As is known that we normally measure creativity by scoring its three dimensions: fluency, flexibility, and originality. However, Some questions arise that: 1) how to distinct these three in explaining creativity， 2) which one of the three is the speaker of creativity?
I see many papers showing that (e.g.) A ( a condition or a person) has higher ( any of the three, e.g., fluency) score than B condition but there is no difference in (another of the three e.g., originality) score (or may even in reverse pattern, A < B), In this situation, How do you explain these results? Can we say condition or person A is more creative than B?
In another case, I sometimes read papers which differentiate high or low creativity of individual difference by measuring fluency only(not other two) ! I do not think this methodology is convinced (if they have no differences in flexibility or originality).
Upon the above question, anyone who would like to discuss or recommend some references?
Dear Jim,
It is true that originality (or novelty) is a common criterion in most definitions of creativity. However, fluency, originality, and flexibility as scores on divergent thinking tests (DTT) are not indices or elements of creativity per se. In the context of the measurement, these indices should be primarily considered as facets of divergent thinking (DT). DTT are not direct measures of creativity, but they are useful instruments in measuring creative potential (see Runco, 2008; Runco & Acar, 2012).
As to your confusion, it is true that sometimes researches rely only on one specific dimension of DT (e.g., fluency). It indeed causes problems because in such cases the measurement lacks validity due to low differentiation between fluency in DTT and verbal fluency (from Thurstone's model of intelligence).
Why do researchers sometimes rely only on fluency? I think that there may be several consideration about the issue. First, this choice may be determined by specific research questions: researchers are interested only in one score (e.g., fluency or originality). I admit that this situation is highly hypothetical, but it is theoretically possible.
Second, sometimes some dimensions of DT are not calculated because they are considered as redundant. For example, flexibility is not computed in the figural form of Torrance's Test of Creative Thinking (TTCT) because it was shown that flexivility highly correlates with fluency (Hebert et al., 2002).
Third, it is true that fluency, originality, and flexibility are highly correlated in DTT, and some researcher prefer to use only one of them. Then, why fluency is chosen as a single score and not originality or flexibility? I think that the choice is determined by the inner voice of objectivity that is usually warranted in scientific research. Perhaps, fluency is seen as more objective measure (number of suggested responses) while originality and flexibility are more vulnerable to subjective bias. Runco (2010, p. 425) comments this in the following way:
"Another kind of discriminant validity involves the various indices of DT (e.g., fluency, originality, flexibility, and elaboration). On the one hand, three lines of investigation bring the discrimination of the various indices into question. First are the factor analyses of DT test scores that often uncover only one factor. Then there are multitrait/multimethod comparisons that suggest that scores within any one method (i.e., one test) are more highly correlated than indices (traits) across tests. And there are the simple product moment correlations, which also suggest redundancy, especially between fluency and originality, and between fluency and flexibility. For these three reasons it has been said that fluency might be used alone. After all, it predicts originality and flexibility."
However, it is worth mentioning that the differentiality of indeces may to some extent depend on testing conditions (e.g., instructions). For instance, "be fluent" instruction stresses that participant should generate many ideas and, thus, the fluency is more congruent with this instruction. But "be creative" instruction lay emphasis on producing highly original ideas and, thus, the originality is more congruent with this instruction (see further detailes in a wonderful review of Reiter-Palmon, Forthmann, & Barbot, 2019). Thus, you should pay attention to the instructions that are used with DTT. Nusbaum, Silvia, and Beaty (2014) argued that the application of "be creative" instruction is a methodological necessity for DT assessment. How do I know which instructions were used if it is not reported? You may look at correlation matrix and find the correlation coefficient between fluency and originality. If the correlation is positive and varies from medium to high, then the "be fluent" instruction was used. If the correlation is close to zero or negative, then it is more probable that the "be creative" instruction was used (although there are another possible instructions leading to similar effects).
Another concern is of confounding effect of fluency on originality. This problem was intensively investigated since late 70s. For example, Hocevar stated that the value of 82% correlation coefficients (N = 89) between fluency and originality exceeded the value 0.5, whereas the average value was 0.69 (Hocevar, 1979). Further, in the reanalysis of Wallach and Kogan (1965), Silvia found that the correlation between fluency and originality was 0.89 (Silvia, 2008). Curiously, in the analysis of the longitudinal study initiated under the guidance of E. P. Torrance, the correlations between the indices of fluency, flexibility, and originality vary in the moderate range (0.39 < r < 0.67), which rather confirms their relative independence from each other (Runco et al., 2010). In the study of Dumas and Dunbar, the latent variable approach was used to study the relationship between fluency and originality (Dumas & Dunbar, 2014). Fluency was determined by counting the total number of ideas, and originality was calculated using the method of latent semantic analysis (LSA). As a result, it was demonstrated that fluency and originality modeled as latent variables correlated with each other only at a moderate level (r = 0.38). This study allows us to suggest that the level of relationship between the variables of fluency and originality may vary depending on the method of modeling the variables. For further interesting details you can look at following works: Runco, Okuda, & Thurston (1987); Plucker & Runco (1998); Snyder et al. (2004). And of course you should consult the recent work of Forthmann, Szardenings, and Holling (2018).
I think that the best way to end is to refer again to the insightful remark from Mark Runco (2010, pp. 425-426):
"The use of only one index of DT is unfortunate on several grounds. First and least important, if only one index were to be used for some reason, fluency is probably not the best one. As noted above, originality is more strongly tied to creativity than is fluency, so if one index were to be used alone, it should probably be originality. Additionally, there is experimental evidence that the indices are independent and represent independent processes. Runco (1985), for example, demonstrated that explicit instructions to be flexible do not necessarily lead to high originality scores, and this fact suggests a kind of operational independence. Similarly, when explicitly directed to be original, fluency scores tend to drop (Harrington, 1975; Runco, 1986). A change in one index, in one direction, accompanied by a change in the opposite direction in another index, would not occur if they were interdependent. In addition, the correlations between fluency and originality or fluency and flexibility are large but not perfect. Hence, there is unique variance, even if it is smaller than that which is shared. Finally, regression techniques have removed variance shared by fluency and originality, allowing the reliability of the unique variance of originality (or flexibility) to be examined (Hocevar, 1980; Runco & Albert, 1985). The unique variance of originality is reliable, at least in some tests and individuals, at moderate or high levels of talent."
I hope this materials will help you!
Best regards,
Kirill
References:
I think this question is relevant to almost every human being who is known to practice critical thinking or who would would like to develop a career along this pathway.
The issues of the three dimensions: fluency, flexibility, and originality, are often seen to be intrinsically interconnect and in some cases appear to be somewhat indistinguishable or inseparable from one another.At least, most people see it that way.
Some people see them as perfect substitutes, whereas not. They are perfectly complementary but perfectly different from one another, just like spouses, family members, or relatives are always different but complimentary to one another or to each other.
My simple answer to these questions follow. to begin with, creativity is the ultimate audible or visible output or result that that involved organised personal inputs or process:
One way of how to distinct these three in explaining creativity, is to identify their individual nature or characteristics as the vital components of creativity itself. For example, fluency is a component of creativity which signifies the smooth flow or vibration of substance or emotion that resonates with a unique symbol or norm.
The element or component of flexibility refers to the emotional, spiritual, or practical ability of a composition that identifies with socio-cultural, economic, environmental, or political issues.
The third component of originality, is the individual personal, group, community, or institutional ownership of a making of a product or an output. All the three put together are bound to initiate or originate an output of creativity. If one of the three is to be appointed to be the speaker of creativity, it likely that, originality will be the official spokesperson.
Best regards
Wilson
I think this question is relevant to almost every human being who is known to practice critical thinking or who would would like to develop a career along this pathway.
The issues of the three dimensions: fluency, flexibility, and originality, are often seen to be intrinsically interconnect and in some cases appear to be somewhat indistinguishable or inseparable from one another.At least, most people see it that way.
Some people see them as perfect substitutes, whereas not. They are perfectly complementary but perfectly different from one another, just like spouses, family members, or relatives are always different but complimentary to one another or to each other.
My simple answer to these questions follow. to begin with, creativity is the ultimate audible or visible output or result that that involved organised personal inputs or process:
One way of how to distinct these three in explaining creativity, is to identify their individual nature or characteristics as the vital components of creativity itself. For example, fluency is a component of creativity which signifies the smooth flow or vibration of substance or emotion that resonates with a unique symbol or norm.
The element or component of flexibility refers to the emotional, spiritual, or practical ability of a composition that identifies with socio-cultural, economic, environmental, or political issues.
The third component of originality, is the individual personal, group, community, or institutional ownership of a making of a product or an output. All the three put together are bound to initiate or originate an output of creativity. If one of the three is to be appointed to be the speaker of creativity, it likely that, originality will be the official spokesperson.
Best regards
Wilson
In general, the three concepts have their own specific definitions; however, FLUENCY can be regarded as the basis upon which creativity is founded since it refers to such mental constructs as eloquence, articulacy, articulateness, expressiveness, communicativeness, coherence, cogency, and intelligibility, all of which are indicators of an individual's creativity.
These constructs can be understood by placing them in the context of a verbal test of creativity developed by E. P. Torrance. One of the most common questions Torrance used to assess creativity was "List all of the uses you can think of for tin cans." Fluency refers to the "total number" of uses the individual can think of - for example, an individual might write "a container for pencils, a container for jelly beans, abstract sculpture, a rolling pin, and using 5 different sized cans to make a Matryoshka doll" before running out of time. In this case, the fluency total is 5 or 5 different uses. Because two of the individual's answers are similar in that they refer to uses of tin cans for containing things, the Flexibility score is 4 because there are 4 "different types" of responses (i.e., container, artwork, tool for flattening, toy). Originality refers to an answer that is "rare but also has some practical merit" - perhaps using tin cans to form a Matryoshka nesting doll would qualify as an example of Originality. In Torrance's model, all three constructs are important aspects of creativity. Although Torrance claimed all three were important aspects of creativity, Originality appears to be the construct that best exemplifies the essence of creativity.
Here is an article that you might find helpful:
Kyung Hee Kim (2006). Can We Trust Creativity Tests? A Review of the Torrance Tests of Creative Thinking (TTCT). Creativity Research Journal, Vol. 18, No. 1, 3–14
Mark B. Scholl Thanks for recommending references~I agree that originality is the best speaker of creativity, though other two are important aspect. Reza Biria I agree with your understanding of fluency in creativity~ Wilson Truman Okaka yes, all three might be complementary in constituting creativity~
My another confusion is why some paper measure only one aspect, which was used to be the agent of creativity. if we put them together for comparisons, they are not consistent: in some case, why some paper only measure fluency (as index of creativity), in another case, why some paper only measure originality to be the index of creativity. That seems unequal or willful or subjective, since all three are basic elements of creativity. in short, all three should measured in the same study
Dear Jim,
It is true that originality (or novelty) is a common criterion in most definitions of creativity. However, fluency, originality, and flexibility as scores on divergent thinking tests (DTT) are not indices or elements of creativity per se. In the context of the measurement, these indices should be primarily considered as facets of divergent thinking (DT). DTT are not direct measures of creativity, but they are useful instruments in measuring creative potential (see Runco, 2008; Runco & Acar, 2012).
As to your confusion, it is true that sometimes researches rely only on one specific dimension of DT (e.g., fluency). It indeed causes problems because in such cases the measurement lacks validity due to low differentiation between fluency in DTT and verbal fluency (from Thurstone's model of intelligence).
Why do researchers sometimes rely only on fluency? I think that there may be several consideration about the issue. First, this choice may be determined by specific research questions: researchers are interested only in one score (e.g., fluency or originality). I admit that this situation is highly hypothetical, but it is theoretically possible.
Second, sometimes some dimensions of DT are not calculated because they are considered as redundant. For example, flexibility is not computed in the figural form of Torrance's Test of Creative Thinking (TTCT) because it was shown that flexivility highly correlates with fluency (Hebert et al., 2002).
Third, it is true that fluency, originality, and flexibility are highly correlated in DTT, and some researcher prefer to use only one of them. Then, why fluency is chosen as a single score and not originality or flexibility? I think that the choice is determined by the inner voice of objectivity that is usually warranted in scientific research. Perhaps, fluency is seen as more objective measure (number of suggested responses) while originality and flexibility are more vulnerable to subjective bias. Runco (2010, p. 425) comments this in the following way:
"Another kind of discriminant validity involves the various indices of DT (e.g., fluency, originality, flexibility, and elaboration). On the one hand, three lines of investigation bring the discrimination of the various indices into question. First are the factor analyses of DT test scores that often uncover only one factor. Then there are multitrait/multimethod comparisons that suggest that scores within any one method (i.e., one test) are more highly correlated than indices (traits) across tests. And there are the simple product moment correlations, which also suggest redundancy, especially between fluency and originality, and between fluency and flexibility. For these three reasons it has been said that fluency might be used alone. After all, it predicts originality and flexibility."
However, it is worth mentioning that the differentiality of indeces may to some extent depend on testing conditions (e.g., instructions). For instance, "be fluent" instruction stresses that participant should generate many ideas and, thus, the fluency is more congruent with this instruction. But "be creative" instruction lay emphasis on producing highly original ideas and, thus, the originality is more congruent with this instruction (see further detailes in a wonderful review of Reiter-Palmon, Forthmann, & Barbot, 2019). Thus, you should pay attention to the instructions that are used with DTT. Nusbaum, Silvia, and Beaty (2014) argued that the application of "be creative" instruction is a methodological necessity for DT assessment. How do I know which instructions were used if it is not reported? You may look at correlation matrix and find the correlation coefficient between fluency and originality. If the correlation is positive and varies from medium to high, then the "be fluent" instruction was used. If the correlation is close to zero or negative, then it is more probable that the "be creative" instruction was used (although there are another possible instructions leading to similar effects).
Another concern is of confounding effect of fluency on originality. This problem was intensively investigated since late 70s. For example, Hocevar stated that the value of 82% correlation coefficients (N = 89) between fluency and originality exceeded the value 0.5, whereas the average value was 0.69 (Hocevar, 1979). Further, in the reanalysis of Wallach and Kogan (1965), Silvia found that the correlation between fluency and originality was 0.89 (Silvia, 2008). Curiously, in the analysis of the longitudinal study initiated under the guidance of E. P. Torrance, the correlations between the indices of fluency, flexibility, and originality vary in the moderate range (0.39 < r < 0.67), which rather confirms their relative independence from each other (Runco et al., 2010). In the study of Dumas and Dunbar, the latent variable approach was used to study the relationship between fluency and originality (Dumas & Dunbar, 2014). Fluency was determined by counting the total number of ideas, and originality was calculated using the method of latent semantic analysis (LSA). As a result, it was demonstrated that fluency and originality modeled as latent variables correlated with each other only at a moderate level (r = 0.38). This study allows us to suggest that the level of relationship between the variables of fluency and originality may vary depending on the method of modeling the variables. For further interesting details you can look at following works: Runco, Okuda, & Thurston (1987); Plucker & Runco (1998); Snyder et al. (2004). And of course you should consult the recent work of Forthmann, Szardenings, and Holling (2018).
I think that the best way to end is to refer again to the insightful remark from Mark Runco (2010, pp. 425-426):
"The use of only one index of DT is unfortunate on several grounds. First and least important, if only one index were to be used for some reason, fluency is probably not the best one. As noted above, originality is more strongly tied to creativity than is fluency, so if one index were to be used alone, it should probably be originality. Additionally, there is experimental evidence that the indices are independent and represent independent processes. Runco (1985), for example, demonstrated that explicit instructions to be flexible do not necessarily lead to high originality scores, and this fact suggests a kind of operational independence. Similarly, when explicitly directed to be original, fluency scores tend to drop (Harrington, 1975; Runco, 1986). A change in one index, in one direction, accompanied by a change in the opposite direction in another index, would not occur if they were interdependent. In addition, the correlations between fluency and originality or fluency and flexibility are large but not perfect. Hence, there is unique variance, even if it is smaller than that which is shared. Finally, regression techniques have removed variance shared by fluency and originality, allowing the reliability of the unique variance of originality (or flexibility) to be examined (Hocevar, 1980; Runco & Albert, 1985). The unique variance of originality is reliable, at least in some tests and individuals, at moderate or high levels of talent."
I hope this materials will help you!
Best regards,
Kirill
References:
Kirill Miroshnik
Dear Kirill,
Thank you very much for your detained explanations as well as the many references. They are really of great help to resolve my confusion.
One point I noticed in your letter is worthy of discussion. On one hand, many studies found great correlation between fluency and originality (may also the case with flexibility), indicating fluency and originality are interdependent. On the other hand, being instructed to be more fluency may decrease originality scores (vice verse), indicating that these two may be independent with each other. So what is the fact? This contradiction sounds very interesting. In my personal bias, I tend to believe that the two should be essentially different with each other: fluency emphasize more on quantity, whereas originality emphasize more on unique quality. they are completely different dimensions.
To resolve the seemingly contradiction, I suggest 1) to reconsider the validity of the method of measurement; and 2) to understand the nature of three and the relationship between them.
Anyway, this sounds an interesting contradiction.
Thanks again!
Best regards,
Jim
Dear Jim,
I will try to answer your questions and to express my point of view.
1. "...many studies found great correlation between fluency and originality (may also the case with flexibility)...".
You are right in your suggestions. It has been shown that fluency and flexibility are also highly correlated (e.g., Cohen & Oden, 1974; Plass, Michael, & Michael, 1974). Likewise, several studies of the nonverbal battery of TTCT (Torrance Test of Creative Thinking) in many countries corroborated that fluency, originality, and flexibility form one single factor (Clapham, 1998; Kim, 2006; Almeida, Prieto, Ferrando, Oliveira, & Ferrándiz, 2008; Krumm, Lemos, & Filippetti, 2014; García, Ferrando, Soto, Sáinz, & Prieto, 2016; Humble, Dixon, & Mpofu, 2017; Yoon, 2017), whereas elaboration forms a separate factor.
2. "On one hand, many studies found great correlation between fluency and originality (may also the case with flexibility), indicating fluency and originality are interdependent. On the other hand, being instructed to be more fluency may decrease originality scores (vice verse), indicating that these two may be independent with each other. So what is the fact?"
Any scientific fact is always theoretically loaded. Therefore, a solution of empirical contradictions lies in finding a common theoretical ground.
First, stating the contradiction, you used the terms fluency, originality, and flexibility without context. Thus, you have made an assumption that fluency, originality, and flexibility are universally defined and calculated. Is this assumption justified? While, in general, it is true for fluency (but see also Snyder, Mitchell, Bossomaier, & Pallier, 2004), that is definitely not so for originality and flexibility. For instance, you can compute originality based on normative data, based on frequencies (only unique responses are counted, Wallach & Kogan, 1965; only those responses that are suggested no more than 1%, 2%, 5%, or even 10% of the sample, Runco, Abdulla, Paek, Al-Jasim, & Alsuwaidi, 2016), as the ratio of originality score to fluency score (Hocevar, 1979), based on latent semantic analysis (e.g., Dumas & Dunbar, 2014), based on subjective scoring techniques (e.g., Silvia et al., 2008), and simply limit the number of possible responses (Guilfrod's Manual for Alternate Uses; Top 2, see Silvia et al., 2008). I did not attempt to provide exhaustive variants of computing originality in divergent thinking tests. I rather suggested those ways which came to mind. As I wrote to you elsewhere, you can also compute flexibility in various ways: (1) the number of used categories of responses or (2) as the number of switches between categories (e.g., Runco, 1986). As for fluency, I believe there is less variation, but nevertheless, some discrepancies also appear. For example, Guilford (1957) differentiated between various types of fluency: word fluency, ideational fluency, associational fluency, expressional fluency. Moreover, if we use the old/new scoring method proposed by Gilhooly et al. (2007) and try to compute the proportion of new ideas (number of new ideas/total fluency), can we consider it as a separate way of computing fluency? Therefore, in my point of view, it would be more appropriate to talk not about fluency, originality, and flexibility, but fluencies, originalities, and flexibilities. It means that we should not forget that these indices can be computed differently and, thus, the contradictions in empirical data arises.
Second, Guilford's indices (fluency, originality, and flexibility) are derived not on theoretical, but on the empirical ground. I am not sure as to whether the relationship between fluency, originality, and flexibility has any theoretical precondition. If it is truly so, there is no hypothetical suggestion how these indices should "behave". Thus, everything that we get as results we take at its face value.
Third, in the case with the application of "be fluent" instruction, it is true that fluency (number of responses) is highly correlated with originality (computed as a summative score). But it does not follow that originality decreases in the condition of "be fluent" instruction. Because if the variables are highly positively correlated, it means that the more ideas are produced, the higher the total originality score. But the total originality score represents only the sum of the scores for each idea. Therefore, it does not mean that highly original ideas were not generated. In fact, these highly original ideas are "dissolved in the sea of many less original ideas". Thus, the question is not in the decrease of originality under "be fluent" condition, but in the fact that originality is confounded by fluency. Using appropriate procedures of computing originality (average/proportion scoring) can solve this problem and provide statistical independence between fluency and originality (please, read the paper of Forthmann, Szardenings, & Holling, 2018). On the other hand, you are absolutely right that under "be original" condition the fluency score decreases (the correlations of fluency with originality are usually moderately negative; see Silvia, Nusbaum, & Beaty, 2015). Interestingly, in a similar vein, Runco and Okuda (1988) used explicit instructions to test the relationship between fluency, originality, and flexibility. As Runco (2010) writes about the results of this study: "They expected originality and flexibility to be particularly strongly related, the assumption being that flexible ideation leads to original ideas. Surprisingly, results indicated that flexibility and originality were unrelated. This pattern was suggested by the flexibility scores in the originality condition, and the originality scores in the flexibility instructional condition. Most important were the originality scores, which were low when the participants were asked to be flexible in their ideation. Participants had been given strategies for proposing flexible ideas, but these did not lead to highly original ideas." (p. 431-432). In sum, fluency, originality, and flexibility can be independent of each other in certain testing conditions and with the application of appropriate computational formulas.
3. " In my personal bias, I tend to believe that the two should be essentially different with each other: fluency emphasize more on quantity, whereas originality emphasize more on unique quality. they are completely different dimensions".
Even though I agree with you that quantity and quality are distinct dimensions, I want to raise one concern regarding this issue. More specifically, distinctiveness does not mean total independence. For instance, height and weight are different variables which are not equivalent. However, they are at least moderately correlated (more than 0.4). Moreover, the types of relationship are not exhausted by correlation. In fact, the more ideas you produce, the higher the probability that you get a highly original one (see D. K. Simonton's works)! Likewise, Linus Pauling said that the best way to get an original idea is to have many ideas. Therefore, the relationship can also be dialectical which means that quantity transfoms into quality.
4. "To resolve the seemingly contradiction, I suggest 1) to reconsider the validity of the method of measurement; and 2) to understand the nature of three and the relationship between them".
I support your desire to get into this problem because it is truly important for the field. Although some issues are still unresolved, you may learn many things from the current literature. I think it will help you to state the problem more specifically.
I hope it helps.
Best regards,
Kirill
References:
In my opinion, every scholar who faces a new project requires all creativity ,fluency , flexibility, and originality
Dr Kamath Madhusudhana
I read some research today out of a European country which clearly showed the Torrance Tests to be able to clearly identify those three main constructs...Other tests do not seem to be able to identify those three factors as succinctly and separately as the TTCT.
- Badges
- Science topic
- Similar topics
- Papers