Thursday, September 13, 2018

Meta-Analytic Investigations of the Relation Between Intuition and Analysis

Authors: Yi Wang (Bowling Green), Scott Highhouse (Bowling Green), Christopher J. Lake (University of Minnesota – Duluth), Nicole L. Petersen (Radford) and Thaddeus B. Rada (Bowling Green)
Journal: Journal of Behavioral Decision Making
Year: 2015

The authors used a meta-analytic study to determine the validity of two relationship models between intuition and analysis. The authors decided to conduct this study because they saw “a lack of consensus about the theoretical relation between the constructs (intuition and analysis).”

First, the authors defined the terms based on generally agreed upon characteristics as noted by the array of research on cognitive processing included in the study. The authors state intuition “refers to reliance on immediate, unconscious judgement based on feelings” and analysis “refers to reliance on deliberate, conscious judgement based on reasoning” citing studies from Allinson & Hayes (1996) and Epstein, Pacini, Danes-Raj, & Heier (1996).

The authors proceed by stating the two relationship models of intuition and analysis. The first model posits that intuition and analysis are opposite ends of the same cognitive spectrum, which are based on “theories of lateralization of brain function (Hines, 1987).” Furthermore, the authors cite Allinson and Hayes (1996), who provide a conceptualization of the lateralization of brain function, who state “intuition is a characteristic of right-brain function, and analysis is characteristic of left-brain function…”

The other model posits that intuition and analysis are independent and divergent modes of cognitive processing. The authors cite the cognitive-experiential self-theory (CEST) framework of Epstein and co., as the primary example whereby “people process information in two parallel interacting systems, rational versus experiential, which operate by different principles.” CEST sees the rational mode as one that requires substantive focus and processing in comparison to the experiential which is automatic. Switching between the two requires effort. The authors also cite “System 1 and System 2” model of Stanovich and West (2000).  

The evidence for the study consisted of previous analyses using Rational-Experiential Inventories (REI) based on CEST, cognitive-style index (CSI) developed by Allinson and Hayes (1996), and General Decision-making Style Inventory (GDMS). REI studies showed that the correlation between rationality (analysis) and experientiality (intuition) was “small and non-significant” suggesting that REI support the CEST conclusion of independence of cognitive processes. Allinson and Hayes developed CSI to assess the validity of the continuum approach, ultimately suggesting the analysis supported the continuum or interdimensional conclusion. GDMS assumed that the processes were independent, but the conclusion of the analysis suggested the opposite.

The authors hypothesized that “a meta-analytic correlation that is significantly negative would provide some support for the bipolar (continuum) model, whereas a lack of relation between the constructs would favor an independence model.” The authors used 80 studies in their meta-analysis. 67 of the 80 studies consisted of the REI (specifically the REI-20), GDMS, and Assessment of Career Decision-making Scale (ACDM) scales. The authors used these scales moderators in their analyses

In the first study, the authors conducted moderator analyses on the three primary scales to determine whether the relationship between intuition and analysis varied between them. The study found that each scale was an important moderator in the correlation between rationality and intuition. A second piece of the first study included examining analysis and intuition as it relates to the Big Five personality traits. The meta-analysis indicated that analysis and intuition tended to relate to different traits, although both analysis and intuition were balanced with regards to the traits of agreeableness and neuroticism. The results of the first study indicated that there was “near-zero correlation between intuition and analysis” which supports the independence model. The authors recognized that since scales were important moderators, especially the ACDM scale, which resulted in a second study.

The second study intercorrelated intuition and analysis across scales. The authors also eliminated the ACDM scale from the second study because of its focus on “career domains.” The second study used data from four scales: REI-31, GMS, DMI, and PID. The scales were randomly combined to create 12 pairs of scale combinations. The authors recruited 511 subjects that were randomly assigned to complete 1 of the 12 different scale pairings. The results of the study suggested that there “was no significant correlation between intuition and analysis across cognitive style subscales.” The authors conducted a hierarchical confirmatory factor analysis (CFA) to calculate the intercorrelation of intuition and analysis while also considering the validity measured by the different scales. According to the authors, a CFA should determine “ whether different scales of analysis and intuition provide equivalent reflection of the intended underlying constructs and (ii) whether there is still no correlation between analysis and intuition even after considering the measurement deficiency across different scales.” The authors found that the four scales included were not interchangeable. Additionally, they found that there is
“no meaningful correlation between intuition and analysis…at the construct level.”

The authors conclude that the studies conducted in the meta-analysis support the independence model of cognitive processing, i.e. that intuition and analysis are separate processes and do not exist on a single thinking continuum. The authors believe that results from the first study regarding the ACDM scale may be due to intuition and analysis have more relatedness in a career context. In the second study, the authors also found that intuition measured across scales consistently while analysis was inconsistent. The authors suggest the different scales may be measuring different underlying analysis constructs.

Disclaimer: Due to limited experience with statistical anaylsis, some of the previous summarization may be difficult to understand. My interpretation of the meta-analysis may use imprecise language or skip elements of the explained statistical analysis. This is because I found it difficult to convert the language into plain English.

My critique may result from my lack of understanding of the scales used. If any critique is evident in the meta-analysis, it appears that the authors heavily relied on REI-20 scales for the first study, which could have skewed the study towards the independence model of cognitive processing. The REI-20 scale represents more than have of the representative sample in the moderator analysis of the first study. Based on their description, it appears they may have controlled for this in the second study by creating mixed scales. Again, I will restate that it appears to me that the meta-analysis does not have an equal sample of analyses to run against. The samples seem to skew in the favor of the independence model from the start. Having read some of Daniel Kahneman’s Thinking Fast and Slow, the independence model does make sense. Based on personal observation, I find myself consciously switching between modes of thinking often, especially when solving complex problems. Unless the authors failed to mention some of the conclusions favored by some of the other scales used, I think that the meta-analysis would benefit from the inclusion of more studies which resulted in the favoring of the continuum model for balance.

Link to Study: 


  1. Based on the study determining that they are separate processes, did the authors discuss how one can improve their intuition based thinking?

  2. The authors did not discuss how one can improve their thinking. The purpose of the study was to determine the validity of the two differing models of cognitive processing.

  3. Harry, I think the final finding that the authors could measure intuition across scales consistently but not analyses poses a strong problem with their conclusions. If these authors are relying on the scales to determine the correlations between intuition and analysis, and the scale is not measuring analysis consistently, then these scales would be a poor measuring tool for this type of analysis. Do they acknowledge any limitations regarding this finding or suggestions on future research to better study correlations between intuition and analysis?

  4. In the first study the authors highlighted some issues that they saw with the data. According to the authors, they tried to resolve said issues in the second study by creating mixed scales, which yielded similar results.

    With regards to future research, the authors state: "Future research may need to explore how analysis and intuition contribute to certain attitudinal and behavioral outcomes that are important to their specific concerns."