This study tries to find a relation between leadership and psychopathy and leadership and hormones. Two groups of respondents were selected from the researcher’s network, leaders and non-leaders. Each subject was tested on leadership factors, psychopathy factors and testosterone and cortisol levels.
96 males were asked to participate in the study. The non-response was 12.5%, resulting in 85 participants. 52 of the participants are leaders (60.7%), with a mean age of 46.98 (SE = 0.951) and 33 of the participants are non-leaders (39.3%), with a mean age of 32.97 (SE = 1.846). The difference in age is significant (t(83) = -7.266, p < 0.01). Of the sample, 54.8% has a MBO degree or lower, the other part of 43.2% has a HBO degree or higher. Running a Chi-square test, shows that leaders are significantly higher educated, than the control group (2= 32.501, p < 0.001). The leaders are found in the network of Ricardo Westendorp and are selected on the size of the company, and the amount of followers. Only leaders in companies with more than 15 employees were selected. The mean years of leadership experience among the leaders is 17.9 years (SE = 1.10) and the mean amount of followers is 14.46 followers (SE = 2.09), the non leaders have no leadership experience and no followers in their corporate environment. Most of the leaders are also the CEO of the company. The non-leaders have randomly been selected and mostly include working individuals, with the exception of a few students. 84.8% of the respondents are married or living together with their partner, the other part of the sample is living by themselves (11.8%), alone with their kid(s) (2.4%) or with their parents (1.2%). The leader group members are significantly more often married or living together with their partner, compared to the control group (2= 20.382, p < 0.001), according to a Chi-square test.
The leaders have been asked to fill in a questionnaire with 128 items and collect saliva at 5 fixed time points at a day. The questionnaire is to be filled in online, using the questionnaire tool from Qualtrics.com. A total of 83 items is used in the analysis. The saliva tubes were to be returned by post. The respondents have been tested on a 83 items that measures both leadership performance and psychopathy and 5 Salivette tubes to measure testosterone and cortisol levels during the day.
A survey has been created, where 25 items measure leadership factors and 58 items measure psychopathy. Leadership is measured by a 7-point Likers scale, whereas psychopathy is measured by a 4-point Likert scale. An example of one of the leadership items is (translated from Dutch): “I often understand how others feel”. This item would have a 7-point Likert scale as an answering scale, varying from “I strongly disagree” (1) to “I strongly agree” (7). Leadership exists of 3 dimensions, including a total of 9 factors. An example of a psychopathy item is (translated from Dutch): “I have hurt people, to see them suffer”. This question is measured by a 4-point likert scale, varying from “Very true” (1) to “Very untrue” (4). Psychopathy consists of 3 factors.
Figure : Leadership model The three leadership dimensions are cognition, social cognition and motivation, as seen in the literature review. Cognition consists of 2 factors, (1) intelligence and (2) creative thinking. Social cognition consists of 3 factors, (3) agreeableness, (4) empathy and (5) charm. Motivation consists of 4 factors, (6) need for achievement, (7) need for affiliation, (8) taking charge and (9) risk taking. The intelligence factor is based on 2 items, which have a significant correlation (r = 0.859, p < 0.01). The creative thinking factor is based on 4 items, Cronbach’s alpha for this scale is 0.811, thus demonstrating good reliability. All alpha’s or correlations can be viewed in table 3. Running a exploratory factor analysis on the cognition cluster shows two factors, dividing the two intelligence items in one factor and the 4 creative thinking items in the other factor, with a minimum factor loading of 0.561. The factor analysis shows the relatedness of the items per factor and it shows that there is enough distance between the two factors, meaning that the factors measure something different. The agreeableness factor consists of 2 items, Cronbach’s alpha for this scale is 0.683, which is very close to the required value of 0.7. The agreeableness scale will still be used, even though the reliability of this scale is marginally under the required value, because factor analysis of the cluster shows that agreeableness is a separate factor. The empathy factor consists of 4 items, Cronbach’s alpha for this scale is 0.883, which indicates good reliability. The charm factor consists of 3 items, Cronbach’s alpha for this scale is 0.718. Running a factor analysis on the social cognition cluster results in three factors. Factor analysis on the social cognition dimension validates the use of the three factors of the social cognition dimension,. Three factors come forward, dividing the items of the three social cognition factors in separate factors, with a minimal factor loading of 0.321. Next, is the motivation dimension. Need for achievement is a 3 item scale, Cronbach’s alpha for this scale is 0.708, indicating good reliability. Taking charge is a 2 factor scale, with a significant correlation (r = 0.883, p < 0.01). Need for affiliation is also a 2 factor scale, with a significant correlation (r = 0.793, p < 0.01). The last factor of motivation, risk taking, is a 2 factor scale, with a significant correlation (r = 0.414, p < 0.01). Running a exploratory factor analysis on the motivation dimension shows 4 factors, which match the 4 motivation factors, with a minimal factor loading of 0.420. The psychopathy questionnaire consists of 3 factors. Boldness, disinhibition and meanness. The three factors represent the traits of psychopathy. Cronbach’s alpha of the boldness factor is 0.855, which indicates good reliability of the scale. Cronbach’s alpha for the disinhibition factor is 0.816, which demonstrates good reliability of the scale. Cronbach’s alpha of the meanness factor is 0.739.
Table 3: Cronbach's alpha or correlations for the leadership and psychopathy factors
Instruments: Saliva collection
The hormone data collection is done by using 5 Salivette tubes for every respondent. The respondents are asked to provide 5 saliva samples, taken on a single day. They have to chew on a swab for 1 minute, and to place it back in the saliva tube afterwards. The respondents were instructed to collect the saliva on fixed times. The first collection took place at the moment of waking, the second collection took place 30 minutes after waking, the third collection took place at 12 o’clock, the fourth at 16 o’clock and the last collection took place at 22 o’clock. The samples will be described in this thesis as the hormone, followed by the nthmeasurement. For example, the fourth cortisol measurement at 16 o’clock will be referred to as cortisol 4. The respondents were asked to register the exact times of saliva collection. The saliva samples were send back by post and analyzed in the laboratory of the Erasmus MC.
All statistical analyses will be done by using SPSS 17. Hypothesis 1 and Hypothesis 2 will be tested by comparing the means of leaders and the control group. The tests that will be used for this are the independent two-tailed t-test, independent one-tailed t-test, the two-tailed Mann-Whitney test and the one-tailed Mann-Whitney test for independent samples. The leadership dummy is set as a grouping variable and the 9 leadership factors and the 3 psychopathy factors as test variables. First, the 9 leadership factors and the 3 psychopathy factors have been tested on normality. Unfortunately, intelligence (D(84) = 0.167, p < 0.001), creative thinking (D(84) = 0.10, p < 0.05), agreeableness (D(84) = 0.13, p < 0.01), charm (D(84) = 0.10, p < 0.05), taking charge (D(84) = 0.156, p < 0.001), need for affiliation (D(84) = 0.168, p < 0.001), risk taking (D(84) = 0.12, p < 0.01) and disinhibition (D(84) = 0.12, p < 0.01) are not normally distributed, according to significant results on the Kolmogorov-Smirnoff test. The variables that are not normally distributed can’t be analyzed by a parametric t-test, because the assumption of normality can’t be made. A non-parametric test, the Mann-Whitney test, will be used to test differences in the means of these variables. An independent t-test will be used for the other normally distributed variables, empathy, need for achievement, boldness and meanness. Another assumption made to be able to perform parametrical tests is equality of variance. Running a Levene’s test on the variables results in several significant results, meaning that the variances of the leader group and the control group are not equal for all variables. The leader group and the control group do not have equal variances for the variables intelligence (F(1, 82) = 8.001, p < 0,01), taking charge (F(1, 82) = 5.058, p < 0.05) and disinhibition (F(1, 82) = 4.536, p < 0.05). All three variables are also not normally distributed. This means that the significant Levene’s test should be taken into account, when building conclusions on the Mann-Whitney tests of these three variables. Hypothesis 3 will be tested by using Spearmans’s correlation coefficients between the leadership factors and the psychopathy factors.
Hypothesis 4 is tested in several ways. First, the saliva data is transformed, so that it can be used. The saliva samples are tested on cortisol and testosterone levels. The data derived from the saliva samples will be used in several ways. Firstly, the individual hormone scores of all 5 collections of both testosterone and cortisol will be used, there is no transformation needed for this data. Secondly, the cortisol to testosterone ratios (C/T) of single time pointswill be used, which is calculated by formula 1, where Ci is the cortisol level at collection point i and Tiis the testosterone level at collection point i.
Thirdly, the area under the curve with respect to the ground level will be used, which is calculated by formula 2, 4 and 5. The AUCg will be calculated for cortisol, testosterone and the cortisol to testosterone ratio. The AUCg of cortisol is calculated by formula 2, with cmi denoting the measurement of cortisol at collection point i, ti denoting the time difference between cmi and cm(i+1) and n the amount of measurements. The calculation of ti is displayed in formula 3, with mi denoting hormone measurements. A standardized version of time difference is used, because the total time difference of the respondents is not equal, resulting in huge differences in the calculation of the area under the curve. Therefore, tidenotes a percentage of the time difference between the first and the last measurement. The AUCg of testosterone is calculated by formula 4, with tmi denoting the measurement of testosterone at collection point i, ti the time difference between tmi and tm(i+1). The AUCg of C/T is calculated by formula 4, where ctmiis measurement of the cortisol to testosterone ratio at collection point i and tiis the time difference between ctmi and ctm(i+1).
Fourthly, the area under the curve with reference to the first measurement (AUCi) will be calculated as a new variable. AUCigives us information about the increase or the decrease of the values. The AUCi is calculated by decreasing the value of AUCg by respectively cm1, tm1, or ctm1 multiplied by the total time difference (t1 + t2 ).
After transformation of the saliva data, statistical analysis can be done. The individual cortisol, testosterone and C/T values will be analyzed by a t-test, with the hormone scores as test variables and leadership as a grouping variable. This will be done by doing 15 t-tests, which will test Hypothesis 4. Hypothesis 5 will be tested by a one-way repeated measures ANOVA, which will check if the cortisol, testosterone and C/T values are significantly different over time within subjects and between groups. This basically tells us if the hormone levels at time point i are different from the hormone levels at time point i+1, i+2, etc. for the two groups, comparing the hormone curve lines on both line shape and line positions. Additionally a hierarchical regression analysis will be done, to check the influence of the hormone scores, leadership factors and psychopathy factors on the leadership dummy and leadership experience in years.