Are Conscientiousness and intelligence inversely correlated, as some studies report? A new paper seems to solve the Conscientiousness/IQ question (they are uncorrelated, the inverse correlations were being induced by a selection effect, https://en.wikipedia.org/wiki/Berkson%27s_paradox), and second, it demonstrates that twin registries/studies are useful even for non-genetic questions - in this case, just getting a large representative sample free of selection effects with both personality and grade covariates.
"How are conscientiousness and cognitive ability related to one another? A re-examination of the intelligence compensation hypothesis", Murray et al 2014 (https://pdf.yt/d/Dfl1N6pbR-4vYaKk)
"Previously, negative associations between intelligence and conscientiousness have been reported and explained in terms of an ‘intelligence compensation hypothesis’ (ICH) whereby higher conscientiousness develops in order to compensate for lower cognitive ability. We argue that conscientious traits, especially those related to achievement, are just as likely to be reinforced by cognitive ability. We evidence this by showing that previous negative associations may be attributable to a compensatory sample selection effect arising because of the use of research samples comprised of participants with achievement above certain thresholds. The associations between conscientiousness and ability in the samples of adolescents and their parents from the Sibling Interaction and Behaviour Study (SIBS) and Minnesota Twin Family Study (MTFS) – which were not selected in this way – were either zero or positive. Further, artificially introducing selection on achievement into these samples biased the associations in the negative direction. Together, these results are consistent with the hypothesis that the true association between these constructs may be zero or positive at the population level but that the use of selected research samples has sometimes resulted in the appearance of a negative association.
A number of studies have reported negative correlations between cognitive ability and conscientiousness-related personality traits (e.g., Furnham, Dissou, Sloan, & Chamorro-Premuzic, 2007; Furnham & Moutafi, 2012; Furnham, Moutafi, & ChamorroPremuzic, 2005; Moutafi, Furnham, & Crump, 2003, 2006; Moutafi, Furnham, & Paltiel, 2004; Soubelet & Salthouse, 2011; Wood & Englert, 2009). Moutafi et al. (2004) proposed an intelligence compensation hypothesis (ICH) to explain this negative association, with subsequent replications often being interpreted as support for the hypothesis. The hypothesis states that individuals of lower cognitive ability become more conscientious in striving for similar levels of achievement to their peers with higher cognitive ability. Individuals higher in cognitive ability are proposed not to increase in conscientiousness because their higher cognitive ability allows them to accomplish more with the same or less effort. Thus, there is no incentive for them to invest in approaching life more conscientiously. However, the evidence for ICH is mixed. Counter to the hypothesis, positive associations between cognitive ability and conscientiousness have been observed (e.g., Baker & Bichsel, 2006; Lounsbury, Welsh, Gibson, & Sundstrom, 2005; Luciano et al., 2006) and other studies have yielded associations that were close to zero or non-significant (e.g., Bartels et al., 2012; Chamorro-Premuzic, Moutafi, & Furnham, 2005; Furnham et al., 2005). Not all studies reporting an association between cognitive ability and conscientiousness did so with the explicit aim of testing the ICH but they nonetheless contribute to the pool of evidence to be considered in evaluating the hypothesis.
A feature which partially distinguishes those studies supporting the ICH from those which do not is sample composition. The majority of studies supporting ICH have been conducted in samples which may be selected with respect to occupational or academic achievement. For example, the studies of Moutafi et al. (2004) and Furnham and Moutafi (2012) used samples of junior to middle managers attending staff development centres, whilst other studies have utilised samples of managerial grade job applicants attending assessment centres (Furnham et al., 2007; Wood & Englert, 2009). Development and assessment centres are costly (Eurich, Krause, Cigularow, & Thorton, 2009). As such, in selection situations, organisations tend only to invite small percentages of the total applicant pools to attend these centres and in training contexts, their use is more common amongst managerial and professional populations (Meriac, Hoffman, Woehr, & Fleisher, 2008; Pepermans, Vloeberghs, & Perkisas, 2003). Another study finding a negative IQ–conscientiousness association used a sample of undergraduate students (Furnham et al., 2005) and entry to university involves selection on prior academic achievement (e.g., Hägglund & Larsson, 2006). Similarly, a study by Soubelet and Salthouse (2011) analysed data from participants who had an average of almost 16 years of education and were approximately 2/3 to 1 standard deviation above the national norms on cognitive ability.
...Thus, one could think of selection into the research sample being based on a composite of IQ and Conscientiousness. Whenever IQ is relatively low, a large enough value on the composite to reach the occupational or educational achievement level necessary for selection into the relevant population can only be achieved by having high Conscientiousness. Conversely, when Conscientiousness is relatively low, IQ must be high to obtain a high enough composite score for selection. Thus, a higher score on one trait necessarily compensates for a lower score on the other. A research sample based on a population selected in this way could yield a negative correlation between IQ and Conscientiousness even if they are un- or positively correlated in the population because it will tend to have a greater proportion of people with discrepant IQ–Conscientiousness scores than the general population.
Conscientiousness was measured using a 198-item version of the multidimensional personality questionnaire (MPQ; Tellegen & Waller, 2008). The MPQ contains two conscientiousness-related traits: Control and Achievement. Here we re-label the Achievement scale ‘Achievement-Striving’ to avoid confusion with our measures of occupational and educational achievement. High scorers on Control describe themselves as reflective; cautious, careful, plodding; rational, sensible, level-headed; liking to plan activities in detail. High scorers on Achievement-Striving describe themselves as working hard, driving themselves; welcoming difficult and demanding tasks; persisting when others give up; ambitious; putting work and accomplishments before many other things; setting high standards; being perfectionistic. Items were measured on a 4-point response scale from ‘Definitely True’ to ‘Definitely False’ and each scale has 18 items. Here we utilised the scale scores for the two measures. Gaughan, Miller, Pryor, and Lynam (2009) reported the highest correlations of MPQ Control to be with the Order (r = .56) and Deliberation (r = .68) facets of Conscientiousness in the NEO-PI-R, whilst MPQ Achievement-Striving correlated most highly with the Achievement Striving (r = .60) and SelfDiscipline (r = .52) facets. The IQ measure completed by participants was an abbreviated version of the Wechsler Adult Intelligence Scale Revised (WAISR; Wechsler, 1974) and included the Vocabulary, Information, Block Design and Picture Arrangement subtests. These subtests were chosen based on their high correlation (.90) with IQ derived from all the subtests.
For the adolescent sample we used grade point average (GPA) as a measure of educational achievement. To avoid problems of comparing grades across different school districts with different testing formats, procedures and standards, GPA was not computed from actual grades. Instead twins and their parents were asked to report, on a 5-point scale from 0 = failed class to 4 = much better than average, the grades typically received in language arts, maths, social studies and science classes. Here, GPA was the average across these ratings. This measure was validated against the actual school grades of a sub-sample of 67 randomly selected participants from the age-11 cohort and found to correlate with these at .89. For the parent sample, we used occupational level according to the Hollingshead’s (1957) occupational scale as a measure of occupational achievement. This is an eight-point scale ranging from ‘unskilled’ to ‘major professional’. Higher ratings on the scale reflect higher levels of occupational achievement.
...We introduced selection by discarding all individuals who were below progressively increasing thresholds of educational or occupational achievement. This was designed to mimic processes of selection into populations (e.g., undergraduate students, or assessment centre participants) to some degree dependent on educational or occupational achievement. We evaluated the correlations between IQ and our conscientiousness measure in the progressively more selected samples.
1.3.2. Moderation analysis
We evaluated whether educational or occupational achievement moderated the effect of IQ on cognitive ability using multiple regression models. One model was estimated for each of the measures of Conscientiousness in each of the samples. In these models the predictors were IQ, Achievement (occupational level for the parent sample and GPA for the adolescent sample) and the interaction between IQ and Achievement. The outcome variable was the Conscientiousness measure (Control or Achievement-striving). IQ and Achievement were both centered prior to analysis. A statistically significant interaction term was considered to be evidence in favour of moderation of the relation between IQ and cognitive
ability by achievement.
...In the unselected adolescent sample there was no statistically significant association between IQ and Control (r = .04, p = .06) but a statistically significant positive association between IQ and Achievement-Striving (r = .14, p < .01). In the unselected parent sample there was a small but statistically significant positive association between IQ and Control (r = .05, p < .01) but no statistically significant association between IQ and Achievement-Striving (r = .03, p = .15).
2.2. Effect of selection on Conscientiousness–IQ association Tables 1 and 2 show the correlations of IQ with the Control and Achievement-Striving personality scales when the full samples were subjected to selection on educational or occupational achievement. They show the downward trajectories of the correlations as samples became increasingly selected on achievement or intelligence
In the adolescent sample, the initial non-significant positive association between IQ and Control in the full sample (r = .04, p = .06) became steadily attenuated and then negative with selection on GPA. At the highest level of GPA, the association was r = .06 (p = .22). A similar albeit more subtle effect occurred in the correlation between IQ and Achievement-Striving, which began at r = .14 (p < .01) and decreased to r = .08 (p = .13) in the most selected group.
In the parent sample, selection on occupational level had little effect on the correlation between IQ and Control. It reduced from .05 to .01 and then rose again to .03 at the highest level of selection. There was a more marked effect of selection on the correlation between IQ and Achievement-Striving. With increasing degrees of selection, it first became steadily attenuated to zero with and then became negative. Although there was no significant association between IQ and Achievement-Striving in the full sample, at the highest level of selection there was a statistically significant negative association (r = .13, p = .03)."
#conscientousness #intelligence #personality
"How are conscientiousness and cognitive ability related to one another? A re-examination of the intelligence compensation hypothesis", Murray et al 2014 (https://pdf.yt/d/Dfl1N6pbR-4vYaKk)
"Previously, negative associations between intelligence and conscientiousness have been reported and explained in terms of an ‘intelligence compensation hypothesis’ (ICH) whereby higher conscientiousness develops in order to compensate for lower cognitive ability. We argue that conscientious traits, especially those related to achievement, are just as likely to be reinforced by cognitive ability. We evidence this by showing that previous negative associations may be attributable to a compensatory sample selection effect arising because of the use of research samples comprised of participants with achievement above certain thresholds. The associations between conscientiousness and ability in the samples of adolescents and their parents from the Sibling Interaction and Behaviour Study (SIBS) and Minnesota Twin Family Study (MTFS) – which were not selected in this way – were either zero or positive. Further, artificially introducing selection on achievement into these samples biased the associations in the negative direction. Together, these results are consistent with the hypothesis that the true association between these constructs may be zero or positive at the population level but that the use of selected research samples has sometimes resulted in the appearance of a negative association.
A number of studies have reported negative correlations between cognitive ability and conscientiousness-related personality traits (e.g., Furnham, Dissou, Sloan, & Chamorro-Premuzic, 2007; Furnham & Moutafi, 2012; Furnham, Moutafi, & ChamorroPremuzic, 2005; Moutafi, Furnham, & Crump, 2003, 2006; Moutafi, Furnham, & Paltiel, 2004; Soubelet & Salthouse, 2011; Wood & Englert, 2009). Moutafi et al. (2004) proposed an intelligence compensation hypothesis (ICH) to explain this negative association, with subsequent replications often being interpreted as support for the hypothesis. The hypothesis states that individuals of lower cognitive ability become more conscientious in striving for similar levels of achievement to their peers with higher cognitive ability. Individuals higher in cognitive ability are proposed not to increase in conscientiousness because their higher cognitive ability allows them to accomplish more with the same or less effort. Thus, there is no incentive for them to invest in approaching life more conscientiously. However, the evidence for ICH is mixed. Counter to the hypothesis, positive associations between cognitive ability and conscientiousness have been observed (e.g., Baker & Bichsel, 2006; Lounsbury, Welsh, Gibson, & Sundstrom, 2005; Luciano et al., 2006) and other studies have yielded associations that were close to zero or non-significant (e.g., Bartels et al., 2012; Chamorro-Premuzic, Moutafi, & Furnham, 2005; Furnham et al., 2005). Not all studies reporting an association between cognitive ability and conscientiousness did so with the explicit aim of testing the ICH but they nonetheless contribute to the pool of evidence to be considered in evaluating the hypothesis.
A feature which partially distinguishes those studies supporting the ICH from those which do not is sample composition. The majority of studies supporting ICH have been conducted in samples which may be selected with respect to occupational or academic achievement. For example, the studies of Moutafi et al. (2004) and Furnham and Moutafi (2012) used samples of junior to middle managers attending staff development centres, whilst other studies have utilised samples of managerial grade job applicants attending assessment centres (Furnham et al., 2007; Wood & Englert, 2009). Development and assessment centres are costly (Eurich, Krause, Cigularow, & Thorton, 2009). As such, in selection situations, organisations tend only to invite small percentages of the total applicant pools to attend these centres and in training contexts, their use is more common amongst managerial and professional populations (Meriac, Hoffman, Woehr, & Fleisher, 2008; Pepermans, Vloeberghs, & Perkisas, 2003). Another study finding a negative IQ–conscientiousness association used a sample of undergraduate students (Furnham et al., 2005) and entry to university involves selection on prior academic achievement (e.g., Hägglund & Larsson, 2006). Similarly, a study by Soubelet and Salthouse (2011) analysed data from participants who had an average of almost 16 years of education and were approximately 2/3 to 1 standard deviation above the national norms on cognitive ability.
...Thus, one could think of selection into the research sample being based on a composite of IQ and Conscientiousness. Whenever IQ is relatively low, a large enough value on the composite to reach the occupational or educational achievement level necessary for selection into the relevant population can only be achieved by having high Conscientiousness. Conversely, when Conscientiousness is relatively low, IQ must be high to obtain a high enough composite score for selection. Thus, a higher score on one trait necessarily compensates for a lower score on the other. A research sample based on a population selected in this way could yield a negative correlation between IQ and Conscientiousness even if they are un- or positively correlated in the population because it will tend to have a greater proportion of people with discrepant IQ–Conscientiousness scores than the general population.
Conscientiousness was measured using a 198-item version of the multidimensional personality questionnaire (MPQ; Tellegen & Waller, 2008). The MPQ contains two conscientiousness-related traits: Control and Achievement. Here we re-label the Achievement scale ‘Achievement-Striving’ to avoid confusion with our measures of occupational and educational achievement. High scorers on Control describe themselves as reflective; cautious, careful, plodding; rational, sensible, level-headed; liking to plan activities in detail. High scorers on Achievement-Striving describe themselves as working hard, driving themselves; welcoming difficult and demanding tasks; persisting when others give up; ambitious; putting work and accomplishments before many other things; setting high standards; being perfectionistic. Items were measured on a 4-point response scale from ‘Definitely True’ to ‘Definitely False’ and each scale has 18 items. Here we utilised the scale scores for the two measures. Gaughan, Miller, Pryor, and Lynam (2009) reported the highest correlations of MPQ Control to be with the Order (r = .56) and Deliberation (r = .68) facets of Conscientiousness in the NEO-PI-R, whilst MPQ Achievement-Striving correlated most highly with the Achievement Striving (r = .60) and SelfDiscipline (r = .52) facets. The IQ measure completed by participants was an abbreviated version of the Wechsler Adult Intelligence Scale Revised (WAISR; Wechsler, 1974) and included the Vocabulary, Information, Block Design and Picture Arrangement subtests. These subtests were chosen based on their high correlation (.90) with IQ derived from all the subtests.
For the adolescent sample we used grade point average (GPA) as a measure of educational achievement. To avoid problems of comparing grades across different school districts with different testing formats, procedures and standards, GPA was not computed from actual grades. Instead twins and their parents were asked to report, on a 5-point scale from 0 = failed class to 4 = much better than average, the grades typically received in language arts, maths, social studies and science classes. Here, GPA was the average across these ratings. This measure was validated against the actual school grades of a sub-sample of 67 randomly selected participants from the age-11 cohort and found to correlate with these at .89. For the parent sample, we used occupational level according to the Hollingshead’s (1957) occupational scale as a measure of occupational achievement. This is an eight-point scale ranging from ‘unskilled’ to ‘major professional’. Higher ratings on the scale reflect higher levels of occupational achievement.
...We introduced selection by discarding all individuals who were below progressively increasing thresholds of educational or occupational achievement. This was designed to mimic processes of selection into populations (e.g., undergraduate students, or assessment centre participants) to some degree dependent on educational or occupational achievement. We evaluated the correlations between IQ and our conscientiousness measure in the progressively more selected samples.
1.3.2. Moderation analysis
We evaluated whether educational or occupational achievement moderated the effect of IQ on cognitive ability using multiple regression models. One model was estimated for each of the measures of Conscientiousness in each of the samples. In these models the predictors were IQ, Achievement (occupational level for the parent sample and GPA for the adolescent sample) and the interaction between IQ and Achievement. The outcome variable was the Conscientiousness measure (Control or Achievement-striving). IQ and Achievement were both centered prior to analysis. A statistically significant interaction term was considered to be evidence in favour of moderation of the relation between IQ and cognitive
ability by achievement.
...In the unselected adolescent sample there was no statistically significant association between IQ and Control (r = .04, p = .06) but a statistically significant positive association between IQ and Achievement-Striving (r = .14, p < .01). In the unselected parent sample there was a small but statistically significant positive association between IQ and Control (r = .05, p < .01) but no statistically significant association between IQ and Achievement-Striving (r = .03, p = .15).
2.2. Effect of selection on Conscientiousness–IQ association Tables 1 and 2 show the correlations of IQ with the Control and Achievement-Striving personality scales when the full samples were subjected to selection on educational or occupational achievement. They show the downward trajectories of the correlations as samples became increasingly selected on achievement or intelligence
In the adolescent sample, the initial non-significant positive association between IQ and Control in the full sample (r = .04, p = .06) became steadily attenuated and then negative with selection on GPA. At the highest level of GPA, the association was r = .06 (p = .22). A similar albeit more subtle effect occurred in the correlation between IQ and Achievement-Striving, which began at r = .14 (p < .01) and decreased to r = .08 (p = .13) in the most selected group.
In the parent sample, selection on occupational level had little effect on the correlation between IQ and Control. It reduced from .05 to .01 and then rose again to .03 at the highest level of selection. There was a more marked effect of selection on the correlation between IQ and Achievement-Striving. With increasing degrees of selection, it first became steadily attenuated to zero with and then became negative. Although there was no significant association between IQ and Achievement-Striving in the full sample, at the highest level of selection there was a statistically significant negative association (r = .13, p = .03)."
#conscientousness #intelligence #personality