33) 50% Elementary school RPM TCIA PolandStudy 4 3398 15.05 (0.28) 49% Middle school BGRT DT (fluency) PolandStudy 5 291 16.75 (0.83) 73% High school APIS DT (originality) PolandStudy 6 429 34.00 (5.65) 43% Adults BGRT DT (fluency, originality) PolandStudy 7 735 38.08 (13.01) 53% Adults ICAR PolandStudy 8 297 30.40 (10.68) 66% Adults INSBAT DT (fluency, originality)

Creative achievementAustria

Note. RPM=RavenMatrices; BGRT= Baddeley Grammatical Reasoning Test; APIS=Multidimensional Crystallized Intelligence Battery; ICAR= International Cognitive Ability ResourceTest; INSBAT = Intelligence Structure Battery; TCT-DP = Test of Creative Thinking–Drawing Production; DT = divergent thinking, TCIA = Test of Creative Imagery Abilities; ICAA =Inventory of Creative Activities and Achievements.


2.1.1. Study 11

In total, 921 students of middle and high schools located in centralPoland participated in the study: 448 men (49%) and 473 women(51%). The participants attended two types of schools: 332 (36%) weremiddle school students and 589 (64%)were high school students. All re-spondents were aged 14 to 20 years (M = 16.2, SD= 3.1).

2.1.2. Study 2The participants were 540 students of elementary schools located in

central Poland: 248men (46%) and 292women (51%). Their age rangedfrom 6 to 12 years (M = 9.36, SD = 1.68).

1 Results of the first (Study 1; Karwowski & Gralewski, 2013) and the last study (Study8; Jauk et al, 2013) have already been published. We decided to re-analyze the data fromthese studies using the NCA approach, as this created a unique opportunity of comparingthe results obtained with the use of different methods.


2.1.3. Study 3A total of 5644 individuals formed a representative sample of Polish

elementary school students (grade 5). The proportion of male and fe-male students was equal. The mean age of the respondents was M =12.82, SD= 0.33.

2.1.4. Study 4The study was conducted on a sample of Polish middle school stu-

dents (N = 3398), attending grade 2 in 89 schools located in citiesabove 100,000 inhabitants. The sample was 50.8% female and almostall participants (89.7%) were born in 1999, which means they were15 years old at the moment of the study.

2.1.5. Study 5This study was conducted on a sample of 291 high school students

(73% females) attending high schools in Warsaw (the capital of



Table 3Relations between intelligence and creative ability: comparison between correlational and Necessary Condition Analysis.

Study Intelligence Creativity Pearson's r (95% CI) Threshold r below threshold r above threshold NCA d CE-FDH NCA d CR-FDH

Study 1 RPM TCT-DP .25 (.19,.31) NO – – .257 .236Study 2 RPM TCT-DP .26 (.16,.35) NO – – .208 .201Study 3 RPM Creative imagination (TCIA) .20 (.17,.23) NO – – .232 .223Study 4 BGRT DT fluency −.02 (−.05,.01) 81 .19a (.05,.33) −.04b (−.07, −.01) .312 .286Study 5 APIS DT originality 0 (−.11, .11) NO – – .174 .133

Study 6 BGRTDT fluency .05 (−.04,.14) NO – – .301 .218DT originality .15 (.06,.24) NO – – .138 .127

Study 7 ICAR Creative activity (ICAA) .29 (.22,.35) 121 .28a (.21, .35) −.06b (−.29, .18) .248 .223

Study 8 INSBAT

DT fluency .22 (.11,.33) 86 .56a (.17, .80) .09b (−.03, .21) .267 .250DT originality .35 (.25,.45) 120 .35a (.23, .46) −.01b (−.25,.23) .188 .173Creative achievement (ICAA) .28 (.17,.38) NO – – .279 .233

Note. RPM=RavenMatrices; BGRT=Baddeley Grammatical Reasoning Test; APIS=Multidimensional Crystallized Intelligence Battery; ICAR= International Cognitive Ability Resource;INSBAT= Intelligence Structure Battery; TCT-DP= Test of Creative Thinking–Drawing Production; DT= divergent thinking, TCIA= Test of Creative Imagery Abilities; ICAA= Inventoryof Creative Activities and Achievements. d = necessary condition effect size.


Poland), Cracow, and Gdansk. The participants' age ranged from 14 to19 years, withM = 16.75 and SD= 0.83.

2.1.6. Study 6The participants were 429 Polish adults (M = 35.17, SD = 5.68,

range = 26–46; men = 284, women = 216). A majority of them hada college or university degree (67%) or a high school diploma (28%).

2.1.7. Study 7The participants were 735 Polish adults (M = 38.06, SD = 13.01,

range = 17–77, men = 349, women = 387). A majority of them hada college or university degree (54%) or a high school diploma (34%).

2.1.8. Study 8The sample consisted of 297 Austrians (101 males) with an average

age of 30.40 years (SD = 10.68); 16% of the participants had at least9 years of schooling, 60% had at least 12 years of schooling, and 24%had a university degree.

2.2. Measures

2.2.1. Measures of intelligence

2.2.1.1. Raven ProgressiveMatrices (RPM) (Studies 1–3).Weused Raven'smatrices (Raven, 2000; Raven, Court, & Raven, 1998), as adapted intoPolish (Jaworowska & Szustrowa, 2000), to measure students' intelli-gence in Studies 1–3. In Study 1, conducted on middle and high schoolstudents, we applied the Standard Progressive Matrices— Plus. In Stud-ies 2 and3, conducted on elementary school students,we used the Stan-dard Progressive Matrices. RPM are widely used as a measure of fluidintelligence. Their items have the form of abstract matrices and thetask is to complete their missing piece. The matrix piece should be se-lected from among the alternative solutions proposed in the test. Eachtask (matrix) has one correct solution. The test is composed of 60matri-ces arranged in the order of increasing difficulty. The reliability of RPMwas α = .87 in Study 1, α = .91 in Study 2, and α = .91 in Study 3.

2.2.1.2. Baddeley Grammatical Reasoning Test (BGRT) (Studies 4 and 6).We used the Baddeley Grammatical Reasoning Test (BGRT: Baddeley,1968; Hunt, 2010) to measure students' intelligence in Studies 3 and4. BGRT measures reasoning about the relationships among differentletters based on deductions from grammatical statements. It is com-posed of 64 items in the form of varying types of grammatical state-ments, with a total of 3 min to solve them. The respondent's task is toassess the truthfulness or falsity of each of them. The reliability BGRTwas α = .93 in Study 4 and α = .73 in Study 6.

2.2.1.3. APIS-Z (Study 5). APIS-Z (Matczak, Jaworowska, Ciechanowicz, &Stańczak, 2006) is a multidimensional battery whose aim is to measure


crystallized intelligence. APIS-Z is composed of eight tests: Behaviors,Squares, Synonyms, Classification, Converting numbers, New words,Building blocks, and Stories. The reliability of the APIS-Z in Study 5,assessed using the split-half method, was r = .70 (α = .71).

2.2.1.4. ICAR Task Battery (Study 7). To measure intelligence, we usedthirty items developed within the International Cognitive Ability Re-source Project (ICAR, http://www.icar-project.com, see also Condon &Revelle, 2014). These were: ten matrix items (α= .74), ten items mea-suringmental rotations (α= .86), and ten items connectedwith overallreasoning (α= .77). The reliability of the overall score was good (α=.88).

2.2.1.5. Intelligence Structure Battery (INSBAT) (Study 8).We assessed in-telligence bymeans of four subtests of the Intelligence Structure Battery(INSBAT; Arendasy et al., 2004), which were constructed to measurefluid intelligence (Gf). Subtests included tests of figural inductive rea-soning, arithmetic flexibility, verbal short-term memory and wordmeaning (Arendasy et al., 2004). The INSBAT is based on item responsetheory (IRT) and allows for tailored testing. The target reliability foreach scale was set to α = .60 (see Jauk et al., 2013, for more detailsabout the INSBAT).

2.2.2. Measures of creativity

2.2.2.1. Test for Creative Thinking–Drawing Production (TCT-DP) (Studies1–2). Test for Creative Thinking–Drawing Production (TCT-DP: Urban &Jellen, 1996; Matczak, Jaworowska, & Stańczak, 2000) was used tomea-sure the extent of students' creative abilities in Studies 1 and 2. The par-ticipants are asked to complete an unfinished drawing. Assessmentusing the TCT-DP (Urban & Jellen, 1996) includes the following criteria:Continuations (Cn), Completions (Cm), New elements (Ne), Connec-tions made with a line (Cl), Connections that contribute to a theme(Cth), Boundary breaking that is fragment-dependent (Bfd), Boundarybreaking that is fragment-independent (Bfi), Perspective (Pe), Humorand affectivity (Hu), Unconventionality with subcriteria (Uc)[(a) manipulation of the test material (Uca); (b) surrealistic or abstractelements (Ucb); (c) use of symbols or signs (Ucc); (d) unconventionalusage of the given fragments (Ucd)] and Speed (Sp). The final TCT-DPscore is the sum of points for these criteria. One score is used as a mea-sure of creative ability. The reliability of the overall scorewasα= .74 inStudy 1 and α = .71 in Study 2.

2.2.2.2. Test of Creative Imagery Abilities (TCIA; Study 3). Test of CreativeImagery Abilities (TCIA; Study 3) was used to measure creative imagi-nation. The TCIA (Jankowska & Karwowski, 2015) test booklet is in A3format and consists of seven tasks. The first stage of solving each taskis of an exploratory character. The participant is taskedwith generating,in an oral or written form, as many images as possible on the basis of a


http://www.icar-project.com



simple graphic sign, called the initial figure. Next, he or she selectsthemost original of the images given and, on its basis, makes a drawingaccompanied by a brief description. The instruction stresses the possi-bility of elaborating and changing the selected image and adding any el-ements to it in such a way as to create something even more original.The drawings anddescriptions of imagerymade in the TCIA are assessedon three scales based on the conjunctional model of creative imagingability (Dziedziewicz & Karwowski, 2015): the Vividness scale, theOriginality scale, and the Transformativeness scale, but the overallscore may be used as well. Each scale is scored according to the criteriadiscussed in detail and illustrated with examples in the test manual(Jankowska & Karwowski, 2015). According to these criteria, it is possi-ble to score 0, 1, or 2 points on each scale for a single drawing. The scoreson scales are computed by adding up the points scored for all the draw-ings. Previous studies (Jankowska & Karwowski, 2015) demonstratethat the TCIA is valid and reliable. Its reliability in this study was high(α = .91).

2.2.2.3. Divergent thinking tests (Studies 4, 6, 8). Different tasks that mea-sure divergent thinking were used in Studies 4, 6, and 8. In Study 4 weused five Guilford-type tasks (Guilford, 1967). These included two “un-usual uses” tasks (generating unusual uses for toothpaste and ice-cream), two consequences tasks (what would happen if there were nocell phones or the Internet and what would happen if people ate onlycandy), and one similarities task (what is similar about a chocolate barand a computer). The fluency index was computed by summing up allthe answers across thefive tasks (α=.86). Study 6 comprised three un-usual uses tasks. Participants were asked to provide unusual uses of abrick, a can, and tape. Answers were scored for fluency (α = .91: thesum of answers given in three tasks) and originality. Each responsewas assessed by three judges blind to the study hypotheses, using a7-point Likert scale. Due to good reliability (α= .77), originality scoreswere averaged. Study 8 consisted of three alternate uses tasks and threeinstances tasks. In the alternate uses tasks, the participants were re-quired to find as many novel and uncommon uses as possible for acan, a knife, and a hairdryer. In the instances tasks, the participantswere instructed to figure out many novel and uncommon solutions tothe following problems: “What can make noise?,” “What can be elas-tic?,” and “What could one use for quicker locomotion?” Four students(three female) rated the originality of the responses (similar to the con-sensual assessment technique proposed by Amabile, 1982) given inboth the alternate uses and the instances task on a four-point scale rang-ing from 1 “not creative” to 4 “very creative.” Mean interrater reliabil-ities were ICC = .80 in the alternate uses tasks and ICC = .69 in theinstances tasks. We used two common scores of creative potential. Ide-ational fluency was defined as the number of ideas given in the task. Toassess ideational originality, we computed an average originality scorethat reflects the mean creativity ratings of all ideas.

2.2.2.4. The uniqueness of Word Association (Study 5). Word AssociationTest (WAT; Chruszczewski, 2009, 2010)was used tomeasure the abilityto produce associations. TheWAT is composed of a sheet with 24word-triggers printed on it. The triggers are randomly selected from the Polishlanguage dictionary. It also comes with instructions on a separate sheetthat encourage providing multiple original and diverse responses,which the participants are expected to write on a separate answersheet. The associations are assessed on the basis of response uniquenessby first selecting the so-called unique associations that would only in-clude the associations that appeared only once in response to theword-trigger. Then, the proportion between the number of unique asso-ciations and all associations produced by a given individual is calculatedand becomes the final measure of uniqueness.

2.2.2.5. Inventory of Creative Activities and Achievements (ICAA; Studies 7–8). To measure creative activity (Study 7) and achievement (Study 8),we used the Inventory of Creative Activities and Achievements (ICAA;


see Jauk et al., 2014). In Study 7 the ICAA was slightly modified in com-parison with the original instrument, as only items related to creativeactivity were used. Thirty-three items dealt with creative activity in dif-ferent domains and askedwhether a person did (coded 1) or did not en-gage in such activity (coded 0) within the past year. This scale yielded areliable index of overall creative activity (α= .80), and, consequently, itwas averaged. Study 8 used the full version of the ICAA across eight do-mains (literature, music, arts and crafts, creative cooking, sports, visualarts, performing arts, and science and engineering) of creative accom-plishment. Participants are presented with statements ranging from “Ihave never been engaged in this domain” (zero points) to “I have al-ready sold some of my work in this domain” (10 points). The internalconsistency of the achievements scale across domains was satisfactory(α = .71).

2.3. Procedure

The procedure varied between the studies. Studies 1–5 and Study 8were conducted face-to-face, while Studies 6 and 7were conducted on-line. Additionally, participants from Study 2 (elementary school stu-dents) were tested individually, while participants from Studies 1, 3,4, 5, and 8 were tested in a group setting, usually in their classroomsor university labs. In all studies, participants were first given the intelli-gence and creativity tests. If additional data were collected during thestudies, we administered the tests in a randomized order. There wereno time restrictions in the case of intelligence tests, despite the timedtests (BGRT — Study 4 and 6 and APIS-Z — Study 5) that originally hadtime restricted to 3 min (BGRT) and 60min (APIS-Z). Solving creativitytests in Studies 1 and 2 (TCT-DP) had no time restriction, although timewas recorded; creativity tests in Study 3 had no time restrictions, whilethe divergent thinking tests used in Studies 4, 6, and 8 had time restrict-ed to 2min per each task (five tasks in Study 4, three in Study 6, and sixin Study 8). WAT's solving time (Study 5) was restricted to 15 min.

Informed written consent was obtained from all participants andfrom parents, teachers, and school principals in case of school stu-dents (Studies 1–5). All the participants were informed that partici-pation was voluntary and that they could withdraw at any time.Participants in Studies 1–5were not rewarded for their participation,participants in Studies 6 and 7 weremembers of the online panel runby the Millward Brown Poland research company (including 90,000Poles — a nationwide representative sample of Internet users), whotake part in various research programs once or twice per year andare rewarded for it. Participants in Study 8 were paid for theirparticipation.

3. Results

Our analyses included three steps: (a) we began with simplePearson's rs, to estimate the correlations between intelligence and cre-ativity in each of the studies and provided a more synthetic, meta-analytically obtained coefficient aggregated across the eight studies.Then (b) we reanalyzed all datasets by applying the segmented regres-sion approach (Jauk et al., 2013), and in the cases where significantbreakpoints were obtained, we compared the correlations between in-telligence and creativity below and above the breakpoints. Finally,(c)we conducted aNecessary Condition Analysis on each of thedatasetsin order to draw the ceiling line in the scatterplot and calculate the ne-cessity effect size.

Correlations between different aspects of creativity and intelligenceranged from r = −.02 (nonsignificant) for BGRT and fluency (Study4) to r = .35 (p b .001) for originality and INSBAT (Study 8) (seeTable 3). The meta-analytically obtained correlation between intelli-gence and creativity estimated with the use of a random effect modelin a meta-analysis (package meta in R; Schwarzer, 2015) of the resultsof our 8 studies was r = .17, 95% CI: .07, .27, p = .0005: exactly thesame as in the previous meta-analysis (see Kim, 2005), but this effect




was also highly heterogeneous, Q(df=7)= 152.96, p b .001, I2 = 95%.2

Despite the heterogeneity, however, out of the 11 correlation coefficientsobtained, eight were statistically significant and seven were above r =.20.

The segmented regression analyses were performed in R using thesegmented package (Mueggo, 2008). Intelligence served as the inde-pendent variable and creativity as the dependent variable. We used aninitial guess parameter for the breakpoints of ψ = 100 IQ points andk = 10 segments. Empirically determined breakpoints were tested forstatistical significance by means of the Davies test (Mueggo, 2008). Sig-nificant curvilinear relationshipswere observed in four cases. In Study4,the threshold of the relationship between fluency and the BGRT resultswas estimated at an IQof 81 in theBGRT. Below this pointwe observed apositive and significant relationship, r = .19, 95% CI: .05, .33, but thiscorrelation was significantly negative, albeit weak, above the threshold,r=−.04, 95% CI:−.07,−.01. In Study 7, which examined the relation-ship between intelligence and creative activity, the threshold was esti-mated at IQ = 121 points, with a robust and significant correlationbelow this point, r = .28, 95% CI: .21, .35, and nonsignificant above it,r = −.06, 95% CI: −.29, .18. In Study 8, we obtained two differentthresholds for fluency and originality. In the case of fluency, thebreakpointwas found at 86 IQ scale points, with strong andpositive cor-relation observed below this point, r= .56, 95% CI: .17, .80, and a lack ofrelationship above that point, r = .09, 95% CI: −.03, .21. In the case oforiginality, the thresholdwas estimated at 120 IQ points, with a positivecorrelation below120 IQpoints, r=.35, 95% CI: .23, .46, and no relation-ship above it, r = −.01, 95% CI: −.25, .23.

Then, we applied the NCA (Dul, 2016) using the NCA package ver-sion 1.1 in R (Dul, 2015). Fig. 2 shows the results. The NCAs effect size(CR-FDH) varied between d = .127 (intelligence and originality inStudy 6) and d = .286 (intelligence and fluency in Study 4), with asample-weighted average of d = .239. According to the recommenda-tions (Dul, 2016) this effect should be consideredmoderate. Additional-ly, Dul (2016) suggests that for the sake of a dichotomous decision,whether the necessary condition is observed or not, a value of d = .10could be treated as a threshold. As Table 2 shows, all the effects wehave obtained were clearly above this value.

The wide scope of measures and aspects of creativity and intelli-gence that we applied across studies gave us the possibility to assessthe generalizability of the NCA approach. As Fig. 3 demonstrates, thevariability of the NCA effects sizes we have obtained was located acrossthe different aspects of creativity rather than across the studies. Morespecifically, similar relationships, and close to the boundary of strong ef-fects (Dul, 2016), were observed in the case of fluency (sample-weight-ed average d = .265) and creative activity or achievement (similareffect size, sample-weighted average d = .226). Similar effects werefound when creativity was measured in a complex way using the TCT-DP test (d = .24 and d = .20) and the creative imagination test (d =.22). Consequently, the weakest NCA effects sizes were observed inthe case of originality (ds between .13 and .17).

4. Discussion

To be tested properly, classic research problems sometimes requiremodern analytical methods. The relationship between intelligence andcreativity is one of them, and previous studies (Jauk et al., 2013) andre-analyses (Silvia, 2008a, 2008b, 2015) showed that these two con-structsmay be closer to eachother thanwaspreviously assumed. There-fore, in this article we explored a problem that is obviously classic

2 Large heterogeneity of the overall effect size of the correlation between intelligenceand creativity across our studies is not surprising given variability of participants' ageand different measures of intelligence and creativity used. Therefore, although we high-light that the overall effect we obtained resembles the previous meta-analysis (Kim,2005), we encourage readers who interpret our findings to focus on separate correlationsrather than on an aggregate.


(Guilford, 1967; Torrance, 1972), yet still interesting for contemporaryresearchers (Gralewski et al., 2012; Jauk et al., 2013; Karwowski &Gralewski, 2013; Silvia, 2015). We argued that although theintelligence–creativity relationship has been analyzed in dozens ofstudies in the last few decades (see Karwowski & Gralewski, 2013 foran overview andKim, 2005 for ameta-analysis), in fact it has very rarelybeen examined in a way that would allow for the conclusion that intel-ligence is or is not a necessary condition of creativity. Consequently,thanks to a recently developed analytical method known as the Neces-sary Condition Analysis (Dul, 2016), we were able to assess this rela-tionship in a way that is more appropriate and robust.

Across the eight studies on an aggregated sample of more than12,000 people at different ages from two countries, using a wideoperationalization of intelligence and creativity, we demonstratedthat, indeed, there are convincing reasons to believe that intelligenceis a necessary condition of creativity. Importantly, however, the NCAwas clearly more consistent and sensitive in finding the necessary-but-not-sufficient patterns of relationships than correlation analysis oreven than the more sophisticated segmented regression analysis. Itshould be noted that although at the most general level the pattern ob-served across studies was consistent with the necessary condition hy-pothesis, it was qualified by the creativity aspect measured. The effectsize of the NCA was clearly the weakest in the case of originality andthe strongest in the case of fluency. When creativity was measuredusing quite a complex creativity test (TCT-DP) or the similarly elaboratecreative imagination test (TCIA), the effects size of the NCAwasmoder-ate. Finally, in the case of creative activity and creative achievement wehave observed clear and medium-to-strong (Dul, 2016) NCA effects.

Although our analyses dealt with a classic research problem inthe psychology of creativity, and although this paper is aimed atmethodological rather than theoretical contribution, both theoreti-cal and methodological aspects of our findings should be empha-sized. Theoretically, this study offers some arguments that differentaspects and facets of creativity may relate to intelligence differently.Methodologically, the NCA seems to be a useful and more appropri-ate method for testing necessity, an alternative to the widely usedcorrelation and regression analysis.

4.1. Intelligence and different aspects of creativity

Creativity is not limited to the ability to generate new and rare ideas.After all, these ideas have to be contextually appropriate or useful to bescored as creative (Kaufman, 2016). Creative ability tests usually focusonly on selected aspect of creative thought, and they most often testthe ability to think fluently and originally (Gajda, Karwowski &Beghetto, in press). However, even if fluency and originality are ex-tremely important for creative thought (indeed, ironically, they maybe named the necessary-but-not-sufficient criteria for creativity), crea-tivity is not reducible to them. This was precisely the reasonwhywe ap-plied a broad array of creative ability measures and, except pure indicesof fluency and originality, we also applied more elaborated figural testsof creative ability (TCT-DP) and creative imagination (TCIA). As thescoring of these tests also involves aspects of elaboration, theymay pro-vide a more valid picture of creative ability. Although we treated thispart of our analyses as exploratory, we were interested in whetherand to what extent such different operationalizations of creative abilitywould be related to intelligence and whether the role played by intelli-gencewould be the same or different for different facets of creative abil-ity. Our results show a large heterogeneity of associations betweenintelligence and creative ability when this association is assessed withthe use of the correlation coefficient. Thesemanifest correlations variedfrom nonsignificant −.02 to highly significant .35, with an average ofr = .17, resembling the meta-analytical estimate (Kim, 2005). Impor-tantly, these correlations are not only heterogeneous, but also not verysystematic. For instance, in the case of originality, the correlations variedbetween 0 and .35. Similar inconsistencies were observed in the case of



Fig. 2.Visualization of theNCA across eight studies and eleven scores. Black lines denote the linear (correlational) function, solid gray lines denote the CR-FDH ceiling line, and broken graylines denote the CE-FDH ceiling line.


fluency. We argue, however, that correlational analysis is inappropriatefor testing the necessary-but-not-sufficient relationship patterns. TheNCA provided much more similar effect sizes across studies using thesame creativity measures, with the strongest effects for fluency andcomplex tests of creative ability (TCT-DP and TCIA) and visibly weakereffects for originality. This latter finding is especially interesting giventhat recent studies demonstrated that intelligence tends to correlate es-pecially strongly with originality (Benedek, Franz, Heene, & Neubauer,2012; Jauk et al., 2014). It confirms the basic NCA assumption that cor-relation and necessity can be completely independent: there can be ne-cessity without correlation and correlation without necessity (see


examples in Dul, 2016). In our studies, the NCA effect size for originalitywas visibly the smallest. It therefore seems that although intelligencemay be strongly correlated with originality, especially when analyzedusing latent variables models, it does not necessarily serve as the neces-sary condition for originality. We have observed that people with lowintelligence are quite able to provide rare ideas (consequently, nostrong NCA pattern is observed), but this does not mean that theseideas are creative (simultaneously original and useful; Eysenck, 1995).

In two of our eight studies we also focused on creative activity andobservable creative achievement that allowed us to go beyond creativeability. This decision was driven by the argument, presented in the



3 Interested readers may find more information about the NCA on the dedicatedwebsite: www.erim.nl/nca.

Fig. 3. Effect sizes of the NCA (CR-FDH) analyses across studies and different aspects ofcreativity. Note. S1–S8 = Study 1–Study 8; TCT-DP = Test of Creative Thinking–Drawing Production; Cr-Imag = creative imagination; Cr-Act = creative activity; Cr-Ach = creative achievement.


introduction, that while creative ability may be perceived as a subfactorof g, creative activity or achievement are not reducible to intellectual po-tential and are being predicted by non-cognitive factors as well, includ-ing Openness and Intellect factors of the openness trait (S. Kaufman,2013; S. Kaufman et al., 2016) or a wide range of creative self-beliefs(Karwowski & Lebuda, in press). Therefore, examining the links be-tween intelligence and creativity understood in such a way makesmore sense. These relationships are relatively rarely studied (seePlucker, 1999; Robertson et al., 2010; Wai et al., 2005 for exceptions),even though they may provide a much more robust answer to NCA-related questions. Across our studies we obtained a very similar andclear pattern of the relationship between intelligence and creative activ-ity/creative achievement, which confirms the role intelligence plays increative endeavors. First, in both cases (creative activity and creativeachievement), positive correlations were found between intelligenceand creativity. Second, in the case of creative activity (but not creativeachievement), the relationship between intelligence and the intensityof creative activity was clearly curvilinear, with a threshold found atthe IQ level of 121 points. This finding underpins the idea that creativeactivity can be considered a form of so-called little-c creativity(Kaufman & Beghetto, 2009) and needs only a minimum intelligence,while creative achievement, more related to pro-C creativity, benefitsfrom intelligence even at high levels (Park et al., 2007; Wai et al.,2005). In both cases, however, intelligence was a clearly necessary-but-not-sufficient condition of creativity— therewere almost no partic-ipants who had high creative activity or achievement and low intelli-gence. Consequently, we found confirmation that intelligence not onlypredicts creativity in general but also constitutes a necessary-but-not-sufficient condition for individually observable creative behaviors andaccomplishments. Previous studies (Jauk et al., 2014) demonstrated thatwhile intelligence indeed predicts creative achievement, it is personality(especially openness) that is associated with creative activity (Conner &Silvia, 2015; Silvia et al., 2014;Wolfradt&Pretz, 2001). As creative activityis relatively frequent, at leastwhen compared to achievement, it is usuallytheorized to be driven by interests (Karwowski & Barbot, 2016) andcreative personal identity (Karwowski & Lebuda, in press). On the otherhand, however, according to investment theories (von Stumm &Ackerman, 2013), although intelligence (especially g) may serve as thecause of creative activity, it is plausible to consider crystallized intelli-gence as the result of previous activity and engagement (von Stumm,2015). Therefore, future studies, optimally longitudinal, should not onlyexplore these relationships further, but also control creative ability, crea-tive activity, and creative achievement in a single study. The intelligence–creativity relationship has also been addressed from a neuroscienceperspective. Jung et al. (2009) observed different correlations between


creative potential and markers of neuronal integrity (i.e., NAA) in groupsof lower and higher intelligence. Similarly, Jauk, Neubauer, Dunst, Fink,and Benedek (2015) found that ideational fluency is related to higher re-gional gray matter volume only in a low IQ group but not in a high IQgroup. The authors speculate that fluent idea production may be tiedmore strongly to visual imagery in less intelligent people, whereas moreintelligent people may employ more complex generation strategies thatare less clearly associated with a single brain structure. These findingsprovide initial insights into how creativity is differently manifested inthe brain on different levels of intelligence. Future research in this fieldis challenged to identify common and distinct neural substrates underly-ing intelligence and creativity and to acknowledge the potentiallymoder-ating role of intelligence in the investigation of the brain correlates ofcreativity.

4.2. The NCA as an analytical alternative

We perceive the main contribution of this paper to be methodolog-ical: the presentation and application of theNCAmethod that allows foramore appropriate analysis of the hypothesized relationships. Althoughthe necessary-but-not-sufficient pattern of the intelligence–creativityrelationship was hypothesized for decades, we argue that it was testedimproperly or, at least, imperfectly. Correlation, regression, segmentedregression, or even latent variable analysis are not fit to test whetheror not one construct may be seen as a necessary condition of another.All these methods, no matter whether they are based on simpleordinary-least-squares estimations or on themore elaborate maximumlikelihood estimations, deal with average relationships — namely, theyestimate the relationship across the distribution of raw scores. Thus,by definition, they test a different type of relationships. Thanks to a re-cently developed analytical method (Dul, 2016), we were able to quan-tify the necessary-but-not-sufficient condition effect and analyze thesize of the “empty-upper-left-corner” — for a long time perceived asproof of this type of relationships. NCA is an intuitive, yet robust analyt-ical tool for estimating such a relationship. This effect was replicatedacross all studies presented in this paper, and its variability was clearlyassociated with and explained by the creativity aspect being measured.Therefore, we do believe that although the NCAmethod is in the begin-ning of its development, the previous studies (Dul, 2016) and the cur-rent findings make it possible to view this method with optimism.3

4.3. Strengths, limitations, and future studies

The studies presented in this article have several obvious advan-tages, but also limitations that should be kept in mind whileinterpreting the results. Among the crucial advantages, we see theirhigh power and the wide and diversified measurement of both con-structs of interest. Even more importantly, as we were able to test therelationship between intelligence and creativity understood as creativeability, creative activity, and creative achievement, this study goes be-yond previous research, which typically focused only on creative ability.The crucial limitation of all the studies presented above very likely liesin their correlational design and lack of potentialmediators. Other stud-ies showed that the association between intelligence and creativity ismediated by cognitive (Benedek et al., 2012; Benedek, Jauk et al.,2014;Nusbaum&Silvia, 2011b) aswell as personality-related processes(Jauk et al., 2013). Another apparent limitation may be the fact thatthese studies did not apply the latent variables approach – so typicalin the contemporary research on the creativity–intelligence relationship(Silvia, 2015). Convincing arguments (Silvia, 2015; Silvia & Beaty, 2012)exist that the relationship between intelligence and creativity is visiblystronger when properly controlled for measurement error. However,


http://www.erim.nl/nca



this limitation is apparent if we bear in mind that the very idea of theNCA assumes analyzing individual-level data. Researchers may, howev-er, consider using latent factor scores, for example estimated in IRTmodels (see Karwowski & Gralewski, 2013), together with their robuststandard errors to test the NCA. Among other limitations, onemay con-sider our use to self-reported creative activity and creative achievement(ICAA). Although creativity researchers have demonstrated that peopleusually have no problem with claiming that they are not creative andhave no creative achievements (Silvia, Wigert, Reiter-Palmon, & Kauf-man, 2012), future studies should replicate our findings with the useof more objective indices of creative activity and creative achievement,including for example experience-sampling measurement or diarystudies (Conner & Silvia, 2015; Silvia et al., 2014).

Finally, future examinations of the necessary condition relationshipwill also benefit from planned further developments of the NCA meth-od, including a formal statistical test of the d parameter and analysesof the robustness with respect to variable distribution characteristics.

4.4. Conclusion

Scholars interested in creative achievement have for years been pos-tulating that intelligence is a conditio sine qua non for creativity; yet,they tested this hypothesis in a suboptimal way. This study providesan example of applying the new methodology of estimating thenecessary-but-not-sufficient condition: the NCA. It demonstrates thatintelligence can indeed be perceived as a necessary-but-not-sufficientcondition of creative ability, creative activity, and creative achievement.

Acknowledgments

This research was partially supported by grants UMO-2011/03/N/HS6/05153 (Study 3) and UMO-2011/03/N/HS6/05073 (Study 2) fromthe National Science Centre, Poland.

References

Amabile, T. M. (1982). Social psychology of creativity: A consensual assessment tech-nique. Journal of Personality and Social Psychology, 43, 997–1013.

Amabile, T. M. (1996). Creativity in context. Boulder, CO: Westview Press.An, D., Song, Y., & Carr, M. (2016). A comparison of two models of creativity: Divergent

thinking and creative expert performance. Personality and Individual Differences, 90,78–84.

Arendasy, M., Hornke, L. F., Sommer, M., Häusler, J., Wagner-Menghin, M., Gittler, G., et al.(2004). Manual intelligence-structure-battery (INSBAT). Mödling: Schuhfried Gmbh.

Baddeley, A. (1968). A 3-min reasoning test based on grammatical transformation.Psychonomic Science, 10, 341–342.

Baer, J. (1993). Divergent thinking and creativity: A task-specific approach. Hillsdale, NJ.:Lawrence Erlbaum Associates.

Bates, T. C., & Shieles, A. (2003). Crystallized intelligence as a product of speed and drivefor experience: The relationship of inspection time and openness to g and Gc.Intelligence, 31(3), 275–287.

Batey, M., & Furnham, A. (2006). Creativity, intelligence, and personality: A critical reviewof the scattered literature. Genetic, Social, and General Psychology Monographs, 132,355–429.

Beaty, R. E., & Silvia, P. J. (2013). Metaphorically speaking: Cognitive abilities and the pro-duction of figurative language. Memory & Cognition, 41, 255–267.

Beghetto, R. A. (2006). Creative self-efficacy: Correlates in middle and secondary stu-dents. Creativity Research Journal, 18, 447–457.

Benedek, M., Borovnjak, B., Neubauer, A. C., & Kruse-Weber, S. (2014). Differences in cre-ativity and personality between classical, jazz, and folk musicians. Personality andIndividual Differences, 63, 117–121.

Benedek, M., Franz, F., Heene,M., & Neubauer, A. C. (2012). Differential effects of cognitiveinhibition and intelligence on creativity. Personality and Individual Differences, 53,480–485.

Benedek, M., Jauk, E., Sommer, M., Arendasy, M., & Neubauer, A. C. (2014). Intelligence,creativity, and cognitive control: The common and differential involvement of exec-utive functions in intelligence and creativity. Intelligence, 46, 73–83.

Brown, T. A. (2015). Confirmatory factor analysis for applied research. New York: GuilfordPublications.

Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytical studies. NewYork: Cambridge University Press.

Carson, S. H., Peterson, J. B., & Higgins, D. M. (2005). Reliability, validity, and factorstructure of the creative achievement questionnaire. Creativity ResearchJournal, 17, 37–50.


Chamorro-Premuzic, T., & Furnham, A. (2008). Personality, intelligence and approaches tolearning as predictors of academic performance. Personality and Individual Differences,44, 1596–1603.

Cho, S. H., Nijenhuis, J. T., van Vianen, A. E. M., Kim, H., & Lee, K. H. (2010). The relationbetween diverse components of intelligence and creativity. Journal of CreativeBehaviour, 44, 125–137.

Chruszczewski, M. H. (2009). Profile uzdolnień [Profiles of giftedness]. Warszawa:Wydawnictwa UW.

Chruszczewski, M. H. (2010). Myślenie skojarzeniowe i jego miejsce wuzdolnieniach artystycznych [Associative thinking and its place within artisticaptitudes]. Ruch Pedagogiczny, 81(3–4), 43–55.

Condon, D. M., & Revelle, W. (2014). The international cognitive ability resource: De-velopment and initial validation of a public-domain measure. Intelligence, 43,52–64.

Conner, T. S., & Silvia, P. J. (2015). Creative days: A daily diary study of emotion, person-ality, and everyday creativity. Psychology of Aesthetics, Creativity, and the Arts, 9,463–470.

Cox, C. (1926). The early mental traits of three hundred geniuses. Stanford: Stanford Univer-sity Press.

DeYoung, C. G. (2015). Cybernetic big five theory. Journal of Research in Personality, 56,33–58.

Dul, J. (2015). NCA: Necessary condition analysis [R Package]. ((https://cran.r-project.org/web/packages/NCA/index.html). Quick Start Guide available at http://papers.ssrn.com/sol3/papers.cfm?abstract_Id=2624981) or http://repub.eur.nl/pub/78323/).

Dul, J. (2016). Necessary condition analysis (NCA): Logic and methodology of “necessarybut not sufficient” causality. Organizational Research Methods, 19, 10–52.

Dul, J., & Ceylan, C. (2011). Work environments for employee creativity. Ergonomics, 54,12–20.

Dul, J., Hak, T., Goertz, G., & Voss, C. (2010). Necessary condition hypotheses in operationsmanagement. International Journal of Operations & Production Management, 30,1170–1190.

Dunbar, K. (1997). How scientists think: online creativity and conceptual change inscience. In T. B. Ward, S. M. Smith, & S. Vaid (Eds.), Conceptual structures and pro-cesses: Emergence, discovery, and change (pp. 461–493). Washington, DC: APAPress.

Dunkel, C. (2013). The general factor of personality and general intelligence: Evidence forsubstantial association. Intelligence, 41(5), 423–427. http://dx.doi.org/10.1016/j.intell.2013.06.010.

Dziedziewicz, D., & Karwowski, M. (2015). Development of children's creative visualimagination: A theoretical model and enhancement programmes. Education, 3–13(43), 382–392.

Dziedziewicz, D., Gajda, A., & Karwowski, M. (2014). Developing intercultural compe-tence and creativity. Thinking Skills and Creativity, 13, 32–42.

Ericsson, K. A. (2014). The road to excellence: The acquisition of expert performance in thearts and sciences, sports, and games. New York: Psychology Press.

Eysenck, H. J. (1995). Genius: The natural history of creativity. New York: Cambridge Uni-versity Press.

Feist, G. J. (1998). A meta-analysis of personality in scientific and artistic creativity.Personality and Social Psychology Review, 2, 290–309.

Feist, G. J., & Barron, F. X. (2003). Predicting creativity from early to late adulthood: Intel-lect, potential, and personality. Journal of Research in Personality, 37, 62–88.

Finke, R. A., Ward, T. B., & Smith, S. M. (1992). Creative cognition: Theory, research, and ap-plications. Bradford: MIT Press.

Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g? The relationship be-tween the scholastic assessment test and general cognitive ability. PsychologicalScience, 15, 641–651.

Fuchs-Beauchamp, K. D., Karnes, M. B., & Johnson, L. J. (1993). Creativity and intelligencein preschoolers. Gifted Child Quarterly, 37, 113–117.

Furnham, A., & Chamorro-Premuzic, T. (2006). Personality, intelligence, and generalknowledge. Learning and Individual Differences, 16, 79–90.

Gajda, A., Karwowski, M., & Beghetto, R. A. Creativity and academic achievement: A meta-analysis. Journal of Educational Psychology 2016 (in press).

Galton, F. (1886). Regression towards mediocrity in hereditary stature. The Journal of theAnthropological Institute of Great Britain and Ireland, 15, 246–263. http://dx.doi.org/10.2307/2841583.

Glăveanu, V. P. (2010). Paradigms in the study of creativity: Introducing the perspectiveof cultural psychology. New Ideas in Psychology, 28, 79–93.

Glăveanu, V. P. (2014). The psychology of creativity: A critical reading. Creativity. Theories– Research – Applications, 1, 10–32. http://dx.doi.org/10.15290/ctra.2014.01.01.02.

Gralewski, J., & Karwowski, M. (2012). Creativity and school grades: A case from Poland.Thinking Skills and Creativity, 7, 198–208.

Gralewski, J., & Karwowski, M. (2013). Polite girls and creative boys? Students' gendermoderates accuracy of teachers' ratings of creativity. Journal of Creative Behaviour,47, 290–304.

Gralewski, J., Weremczuk, E., & Karwowski, M. (2012). Intelligence and creativity of Polishmiddle-school students: Looking for the threshold hypothesis. New EducationalReview, 29, 328–338.

Guilford, J. P. (1967). The Nature of human intelligence. New York: McGraw-Hill.Hogan, R., & Hogan, J. (2007). Hogan personality inventory manual (3rd ed.). Tulsa, OK:

Hogan Assessment Systems.Hume, D. (1777). An enquiry concerning human understanding. London: Pearson.Hunt, E. (2010). Human intelligence. New York: Cambridge University Press.Jäger, A. O. (1984). Intelligenz struktur forschung: Konkurrierende Modelle neue

Ent-wicklungen Perspektiven [Research on the structure of intelligence: com-peting models developments perspectives]. Psychologische Rundschau, 35,21–35.


http://refhub.elsevier.com/S0160-2896(16)30028-9/rf0005


























































https://cran.r-project.org/web/packages/NCA/index.html

https://cran.r-project.org/web/packages/NCA/index.html

http://repub.eur.nl/pub/78323/




































http://dx.doi.org/10.2307/2841583

http://dx.doi.org/10.2307/2841583



http://dx.doi.org/10.15290/ctra.2014.01.01.02




















Jankowska, D. M., & Karwowski, M. (2015). Measuring creative imagery abilities. Frontiersin Psychology, 6, 1591. http://dx.doi.org/10.3389/fpsyg.2015.01591.

Jauk, E., Benedek, M., Dunst, B., & Neubauer, A. C. (2013). The relationship between intel-ligence and creativity: New support for the threshold hypothesis by means of empir-ical breakpoint detection. Intelligence, 41, 212–221.

Jauk, E., Benedek, M., & Neubauer, A. C. (2014). The road to creative achievement: A latentvariable model of ability and personality predictors. European Journal of Personality,28, 95–105.

Jauk, E., Neubauer, A. C., Dunst, B., Fink, A., & Benedek, M. (2015). Gray matter correlatesof creative potential: A latent variable voxel-based morphometry study. NeuroImage,111, 312–320.

Jaworowska, A., & Szustrowa, T. (2000). Test Matryc Ravena w wersji Standard. Formy:Klasyczna, Równoległa, Plus. Polskie standaryzacje [Raven's matrices test in standardform. The forms: classic, parallel, plus. Polish standardizations]. Warszawa: PracowniaTestów Psychologicznych Polskiego Towarzystwa Psychologicznego.

Jung, R. E., Gasparovic, C., Chavez, R. S., Flores, R. A., Smith, S. M., Caprihan, A., & Yeo, R. A.(2009). Biochemical support for the “threshold” theory of creativity: A magnetic res-onance spectroscopy study. Journal of Neuroscience, 29, 5319–5325.

Karwowski, M. (2011). It doesn't hurt to ask… but sometimes it hurts to believe. Predictorsof Polish students' creative self-efficacy. Psychology of Aesthetics, Creativity and the Arts, 5,154–164.

Karwowski, M., & Barbot, B. (2016). Creative self-beliefs: Their nature, development, andcorrelates. In J. C. Kaufman, & J. Baer (Eds.), Cambridge companion to reason and devel-opment (pp. 302–326). New York: Cambridge University Press.

Karwowski, M., & Gralewski, J. (2013). Threshold hypothesis: Fact or artifact? ThinkingSkills and Creativity, 8, 25–33.

Karwowski, M., & Lebuda, I. (2013). Extending climato-economic theory:When, how, andwhy it explains differences in nations' creativity. Behavioral and Brain Sciences, 36,493–494.

Karwowski, M., & Lebuda, I. (2015). The big five, the huge two, and creative self-beliefs: Ameta-analysis. Psychology of Aesthetics, Creativity, and the Arts. http://dx.doi.org/10.1037/aca0000035 (Oct 19, 2015).

Karwowski, M., & Lebuda, I. (2016). Creative self-concept: A surface characteristic of cre-ative personality. In G. Feist, R. Reiter-Palmon, & J. C. Kaufman (Eds.), Cambridgehandbook of creativity and personality research. New York: Cambridge UniversityPress (in press).

Kaufman, J. C. (2016). Creativity 101 (2nd ed.). New York: Springer.Kaufman, J. C., & Beghetto, R. A. (2009). Beyond big and little: The four C model of crea-

tivity. Review of General Psychology, 13, 1–12.Kaufman, S. B. (2013). Opening up openness to experience: A four-factor model and rela-

tions to creative achievement in the arts and sciences. Journal of Creative Behaviour,47, 233–255.

Kaufman, S. B., Quilty, L., Grazioplene, R., Hirsh, J., Peterson, J., & DeYoung, C. (2016).Openness to experience and intellect differentially predict creative achievement inthe arts and sciences. Journal of Personality, 84, 248–258.

Keith, T. Z., & Reynolds, M. R. (2010). Cattell–Horn–Carroll abilities and cognitive tests:What we've learned from 20 years of research. Psychology in the Schools, 47, 635–650.

Kharkhurin, A. V. (2014). Creativity.4in1: Four-criterion construct of creativity. CreativityResearch Journal, 26, 338–352.

Kim, K. H. (2005). Can only intelligent people be creative? A meta-analysis. Journal ofSecondary Gifted Education, 16, 57–66.

Lubart, T., Pacteau, C., Jacquet, A., & Caroff, X. (2010). Children's creative potential:An empirical study of measurement issues. Learning and Individual Differences,20, 388–392.

Madsen, A. M., Hodge, S. E., & Ottman, R. (2011). Causal models for investigating complexdisease. I. A primer. Human Heredity, 72, 54–62.

Matczak, A., Jaworowska, A., Ciechanowicz, A., & Stańczak, J. (2006). Bateria Testów APIS-Z.Podręcznik Wydanie II [An APIZ-Z tests battery. A manual (2nd ed.). Warszawa:Pracownia Testów Psychologicznych Polskiego Towarzystwa Psychologicznego.

Matczak, A., Jaworowska, A., & Stańczak, J. (2000). Rysunkowy Test Twórczego MyśleniaK.K. Urbana i H.G. Jellena TCT-DP. Podręcznik. (A drawing test of creative thinking byK.K. Urban and H.G. Jellen TCT-DP. A Manual). Warsaw: Pracownia TestówPsychologicznych Polskiego Towarzystwa Psychologicznego.

McGrew, K. S. (2009). CHC theory and the human cognitive abilities project: Standing onthe shoulders of the giants of psychometric intelligence research. Intelligence, 37,1–10.

Mourgues, C. V., Tan, M., Hein, S., Al-Harbi, K., Aljughaiman, A., & Grigorenko, E. L. (2015).The relationship between analytical and creative cognitive skills from middle child-hood to adolescence: Testing the threshold theory in the Kingdom of Saudi Arabia.Learning and Individual Differences. http://dx.doi.org/10.1016/j.lindif.2015.05.005.

Mueggo, V. M. (2008). Segmented: An R package to fit regression models with broken-line relationships. R News, 8, 20–25.

Mullineaux, P., & Dilalla, L. (2009). Preschool pretend play behaviors and early adolescentcreativity. Journal Of Creative Behavior, 43, 41–57.

Nusbaum, E. C., & Silvia, P. J. (2011a). Are openness and intellect distinct aspects of open-ness to experience? A test of the O/I model. Personality and Individual Differences, 51,571–574.

Nusbaum, E. C., & Silvia, P. J. (2011b). Are intelligence and creativity really so different?Fluid intelligence, executive processes, and strategy use in divergent thinking.Intelligence, 39, 36–45.

Park, G., Lubinski, D., & Benbow, C. P. (2007). Contrasting intellectual patterns for creativ-ity in the arts and sciences: Tracking intellectually precocious youth over 25 years.Psychological Science, 18, 948–952.

Park, G., Lubinski, D., & Benbow, C. P. (2008). Ability differences among people who havecommensurate degrees matter for scientific creativity. Psychological Science, 19,957–961.


Plucker, J. A. (1999). Is the proof in the pudding? Reanalyses of Torrance's (1958 to pres-ent) longitudinal data. Creativity Research Journal, 12, 103–114.

Plucker, J., Esping, A., Kaufman, J. C., & Avitia, M. A. (2015). Creativity and intelli-gence. In S. Goldstein, D. Princiotta, & J. A. Naglieri (Eds.), Handbook ofintelligence: Evolutionary theory, historical perspective, and current concepts(pp. 283–291). New York: Springer.

Preckel, F., Holling, H., & Wiese, M. (2006). Relationship of intelligence and creativity ingifted and non-gifted students: An investigation of threshold theory. Personality andIndividual Differences, 40, 159–170.

Raven, J. (2000). The Raven's progressive matrices: change and stability over culture andtime. Cognitive Psychology, 41, 1–48.

Raven, J., Court, J., & Raven, J. (1998). Raven's progressive matrices.Oxford: Oxford Psychol-ogists Press.

Rindermann, H., & Neubauer, A. (2004). Processing speed, intelligence, creativity, andschool performance: Testing of causal hypotheses using structural equation models.Intelligence, 32, 573–589.

Robertson, K. F., Smeets, S., Lubinski, D., & Benbow, C. P. (2010). Beyond the thresh-old hypothesis: Even among the gifted and top math/science graduate students,cognitive abilities, vocational interests, and lifestyle preferences matter for ca-reer choice, performance, and persistence. Current Direction in PsychologicalScience, 19, 346–351.

Runco, M. (2003). Education for creative potential. Scandinavian Journal of EducationalResearch, 47, 317–324.

Runco, M. A. (1991). The evaluative, valuative, and divergent thinking of children. Journalof Creative Behaviour, 25, 311–319.

Runco, M. A. (2007). Creativity. Theories and themes: Research, development, and practice.San Diego, CA: Elsevier Academic Press.

Runco, M. A., & Albert, R. S. (1986). The threshold theory regarding creativity and intelli-gence: An empirical test with gifted and nongifted children. Creative Child and AdultQuarterly, 11, 212–218.

Runco, M. A., & Jaeger, G. J. (2012). The standard definition of creativity. CreativityResearch Journal, 24, 92–96.

Runco, M. A., Millar, G., Acar, S., & Cramond, B. (2010). Torrance tests of creative thinkingas predictors of personal and public achievement: A fifty-year follow-up. CreativityResearch Journal, 22, 361–368.

Schwarzer, G. (2015). Meta: general package for meta-analysis. R package version, 4–1.Silvia, P. J. (2008a). Creativity and intelligence revisited: A latent variable analysis of Wal-

lach and Kogan (1965). Creativity Research Journal, 20, 34–39.Silvia, P. J. (2008b). Another look at creativity and intelligence: Exploring higher-order

models and probable confounds. Personality and Individual Differences, 44,1012–1021.

Silvia, P. J. (2015). Intelligence and creativity are pretty similar after all. EducationalPsychology Review, 27, 599–606.

Silvia, P. J., & Beaty, R. E. (2012). Making creative metaphors: The importance of fluid in-telligence for creative thought. Intelligence, 40, 343–351.

Silvia, P. J., Beaty, R. E., Nusbaum, E. C., Eddington, K. M., Levin-Aspenson, H., &Kwapil, T. R. (2014). Everyday creativity in daily life: An experience-samplingstudy of “little c” creativity. Psychology of Aesthetics, Creativity, and the Arts, 8,183–188.

Silvia, P. J., Nusbaum, E. C., Berg, C., Martin, C., & O'Connor, A. (2009). Openness to expe-rience, plasticity, and creativity: Exploring lower-order, high-order, and interactiveeffects. Journal of Research in Personality, 43, 1087–1090.

Silvia, P. J., Wigert, B., Reiter-Palmon, R., & Kaufman, J. C. (2012). Assessing creativity withself-report scales: A review and empirical evaluation. Psychology of Aesthetics,Creativity, and the Arts, 6, 19–34.

Silvia, P. J., Winterstein, B. B., Willse, J. T., Barona, C. M., Cram, J. T., Hess, K. I., ... Richard, C.A. (2008). Assessing creativity with divergent thinking tasks: Exploring the reliabilityand validity of new subjective scoring methods. Psychology of Aesthetics, Creativity,and the Arts, 2, 68–85.

Simonton, D. (1994). Greatness. New York: Guilford.Simonton, D. K. (1997). Creative productivity: A predictive and explanatory model of ca-

reer trajectories and landmarks. Psychological Review, 104, 66–89.Simonton, D. K. (2012). Taking the U.S. Patent Office criteria seriously: A quantitative

three-criterion creativity definition and its implications. Creativity Research Journal,24, 97–106.

Simonton, D. K. (2013). After Einstein: scientific genius is extinct. Nature, 493(7434), 602.Simonton, D. K. (2014). Creative performance, expertise acquisition, individual differ-

ences, and developmental antecedents: An integrative research agenda. Intelligence,45, 66–73.

Sligh, A. C., Conners, F. A., & Roskos-Ewoldsen, B. (2005). Relation of creativity to fluid andcrystalized intelligence. Journal of Creative Behaviour, 39, 123–136.

Spearman, C. E. (1927). The abilities of man. London: Macmillan.Sternberg, R. J. (2002). Creativity as a decision: Comment. American Psychologist, 57,

376.Sternberg, R. J., & Lubart, T. I. (1999). The concept of creativity: Prospects and paradigms.

In R. J. Sternberg (Ed.), Handbook of creativity (pp. 3–16). New York: Cambridge Uni-versity Press.

von Stumm, S. (2015). Investment traits and intelligence in adulthood. Journal ofIndividual Differences, 34, 82–89.

von Stumm, S., & Ackerman, P. L. (2013). Investment and intellect: A review and meta-analysis. Psychological Bulletin, 139, 841–869.

Torrance, E. (1972). Can we teach children to think creatively? Journal of CreativeBehaviour, 6, 114–143.

Torrance, E. P. (1962). Guiding creative talent. Englewood Cliffs, NJ: Prentice-Hall.Urban, K. K., & Jellen, H. G. (1996). Test for creative thinking-drawing production (TCT-DP).

Lisse, Netherland: Swets and Zeitlinger.


http://dx.doi.org/10.3389/fpsyg.2015.01591



























http://dx.doi.org/10.1037/aca0000035

http://dx.doi.org/10.1037/aca0000035


































http://dx.doi.org/10.1016/j.lindif.2015.05.005







































































































Vartanian, O., Martindale, C., & Kwiatkowski, J. (2003). Creativity and inductive rea-soning: The relationship between divergent thinking and performance onWason's 2–4–6 task. The Quarterly Journal of Experimental Psychology: Section A,56, 641–655.

Wai, J., Lubinski, D., & Benbow, C. P. (2005). Creativity and occupational accomplishmentsamong intellectually precocious youths: An age 13 to age 33 longitudinal study.Journal of Educational Psychology, 97, 484–492.


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

Weisberg, R.W. (2006). Creativity: Understanding innovation in problem solving, science, in-vention, and the arts. New York: John Wiley & Sons.

Wolfradt, U., & Pretz, J. E. (2001). Individual differences in creativity: Personality, storywriting, and hobbies. European Journal of Personality, 15, 297–310.
















is creativity without intelligence possible? a necessary ......1.1. a look at the...

Documents