Vilfredo Pareto's Sociology : a Framework for Political Psychology

Vilfredo Pareto’s Sociology
166
5.4 Scale 
Analysis
Before comparing the three subpopulations (Labour, Conservative and Liberal 
Democrat MPs) on each of the ten personality measures, checks were made to ensure 
adequate internal consistency validity and reliability. Part-whole correlations were 
examined for each of the three subpopulations in turn. Crucially, for an item to pass 
muster it had to contribute to its associated scale with a reasonably high part-whole 
correlation for all three subpopulations. This provided a rigorous three way split 
reliability standard for each scale; furthermore, it ensured all three subpopulations 
could be compared on all ten personality measures. 
Scales measuring conservatism-liberalism, conviction-relativism, caution-risk, 
dissociative experiences and social anxiety all fared well when subjected to this 
test. Items contributing to the locus of control scale also held up well across the 
three subpopulations. Most part-whole correlations were in the r=
±.5-.7 range, with 
the poorest performing item, ‘trusting to fate has never turned out as well for you 
as making a decision to take a definite course of action’ still managing a moderate 
part-whole correlation (r=-.35, p=.023) for the Conservative subpopulation. Items 
contributing to the political aloofness scale did almost as well, with most part-whole 
correlations retaining high values across subpopulations. The single exception to this 
rule was ‘politicians who don’t take risks have little to contribute’. This item correlated 
with its set score at just r=.32 (p=.145) for the Liberal Democrat subpopulation. This 
failure may reflect skewdness arising from a tendency for Liberal Democrats to score 
particularly highly (18 cases out of 24 either ‘agreed’ or ‘strongly agreed’). Hence 
it was decided to retain the item. The convention-innovation scale gave more cause 
for concern, because the ‘children should be taught to respect and obey authority’ 
item failed badly for both Conservative and Liberal Democrat subpopulations.
7
 This 
item was therefore jettisoned. The resulting balanced four item scale was found to 
perform acceptably for all three subpopulations. The individualism-collectivism 
scale’s performance was acceptable overall, although one item was problematic. The 
item ‘You would probably agree with the proverb “too many cooks spoil the broth”’ 
performed poorly for the Conservative subpopulation owing to an unusual degree of 
skewdness (35 out of 42 Conservatives either agreed or strongly agreed). This item’s 
part-whole correlation for the Liberal Democrats was also low, although this time 
the failure appeared less attributable to skewdness. Hence it was judged necessary to 
be wary of the resulting individualism-collectivism variable when analysing Liberal 
Democrat responses. The poorest performing scale was that tapping aggression. 
Part-whole correlations were low for all three subpopulations. This seemed due to 
skewdness towards low scores. 125 out of 150 MPs either ‘disagreed’ or ‘strongly 
disagreed’ with ‘you almost never dare express anger to people for fear you may 
lose their love or approval’. 107 out of 153 cases either ‘disagreed’ or ‘strongly 
of 2. Back benchers were given weightings of 1. This variable was only calculated for Labour 
and Conservative MPs. 
7  As one MP pointed out on his questionnaire, this item’s meaning was contestable 
because ‘respecting’ and ‘obeying’ authority need not coincide. 

---

Testing Pareto’s Theory
167
disagreed’ with ‘there are two kind of people in this world: the weak and the strong’. 
Hence caution was necessary with this variable.
5.5 Population 
Diversity
The population contained good diversity. The 153 respondents included 81 Labour 
MPs, 42 Conservative MPs and 24 Liberal Democrat MPs. No other political parties 
were represented in numbers sufficient for statistical analysis, although two Ulster 
Unionists, one Welsh Nationalist and one Scottish Nationalist did contribute. 13 MPs 
indicated they were younger than 35. 97 were between the ages of 35 and 55. 43 MPs 
were aged over 55. 48 were raised in lower income bracket households, whilst 74 
were from middle income brackets and 23 were from upper income brackets. 
Only 28 MPs (18%) were female. Of these, 21 identified themselves as Labour 
(26% of the Labour subpopulation), 4 identified themselves as Conservative (9% of 
the Conservative subpopulation) and only 1 identified herself as Liberal Democrat 
(4% of the Liberal Democrat subpopulation). This meant gender differences could 
only be studied within a Labour Party context; specifically that cohort of female 
Labour MPs popularly known as ‘Blair’s Babes’. 
71 MPs (46% of the whole population) had entered Parliament for the first time 
within the 1997 intake. The remainder, bar one missing case, had been there for 
between 3 and 39 years. Of these, 35 MPs (around 23 %) had been in Parliament for 
15 years or more. Of the Conservative and Labour MPs who replied giving some 
indication of the levels of political seniority which they had managed to attain, 62 
(just under 50% of the whole population) had not risen beyond the back benches. 23 
MPs (19%) had achieved a rank of P.P.S. or equivalent. 21 MPs (17%) had achieved 
the rank of Junior Minister or equivalent. A further 16 MPs had risen to full cabinet 
(or shadow cabinet) level.
5.6  Comparing the Three Parliamentary Parties
Mean scores were calculated on both demographic and personality measures for 
each of the three political parties. Rather than express differences between party 
subpopulations using these scores, it was decided that a better indication of effect 
size would be conveyed by expressing differences as proportions of mean standard 
deviations which were calculated for the whole population. Intervals between the 
parties on all variables are set out in table 5.6(a) below. Subpopulations are ordered 
on each variable according to ascending mean scores. Intervals between low and 
middle positions, and between middle and high positions, are also given for each 
variable. Statistically significant intervals are asterisked. Ns are all in the region of 
80 (for Labour), 40 (for the Conservatives) and 20 (for the Liberal Democrats).
The patterns highlighted by table 5.6(a) reveal the extent to which differences 
between the political parties correspond to the multi-trait continuum which comprises 
Pareto’s model. The richest of these patterns, which will provide our starting point, 
certainly echoes much of Pareto’s distinction between the lion and the fox. This pattern 
consists of a similar ordering of the parties on seven of the ten personality variables. 

---