Sie sind auf Seite 1von 18

Operationalizing Max Webers probability

concept of class situation: the concept of


social class
1
Ken Smith
Abstract
In this essay I take seriously Max Webers astonishingly neglected claim that class
situation may be dened, not in categorial terms, but probabilistically. I then apply
this idea to another equally neglected claim made by Weber that the boundaries of
social classes may be determined by the degree of social mobility within such
classes. Taking these two ideas together I develop the idea of a non-categorial
boundary surface between classes and of a social class corridor made up of all
those people who are still to be found within the boundaries of the social class into
which they were born. I call social mobility within a social class intra-class social
mobility and social mobility between classes inter-class social mobility. I also
claim that this distinction resolves the dispute between those sociologists who
claim that late industrial societies are still highly class bound and those who think
that this is no longer the case. Both schools are right I think, but one is referring to
a high degree of intra-class social mobility and the other to an equally high degree
of inter-class mobility. Finally I claim that this essay provides sociology with only
one example among many other possible applications of how probability theory
might usefully be used to overcome boundary problems generally in sociology.
Keywords: Weber; social class; social mobility; probability; boundary surface; social
class corridor
A social class makes up the totality of those class situations within which
individual and generational mobility is easy and typical
(Max Weber [1920] 1978: 302; emphasis added).
The boundary problemdogs the concept of social class. Where todrawthe line
between one social class and another and howto justify the apparently arbitrary
choice of selecting only one or more of the various measures that are commonly
Smith (Faculty of Humanities and Social Sciences, Buckinghamshire Chilterns University College) (Corresponding about email:
ken.smith@bcuc.ac.uk)
London School of Economics and Political Science 2007 ISSN 0007-1315 print/1468-4446 online.
Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden,
MA 02148, USA on behalf of the LSE. DOI: 10.1111/j.1468-4446.2007.00140.x
The British Journal of Sociology 2007 Volume 58 Issue 1
employed in studies of social class: family background, education, occupation,
income, life-chances, relationtothe means of production, etc. Evenif all of these
conditions are selected, so that, apparently, no choice has been made (Smith
2002a: 3369), this decision would still need to be justied theoretically.
One possible solution to this problem might be to drop the idea altogether
that social classes have boundaries of any kind, as the Cambridge school
advocates (Stewart, Prandy and Blackburn 1980). Classes, we might well say,
are simply not the type of phenomena that has boundaries. But to do this ies
in the face of an intuitive concept of social class which, at least in sociology, has
traditionally been seen very much as a determinate phenomenon. The idea
then that even social classes might be indeterminate, or in some other way
undenable, opens up the very real possibility of the abandonment, not just of
the concept of social class itself (Lee andTurner 1996), but perhaps even of the
study of sociology altogether. After all, if even social classes do not have
boundaries, then what does? And if sociologists cannot even say what are the
exact boundaries of a social class one of the most fundamental of all socio-
logical concepts then what is the point of sociology?
However one fairly obvious possibility seems to have been overlooked in
traditional discussions of this question. This is the idea that sociology might do
better, especially when it comes to the statistical analysis of social scientic
data, to follow the lead of the natural sciences and try to develop a probability
concept of social class. Boundary problems are common in the natural sciences
but these can usually be overcome by making use of probability theory. For
example, since 1926, when Erwin Schrodinger (18871961) devised his famous
equation,
2
rather than trying to dene the precise location of an electron at any
particular point in time, modern physics has accepted that it can only know the
probable location of any particular electron at a given point in space. As Peter
Atkins says on this point:
To understand the results of Schrodingers calculation we have to know that
the solutions of his equation predict the probability that the electron will be
found at each point of space, not, as in classical physics, the precise location
of the electron at any instant [in time]. (Atkins 2003: 146; emphasis in
original)
What is more, Schrodingers equation is understood to provide physicists not
only with a probability reading, but also with what is known in physics as a
boundary surface for this reading; in the case of hydrogen, that point which
captures about 90 per cent of the probability of nding the electron at any
given point in time (see Figure I).
If the position of the electron in a hydrogen atom can only be determined
probabilistically, then why not the position of human individuals in social
classes too? And, if even physics cannot solve the problem of nding the exact
location of an electron at any one point in time (and, of course, with a 90 per
88 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
cent boundary there must be a 10 per cent chance at any one time that the
electron is not within the predicted area), why then should sociology be
required to say exactly where particular human individuals stand in the class
structure at any given point in time, especially when one considers that, with
the relatively high degree of social mobility in late industrial society, the social
space that many individuals occupy in the class structure is likely to change a
number of times during their lifetime? And, of course, this objection to class
boundaries also has much wider implications for the social sciences generally,
since this problem is generic to all social boundaries (Smith 2002b: 4059).
Perhaps then sociology too might do better to develop some kind of probabi-
listic concept of social class?
But if all this talk of probability sounds a little bit remote from sociology
(why, after all, should sociology follow the lead of the natural sciences, even
in the area of statistical analysis?) I would argue that this is not the case since
there already exists within sociology a classical discussion of just such a prob-
ability concept of social class, and this in the work of none other than one of
the founders of modern sociology, Max Weber. In the later and most intriguing
of his two famous discussions of the concept of class in Economy and Society
(the discussion which, somewhat perversely, appears rst in the by now stan-
dard University of California Press edition of this work), Weber denes class
situation probabilistically in the following all too familiar way:
Class situation means the typical probability of
1. procuring goods
2. gaining a position in life and
3. nding inner satisfaction
FIGURE I. The probability of nding an electron in a hydrogen atom as expressed by the
density of shading, the exponential decrease in probability with distance from the nucleus, and
the boundary surface that captures about 90 per cent of this probability
Source: Atkins 2003: 147; Fig. 5.5.
Operationalizing Max Webers probability concept of class situation 89
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
a probability which derives from the relative control over goods and skills
and their income-producing uses within a given economic order. (Weber
[1920] 1978: 302; emphasis added)
And he then goes on to further dene class (as opposed to class situation)
3
as
follows:
Class means all persons in the same class situation.
a) A property class is primarily determined by property differences,
b) A commercial class by the marketability of goods and services
c) A social class makes up the totality of those class situations within
which individual and generational mobility is easy and typical.
(Weber 1978 [1920]: 302; emphasis in original)
This probabilistic denition of class situation, and Webers subsequent identi-
cation of a property class, a commercial class and, most intriguingly of all, a
social class, may be contrasted with his earlier far more categorical, and much
less sophisticated (no mention of probability here
4
) denition of class and class
situation as follows:
We may speak of a class when (1) a number of people have in common a
specic causal component of their life chances, insofar as (2) this component
is represented exclusively by economic interests in the possession of goods
and opportunities for income, and (3) is represented under the conditions of
the commodity or labor markets. This is class situation. (Weber 1978: 927;
emphasis added)
We would seem to have here then, within Webers sociology, a classical (albeit
relatively underdeveloped) statement of the possibility of establishing a
probabilistic denition of class situation and, intriguingly, a concept of social
class to be distinguished from something called a property class and a
commercial class based on individual and intergenerational social mobility
that is strikingly different from Webers earlier far more categorical denition
of class in terms of market situations (i.e., what one has to sell in the market
place). Such a probabilistic concept of social class would also be very different
indeed from the now conventional (Goldthorpe 1987; Runciman 1990; Roberts
2001) way of dening class situation generally which attempt to draw a hard
and fast boundary in fact, a categorial distinction between one social class
and another, based it seems, as John Scott says, on little more than the
informed professional judgement of the researcher (Scott 1996a: 131) and
irrespective of however arbitrary the resulting boundary line might seem to be.
A probabilistic concept of social class would therefore seem to offer us a
clear alternative a third way as it were between the conventional view
which seeks to dene classes in terms of hard and fast boundaries and the
90 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
alternative view which denies that classes have any such boundaries at all. A
probability concept of class would still try to establish boundaries, but these
would be of a rather more fuzzy and indeterminate kind. It would seek to say
something about the probability that is to say, the chances, the odds, the
possibility that a representative individual might be expected to be found
within a given social space in this case a social class at any one point in time,
and what the boundary surface of this particular social class might be. For
example, since Webers concept of social class is dened in terms of social
mobility, we might well ask ourselves what the chances are that someone born
into a given class situation (the daughter of a stone mason for example) would
stay within their social class of origin or would be likely to be mobile out of it.
Such a concept of social class would not attempt to say anything about the class
location of any particular human individual, but only what the odds are at any
one time that such a person would stay in or leave a given class situation.
Operationalizing Webers concept of social class
My rst reason then for advocating a probability concept of social class is that
this might well provide us with a more sensible way of determining the bound-
aries of a social class than a more determinate concept of class has so far been
able to do, while my second reason is that such a concept has already been
suggested and by none other than Max Weber. These two observations then
lead on to my third point, which of course concerns the question of exactly how
contemporary sociology might go about developing such a probabilistic
concept of social class (or at least of class situation?) of the kind that Weber
has so intriguingly suggested. Obviously it will be necessary to say a little bit
more about probability theory itself; this seems unavoidable. But, apart from
this, I want to make a fairly startling claim here. Slight though Webers discus-
sion of this topic is, I want to suggest that everything we need to know to
develop such a probability concept of class situation is already contained in
Webers later and more sophisticated denition of the concept of class situa-
tion (Weber 1978 [1920]: 302) and, in particular, in the truly astonishing de-
nition that he gives here of the concept of a social class.
In dening the concept of social class Weber makes a very unusual and, as far
as I am aware,
5
little noticed claim. Within contemporary sociology, social
mobility is normally understood to take place between one social class and
another and, in fact, one would not normally say that social mobility had
occurred at all unless someone had left the boundaries of one social class to
join another. Where it occurs, social mobility is therefore normally thought of
as breaking through the boundaries of social classes dened in this hard and
fast (categorial) way. However, as we have seen, this is not the way that Weber
denes the concept of a social class in his later discussion of class situation. On
Operationalizing Max Webers probability concept of class situation 91
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
the contrary, a social class, he says, is made up of the totality of those class
situations within which individual and generational mobility is easy and
typical (Weber 1978 [1920]: 302; emphasis added). The key word in this de-
nition is within. For Weber social mobility not only takes place between social
classes, but also within a social class as well, and he actually seems to suggest
that the degree of mobility within a social class denes the boundaries of this
concept. This boundary is said to be up to, but not beyond, that point where
mobility is easy and typical.
6
According to Weber then, a social class is not a
class at all unless mobility takes place within its borders and, crucially, this type
of social mobility does not therefore undermine the existence of social classes,
but rather denes what these classes are.
7
Providing only then that we can
plausibly say what Weber might have meant by the rather vague and imprecise
expression easy and typical and this I would suggest is where an elementary
understanding of probability theory comes into play Webers highly unusual
denition of social class would seem to provide us with an alternative basis for
redening the boundaries of an innovative new probabilistic concept of social
class. Where individual and intergenerational mobility from one class situation
to another is easy and typical let us say for the moment, where individuals
commonly move from one occupation to another, either within their own life
course and/or compared to their parents then we have a social class.
However, where mobility is typically observed to be uncommon between class
situations, a university lecturer and a painter and decorator for example, then
these class situations do not form a social class. Some painters and decorators
may become university lecturers and some university lecturers painters and
decorators, but these are exceptions to the rule and do not alter the fact that
such mobility is very uncommon indeed. Therefore people occupying these
very different class situations quite clearly do not constitute a social class.
What then does the phrase easy and typical mean and how are we ever
supposed to understand what Weber might have meant by this term? The
answer to this question would undoubtedly have been a major problem for us
but for the context in which Weber uses this phrase: as part of his denition of
social class, which is itself part of the wider denition of class and class
situation. Since Webers denition of social class occurs in the context of his
discussion of class situation, and since this concept is dened probabilistically,
I suggest that the answer to this question is in fact relatively straight-forward;
easy and typical quite simply means probable. Something may reasonably be
said to be easy and typical when it is probable. Our question then becomes
what does probable mean? In statistics probability is a measure of the like-
lihood that one particular event or outcome will occur relative to all other
possible events or outcomes (McCall 1994: 441). The probability of something
occurring might well be as low as only one in one hundred, but this is still its
probability. However, something is usually only said to be probable once it is
more likely to occur than not, and something may clearly only be said to be
92 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
more likely to occur than not if it occurs just slightly more than 50 per cent of
the time. The boundary surface of a probability concept of social class then, the
point which we must remain within if something is said to be more likely to
occur than not, may therefore reasonably be said to be just over 50 per cent.
8
This denition may well seem arbitrary. Why, after all, should we draw the
line between one class and another at 50 rather than 40 or 30 per cent?
However, as we have already seen, a boundary surface is not the same thing at
all (and, in fact, is not nearly so categorical) as a boundary line, since it does not
claim to incorporate all of the people who may in fact still be within their
original class situation. A boundary surface also does not establish the lower
limit of class formation social mobility might reasonably be said to be easy
and typical where only 20 or 30 per cent of people are socially mobile but
only the upper limit; the point beyond which (mobility from the original class
situation now having become so common) a social class may no longer be said
to exist. Finally, the whole point of probability theory in statistics is precisely to
allow us to establish decision rules when, due to a lack of any other informa-
tion, such decisions would otherwise be difcult to make (tossing a coin would
be an example of just such a decision making process). If therefore just over 50
per cent of people who were themselves at one time in a particular class
situation are found to be in other class situations during their own life course
(for example if, as we might well expect, a majority of the sons and daughters
of plumbers are not themselves plumbers but occupy other class situations)
then up to and including 50 per cent of the individuals who occupy these class
situations may reasonably be said to constitute a social class so dened. While,
conversely, where there are class situations into which some people who are
the sons and daughters of plumbers are socially mobile but, crucially, where
mobility to these positions is less than 50 per cent of the total of all individuals
from the original class situation, then these class situations do not form a social
class with the initial class location. Crucially then the boundary surface of such
a concept of social class, and the social mobility with which we are concerned
here, is that within which 50.1 per cent of the people from such an initial class
location still nd themselves and the other 49.9 per cent of people who at one
time occupied such a class situation but now no longer do so, are no longer
members of such a social class.
Social class and social mobility
If then we now have a reasonable idea of what Weber might have meant by a
concept of social class, the boundary of which is said to be given by those class
situations within which social mobility is easy and typical, the problem we
now have to face is one of how to actually apply such a concept. What in fact
are the odds what are the chances, what is the probability of somebody who
Operationalizing Max Webers probability concept of class situation 93
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
is born into any given class situation moving from that situation into another,
and between which class situations exactly are the chances of this happening
better than even (better than 50:50)? By looking at any good recent study of
social mobility it should now be a relatively easy matter for us to say exactly
what the odds are of a given person who is born into any given class situation
staying within a particular social class and where the boundary surface of such
social classes actually are (i.e., at what point just slightly more than 50 per cent
of all such people have been accounted for). This is therefore what I will try to
do now.
The study of social mobility that I have chosen to look at in detail is that by
Mike Savage and Muriel Egerton (1997) which is itself based on a subset of the
National Child Development Study, a panel survey of 17,414 children born
between 3rd and the 9th March 1958 (Savage and Egerton 1997: 647). This
study is not without its problems from our present point of view. In particular
(see Table I below) it does not tell us anything about the precise class situation
of the two sets of 3,149 fathers and their sons, or the 3,181 fathers and their
daughters who are represented in the study but already aggregates these
multiple class situations into something very much very like Goldthorpes
seven class scale.
9
However, despite this, I have decided to use this study for
two reasons. Firstly because it does provide us with a representative sample of
the social mobility of the general population of England and Wales between
19581991 and we know exactly how many people took part in the study (and
therefore the total number of all the required outcomes which it was possible
for them to make). Since probability is a measure of required outcomes com-
pared to all other possible outcomes, this should then allow us to say some-
thing about the exact chances of social mobility by the age of 33 of a cohort of
children who were born in 1958 compared to their fathers occupational class
at the time of their birth. And secondly because it does still allow me to
illustrate one of the main things that I am trying to emphasize here; namely, the
contrast between a Goldthorpe type formulation of social class (see Tables I
and II below) and my own boundary surface or, as I will also call it, social
class corridor concept of class (see Table III). Therefore, by using Savage and
Egertons data, I will be able to bring out the contrast more clearly between a
conventional account of social mobility which sticks closely to Goldthorpes
model and, as I would argue, imposes arbitrary boundaries on social classes,
and the more empirical approach that I am advocating here which seeks to
discover exactly where the boundaries of such social classes actually lie at the
50 per cent level of social class formation.
From our present point of view the rst thing we can note about Savage and
Egertons data (see Table I for the social mobility of sons in 1991 compared to
their fathers occupational class in 1958 and Table II for the social mobility of
daughters) is that there is not even one single occupational group not the
owners and managers of large private rms, professionals, intermediate, skilled
94 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
or unskilled manual workers etc. in which 50 per cent reproduction of the
occupational group is achieved. As far as sons are concerned, the most highly
reproduced occupational status, the group most likely to pass on their status to
their sons, is that of skilled working men at only 34 per cent (see Table I,
column 6, row eight; note: columns run vertically, like the columns of a
classical building, while rows are read horizontal). The chances, the odds, of
such sons staying within the same broad occupational class situation as their
father is therefore 34:66, which reduces to one in three or one for and two
against. While for daughters, the most highly reproduced occupational status is
that of the daughters of intermediate fathers, 41.7 per cent of whom are
themselves in intermediate occupations (see Table II, column 5, row 6), a
probability of 41:59 [two to three], or two chances in favour compares to three
against. In short, none of these large occupational aggregates amounts to a
social class distinction, or even anything like this, at the 50 per cent level of
social class formation. The boundary surface of the concept of social class
formation that we are employing here therefore supports the conventional
view in sociology that such large occupational groups, though undoubtedly
signicant social phenomena, do not constitute anything like what might
meaningfully be called a social class.
At rst sight, this very low level of general occupational reproduction would
seem to present an extremely distressing prospect for the sons, and particularly
for the daughters, of high status occupation fathers. For example, only 19 per
cent of the sons and less than 8 per cent of the daughters of professional
fathers were themselves in professional occupation by the age of 33 (although
this might just have been as a result of the long training that is increasingly
involved in achieving professional status), while only 34.7 per cent of the sons
and 10.5 per cent of the daughters with fathers in the large business category
TABLE I. Savage and Egerton: social mobility of sons (column percentages)
Sons group Fathers group
Large bus. Prof. Lower
service
Small bus. Inter n-m Skilled
manual
Semi &
unskilled
Total %
Large bus. 34.7 27.2 20.3 19.2 16.9 13.9 9.0 16.7
Professional 12.5 19.01 11.7 7.6 8.0 4.6 5.3 7.3
Lower service 16.5 32.9 20.8 12.8 14.7 9.7 7.5 12.3
All service 63.7 70.1 52.8 39.6 39.6 28.2 21.8 36.3
Small bus 8.5 7.6 7.8 19.5 7.6 9.9 7.9 10.8
Inter. n-m. 10.2 8.7 16.5 8.1 15.1 10.2 7.1 10.0
All intermed. 18.7 16.3 24.3 27.6 22.7 20.1 15.0 20.8
Skilled man. 9.7 9.8 15.6 24.2 28.9 34.0 36.5 28.1
Semi / unskill 8.0 3.8 7.4 8.6 8.9 17.7 26.7 14.8
All manual 17.7 13.6 23.0 32.8 37.8 51.7 63.2 42.9
N 176 184 231 579 225 1,222 532 3,149
Source: Savage and Egerton 1997: 654.
Operationalizing Max Webers probability concept of class situation 95
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
were in the same occupational sector as their fathers by the age of 33.
However, when we come to consider broader occupational groupings this
picture changes dramatically. Following the practice adopted by Goldthorpe
(1987), Savage and Egerton amalgamate each of Goldthorpes seven occupa-
tional categories into three much larger social classes (see rows 4, 7 and 10 in
Tables I and II shown in bold type). When this is done, a very different picture
emerges. We can see that while only 19 per cent of the sons of professional
fathers had themselves achieved professional occupational status by age 33, a
staggering 70.1 per cent (Table I, column 2, row 4) of them that is, seven out
of ten, or 1 in every 1.42 such sons had achieved Goldthorpes broad service
class status by 1991. Similarly, fully 63.7 per cent of the sons with fathers in
large businesses (Table I, column 1, row 4) and 52.8 per cent of the sons of
lower service class fathers (Table I, column 3, row 4) managed to stay within
the broad general occupational area of the service class, while fully 63.2 per
cent of the sons of semi and unskilled fathers (Table I, column 7, row 10)
remained within the broad occupational grouping of the manual working class.
What this means is that the actual (empirically determined at the 50 per cent
level of signicance), social class boundary of these ve large male occupa-
tional groupings is lower than i.e. falls within the boundary of Goldthorpes
purported social class boundary, while the actual social class boundary of all
the other male categories, and every single one of the female categories, lies
somewhere outside the Goldthorpe social class boundaries.
Taking sons and daughters separately then, not one of Goldthorpes social
class boundaries actually coincides with an empirically determined boundary of
social class at the 50 per cent level of statistical signicance (although the 52.8
per cent lower service fathers and all service sons and the 51.7 per cent skilled
manual fathers and all manual sons come very close to this). Goldthorpes
TABLE II. Savage and Egerton: social mobility of daughters (column percentages)
Daughters
group
Fathers group
Large bus. Prof. Lower
service
Small bus. Inter n-m Skilled
manual
Semi &
unskille
Total %
Large bus. 10.5 8.9 9.5 9.4 4.6 5.9 4.7 6.9
Professional 5.0 7.8 3.9 3.4 3.2 1.0 0.7 2.4
Lower service 24.9 39.1 26.8 23.9 21.3 18.6 13.9 21.1
All service 39.9 55.8 40.2 36.7 29.1 25.5 19.3 30.4
Small Bus 4.0 7.3 3.5 6.3 3.2 4.2 4.0 4.6
Inter n-m. 41.8 27.4 35.1 36.5 41.7 36.0 32.9 35.7
All intermed. 45.8 34.7 38.6 42.8 44.9 40.2 36.9 40.3
Skilled man. 1.0 0.6 3.5 1.2 0.5 3.7 5.6 3.0
Semi / unskill 13.4 8.9 17.7 19.4 25.5 30.6 38.3 26.3
All manual 14.4 9.5 21.2 20.6 26.0 34.4 43.9 29.3
N 201 179 231 587 216 1,213 554 3,181
Source: Savage and Egerton 1997: 655.
96 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
professional judgment then of where the boundaries of large occupational
social classes are to be found would seem to be somewhat mistaken, and this is
especially so in the case of women whose employment status is strikingly
different to that of their fathers.
Where then do the actual, empirically determined, boundaries of each of
these social classes lie? We know that for the sons of large business, profes-
sional, and lower service class fathers, and skilled manual and unskilled manual
fathers this lies within the boundaries of Goldthorpes service and manual
working classes, and that for all other groups, the sons with fathers in small
businesses, non-manual fathers, and for all of the daughters of every single
occupational status group, this boundary is outside or exceeds Goldthor-
pes estimation. But where exactly does this boundary lie at the 50 per cent
level of signicance? In other words, within what range of positions then must
we travel until at least 50 per cent of all the people in any given initial class
situation have been accounted for? Once we know this then we will have
discovered the probability boundary of these actual, empirically determined,
social classes.
Range is one of three major measures of variability in statistics, the other
two being variance and standard deviation. While variance and standard devia-
tion are normally measured from one of the three main measures of central
tendency in a distribution (the mode, the median and the mean), with range no
such statistical measure of centrality is required. Range is in fact simply the
difference between the highest and the lowest score or value in any
distribution. Starting from any given point in a distribution one simply looks to
see how far, plus or minus, the distribution is spread out from this initial point
of departure. In our case, this means looking to see how far we need to travel
until just over 50 per cent of all the people from a given class situation have
been accounted for. Now unfortunately for us, as far as the children with
fathers in large businesses are concerned, we do not have any data for people
who by 1991 have risen above their fathers occupational status since Savage
and Egerton, following Goldthorpe 1980, do not include an elite class above
the occupational class structure even though the existence of such a class has
apparently been recognized by Goldthorpe himself (Scott 1996b: 212).
10
However, if we say for the moment that class situation simply relates to
occupational status (and that therefore people who do not have an occupation
also do not have a class situation) and therefore only move down the class
hierarchy we can see that, since a total of 176 men (Table I, column 1, row 11)
and 201 women (Table II, column 1, row 11), and therefore 377 men and
women in all in Savage and Egertons sample, were born into large business
class situations in 1958 (see Table III below for the absolute numbers of such
people rather than the percentage these individuals represent as given by
Savage and Egerton). Taking sons and daughters together then (since we no
longer have any reason to disaggregate these two groups), we can see that we
Operationalizing Max Webers probability concept of class situation 97
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
need to account for at least 189 (equal to 50.1 per cent) of these 377 sons and
daughters until we can say we have reached the social class boundary within
which people born into this broad general social class situation are commonly
more likely than not to be found by age 33. When we do this (see Table III,
columns 1 and 2) we can see that the 50 per cent boundary very nearly equals
and in fact only exceeds by three individuals (192 persons or 50.9 per cent) the
boundary of Goldthorpes broader service class for this group since, as far as
the sons of such fathers are concerned, 63.7 of 176 such sons (Table I, column
1, row 4) gives a total of 112 men while 39.9 per cent of 201 such women
(Table II, col.1, row 4) gives a total of 80 women which, when added together
equal 192 sons and daughters with fathers in the large business category, who
stayed within the broad general service class sector by age 33. When therefore
both the sons and daughters of fathers in large businesses are taken together
as a single group Goldthorpes professional judgement (Scott 1996a: 1312)
of the boundary of this particular social class is therefore conrmed in this case
by our probability measure of the boundary surface of this class.
Similarly, when we come to consider the actual social class boundaries in
1991 of people who were born into the skilled manual working-class situations
in 1958 (the other occupational group which, for men at least, has a very high
degree of inter-generational social reproduction), we can see that since 1,222
men in Savage and Egertons sample (Table I, column 6, row 11; N = 1,222) and
Table III. Social class boundary surface of men and women born into seven different broad
general occupational class situations (absolute numbers)
Source: Modied version of Savage and Egerton 1996 Tables I and II; showing absolute gures
rather than percentages and actual rather than Goldthorpe class boundaries.
Note*: Actual G refers to the actual numbers still within the boundary of Goldthorpes Service,
Intermediate or Manual Classes in 1991.
98 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
1,213 women (Table II, column 6, row 11; N = 1,213) were born into this
general class situation, we have a total of 2,435 people in this broad general
class situation and we therefore need to account for the social mobility of 1,218
(or 50.02 per cent) of these individuals before we might reasonably say that we
have found the probable boundary surface of people born into this class
situation. What in fact do we then nd in this case? Of the men who were born
into this class situation 51.7 per cent, or 632, were still within the boundaries of
Goldthorpes manual class (Table I, column 6, row 10) in 1991, however, only
34.3 per cent (Table II, column 6, row 10), or 416 of all the women born into
this class situation were still in manual-class occupations at this time, giving us
a total of 1,048 men and women from this particular class situation who were
still within Goldthorpes manual working-class boundary by 1991. This total
therefore falls short by 170 people of our required outcome of 1,218 persons
still within this social class if Goldthorpes manual working-class boundary
was in fact empirically supported for people born in this general class situation.
In order then to nd the actual social class boundary at the 50 per cent
probability level for people born in this class situation we therefore have to
leave the realms of Goldthorpes, as we now see, too hard and fast manual
class category Goldthorpe clearly sets this too low and move up into the
so-called intermediate class, which we now see is therefore not a distinct
social class at all for people born into skilled manual occupational
backgrounds. Since we have a more or less equal number of sons and daugh-
ters of skilled working men in Savage and Egertons sample (1, 222 sons to
1,213 daughters or 50.185 per cent:49.815 per cent) it seems reasonable to take
an even number of our missing 170 sons and daughters of skilled working
men (therefore 85 each) from both Table I and Table II. If we then look at
Table I (column 6, row 6) rst we can see that a further 10.2 per cent of the
1,222 men who were born into manual-class situations (in other words, a
further 125 of these 1,222 men) were in so-called intermediate non-manual
occupations by the age of 33. We therefore have 40 more of these men
(125 - 85 = 40, or 32 per cent) in the so-called intermediate non-manual class
than we need to determine the actual social class boundary of men born in this
class situation. Similarly, if we look at Table II (column 6, row 6), we can see
that fully 36 per cent of the total 1,213 women born into manual-class situa-
tions (= 437 women), were by 1991 in so-called intermediate non-manual
occupations, and that this is therefore (437 - 85 =) 352 more women in this
group than we need to determine the actual social class boundaries of such
women in 1991. We can therefore say that the probable empirical boundary of
the social class into which more people than not who were born into a skilled
manual working-class situation in 1958 is actually about (85/1.25) = 68 per
cent, or two-thirds of the way up the so-called intermediate non-manual class
for men, but only about (85/4.37) = 19.45 per cent, or one-fth the way into the
so-called intermediate non-manual class for women. The actual social class
Operationalizing Max Webers probability concept of class situation 99
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
boundaries of people born into skilled manual-class situations therefore
extends downward to incorporate all the semi- and unskilled people in 1991
who were born into the skilled-class situation in 1958 and upwards to pen-
etrate about two thirds of the way into intermediate non-manual occupations
for men, but only about one fth, or 20 per cent, of the way in the (so-called)
intermediate class for women from this particular class situation. (Note, if
Savage and Egertons sample had included data on skilled manual workers
who had fallen out of the occupational class structure altogether and had
therefore entered the so-called underclass as I think it might well have done
then the upper limits of this particular social class would undoubtedly have
to be revised downwards).
It is then a relatively simple matter to calculate the boundary surface (at the
50 per cent level of signicance) of the social class of all the other occupational
groups (male and female combined) represented in Savage and Egertons data.
And, with the exception of the children with fathers in large businesses men-
tioned above, we nd that not one of these groups actually falls on the hard and
fast social class boundary (service, intermediate or manual) imposed by the
Goldthorpes class schema (see the nal row of Table III above for the actual
difference from Goldthorpes estimate). Two groups (the children of profes-
sional and semi and unskilled workers) are actually under the Goldthorpe line
(by which I mean that the social class boundary of these two groups the range
of positions within which people born into these two broad general class
situations typically exchange occupations with others is less than (lower than)
Goldthorpes line), while the remaining four groups (as with the children of
skilled manual workers) are all above, and sometimes well above, Goldthor-
pes line. What all of this means of course is that Goldthorpes theoretically
imposed class divide is actually (and really we should perhaps say, not too
surprisingly?) too rigid when it is subjected to an empirical test of this kind. In
particular, Goldthorpes so-called service class boundary badly overstates
(sets too far down the class hierarchy) the actual social class boundary of
people originally from professional class situations (which actually only
extends something like 122/40 = 35 per cent of the way down into the lower
service class) and, as with the skilled manual working class, badly understates
(sets too high up the class hierarchy) the actual lower social class boundaries
of people from small business and intermediate non-manual class situations
(which, for men at least, actually extends into the upper reaches of skilled
working-class situations).
Finally, in Table III above I have emphasized the concept of what I call a
social class corridor. What I mean by this term are all those people who may
still be said to be within the boundary of their social class of origin at the 50 per
cent level of social class formation. Where social class is dened in terms of
frequent mobility between a number of closely related class situations, such
people may be said to have moved on in their lives, from one class situation to
100 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
another, but not up. For these people social class of origin largely a matter of
family background really does still mean something. There is however a
second group which has broken through the 50 per cent boundary surface.
They are more mobile than the rst group and may be said to have escaped the
boundaries of their social class of origin. The rst group those within the
social class corridor may therefore be said to be a highly class bound/class
dependent group in so far as they still to be found in much the same class
situation as their parents, while for the second group those outside the
corridor social class of origin is now of little or no importance.
Conclusion
In this essay I have taken seriously Max Webers astonishingly neglected claim
that the boundaries of a social class can be dened empirically in terms of the
degree of individual and generational mobility within which such mobility may
reasonably be said to be easy and typical. Since Weber claimed that class
situation is to be dened probabilistically, I have set the upper boundary of
easy and typical social mobility at just over 50 per cent, since, statistically
speaking, something may be said to be more likely to occur than not when it
occurs just slightly more than 50 per cent of the time. I have then applied this
boundary surface to a typical discussion of social mobility in Britain today, that
by Savage and Egerton, in order to discover what I claim are the actual
empirical boundary surfaces of a number of broad general social classes
dened in terms of this degree of mobility. Not surprisingly this boundary
surface is found to be much less clear-cut than the three categorial distinctions
(Service, Intermediate and Manual) that the Goldthorpe scale suggests. In fact,
with the possible exception of the children of the owners and managers of
large businesses, none of Goldthorpes boundaries are supported, with both
professional and semi and unskilled classes being well within Goldthorpes
boundary (by which I mean that more than 50 per cent of all individuals from
these original class situations are already accounted for well before we meet
Goldthorpes boundaries), while all of the other four class situations (lower
service, small business, intermediate non-manual and skilled manual) are
found to break through the boundaries imposed by Goldthorpes scale. The
above discussion therefore applies an empirical measure to and in fact, as
John Scott says (Scott 1996b: 21314), a test of the adequacy of Goldthorpes
widely used but highly intuitive class scale and thereby makes a contribution to
the project rst described by Scott in 1994 that a promising future for class
analysis can be achieved only if we go back to very theoretical foundations of
the Weberian programme and apply a direct empirical measure to demo-
graphic process (Scott 1996a: 12731).
In writing this essay I also claim to have provided one possible illustration of
a much wider project rst alluded to by Goldthorpe and Marshall in their 1992
Operationalizing Max Webers probability concept of class situation 101
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
essay The Promising Future of Class Analysis (in Lee and Turner 1996), but
not taken seriously by them at that time. In this essay Goldthorpe and Marshall
are strikingly dismissive of the idea (which, in my opinion, they incorrectly
attribute to Breiger, 1981) that occupational mobility rates might actually be
said to determine the boundaries of a social class. They also claim to be unsure
whether any such idea could claim serious endorsement in Webers work and,
furthermore, they say that they think any such scheme must be excessively
empiricist and likely to lead to major interpretative problems (Lee andTurner
1996: 105). However I do not myself see why we should agree with any of these
objections. I have shown that Weber does endorse just such a project, while I
think that a probability concept of social class boundaries, although empirical,
must lead to fewer problems than a purely theoretical concept which is not
supported empirically in this way and where therefore any boundary estab-
lished cannot be tested and must seem to be arbitrary. Furthermore, I believe
that the concept of social class that I am advocating here one that allows for
a relatively high degree of social mobility from one class situation to another
but crucially without breaking through the boundaries of a social class dened
in terms of such mobility resolves the long standing dispute within sociology
between those, like Goldthorpe and Marshall, who think we still live in a highly
class bound society and those, like Peter Saunders (Saunders 1992: 7980) and
Ray Pahl (Lee and Turner 1996: 8997), who think this is no longer the case.
Both arguments are correct I think. What Goldthorpe and Marshall describe as
relatively high rates of social mobility would, I suggest, be better understood in
terms of what Weber means by the mobility that takes place within the bound-
aries of social classes, and what Goldthorpe and Marshall call low rates of
absolute social mobility would be better understood as social mobility that
takes place between social classes. There is indeed a very high degree of social
mobility within social classes in late industrial society, but crucially, as Gold-
thorpe and Marshall insist, such mobility does not undermine the boundaries
of Weberian concept of social class.
During the course of this essay I have made use of a number of new terms
which I have suggested might usefully be employed in any future discussion of
aWeberian concept of social class, in particular the idea of a boundarysurface
rather than a boundary line between one social class and another, and the idea
of a social class corridor that incorporates all those individuals who are still to
be found within their social class of origin. In view of the importance that
Weber attaches to mobility within a social class and the neglect that this
concept has received within contemporary sociology, I wish nally to suggest a
further new term to emphasize the importance of this key point. As we have
already seen, in dening his concept of a social class Weber explicitly refers to
something called individual and generational mobility. Nowadays these two
types of mobility are more usually called intra-generational and inter-
generational mobility (respectively, social mobility which takes place within an
102 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)
individuals own career and social mobility compared to ones parents). By
extension therefore it seems to me that we might well refer to that social
mobility which takes place within the boundaries of a social class as intra-class
social mobility and that which takes place between one social class and another
as inter-class social mobility and, unless the similarity between these two new
terms and the already existing distinction between intra and inter-generational
mobility is too confusing, in future sociologists might then try to analyse both
intra and inter-generational social mobility and intra and inter-class social
mobility.
(Date accepted: July 2006)
Notes
1. I would like to thank Professor John
Scott of Essex University for reading an
earlier very much longer version of this
essay for me and for all of the very many
helpful comments and suggestions that he
made which I have incorporated into this
essay.
2. For the record, Shrodingers formula is:
V: h
2
y / m = (V - E)y, where E is the
energy of a particle of mass m, y is the wave-
function we are trying to nd, and y is its
curvature (Atkins 2003: 215).
3. For Weber, as I understand this, class
situation is an individual attribute. It refers to
the position of any given individual human
being in society according (a) to the skills
they have to offer and (b) to the income this
generates. Class, on the other hand, is a more
collective appellation. It refers to all of those
individual people who may be said in some
theoretically meaningful way to be in the
same general class situation. Class then is a
totality, as Weber says, of any number of indi-
vidual class situations
4. Although it might be noted here that
even in this earlier denition Weber still
denes class in terms of life chances.
5. For a possible [partial?] exception to
this rule, see the essay by Ronald Breiger
(1981: 584) and especially Goldthorpe and
Marshalls comment on this (in Lee and
Turner 1996: 105). However, Goldthorpe
and Marshalls description of Breigers
project seems to me to take this point very
much further than anything Breiger either
intended himself or does in fact do.
6. According to John Scott, Webers
actual words are that mobility is leicht
moglich and typisch stattzunden pegt
(Scott 1996b: 247 fn.13)
7. All of this of course the nature of the
concept of social class is to say nothing
about the entirely separate concepts of
property or commercial class situations,
which however I think we might reasonably
expect to be more conventionally dened by
Weber; e.g., in terms of what one has to sell
types of property, labour, goods, or services
etc. in the market place.
8. Other denitions of this highly
ambiguous phrase are clearly possible. For
example, one might reasonably say that the
phrase easy and typical means frequent
or common, and something being a veg-
etarian, for example might easily be said
to be a common thing if only as few as 10
or 20 per cent of the population do this.
However, I want to suggest that this is not
the point. What we need to do here is not
to say when something is easy and typical,
but rather when it may no longer reason-
ably be said to be so. Therefore, although
something obviously is common when a
100 per cent of the population do this, or
even 90 or 80 or 70 or 60 per cent, the point
is that it may no longer reasonably be said
to be easy or typical once less than 50 per
cent of people do this.
Operationalizing Max Webers probability concept of class situation 103
British Journal of Sociology 58(1) London School of Economics and Political Science 2007
9 Savage and Egerton employ what they
themselves describe as a modied version of
Goldthorpes model (Savage and Egerton
1997: 6501).
10 This is also a problem for the children
of semi and unskilled parents, some of whom
might well have fallen out of, or below the
occupational class structure altogether; a
potential underclass in fact.
Bibliography
Atkins, Peter 2003 Gallileos Finger: The Ten
Great Ideas of Science, Oxford: Oxford Uni-
versity Press.
Breiger, Ronald 1981 The Social Class
Structure of Occupational Mobility, Ameri-
can Journal of Sociology 87(3): 578607.
Goldthorpe, John 1980 Social Mobility and
Class Structure, Oxford: Clarendon Press.
Goldthorpe, John 1987 Social Mobility and
Class Structure in Modern Britain, Oxford:
Clarendon Press.
Goldthorpe, John and Marshall, Gordon
1996 The Promising Future of Class Analy-
sis, in David J. Lee and Bryan S. Turner
1996 Conict About Class: Debating
Inequality in Late Industrialism, Harlow,
Essex: Pearson Education.
Lee, David J. and Turner, Bryan S. 1996
Conict About Class: Debating Inequality in
Late Industrialism, Harlow, Essex: Pearson
Education.
McCall, Robert 1994 Fundamental Statistics
for Behavioral Sciences, Fort Worth, USA:
Harcourt Brace College Publishers.
Pahl, Ray 1989 Is the Emperor Naked?
Some Questions on the Adequacy of Socio-
logical Theory, in David J. Lee and Bryan S.
Turner 1996 Conict About Class: Debating
Inequality in Late Industrialism, Harlow,
Essex: Pearson Education
Roberts, Ken 2001 Class in Modern Britain,
Basingstoke, Hampshire: Palgrave.
Runciman, W. G. 1990, How Many Classes
Are There in Contemporary British
Society?, Sociology 24(3): 37796.
Saunders, Peter 1992 Social Class and Strati-
cation, London and New York: Routledge.
Savage, Mike and Egerton, Muriel 1997
Social Mobility, Individual Ability, and the
Inheritance of Class Inequality, Sociology
31(4): 64572.
Scott, John 1996a Class Analysis: Back to
the Future, in David J. Lee and Bryan S.
Turner 1996 Conict About Class: Debating
Inequality in Late Industrialism, Harlow,
Essex: Pearson Education.
Scott, John 1996b Stratication and Power,
Cambridge: Polity Press.
Smith, Ken 2002a Mutually Contested Con-
cepts and Their Standard General Use,
Journal of Classical Sociology 2(3): 32943.
Smith, Ken 2002b Some Critical Comments
on the use of the Concept of Ethnicity , in
Modood et al. Ethnic Minorities in Britain,
Sociology 36(2): 399417.
Stewart, S., Prandy, K. and Blackburn, R. M.
1980 Social Stratication and Occupations,
London: Macmillan.
Weber, Max 1978 [1920] Economy and
Society, in two volumes, Berkeley and Los
Angeles, California: University of California
Press. [Previously published in 1920].
104 Ken Smith
London School of Economics and Political Science 2007 British Journal of Sociology 58(1)

Das könnte Ihnen auch gefallen