In her more recent work Pahl has further divided the categories by identifying female and male dominated 'patterns of control' ( Pahl , 1989 and 1990 ; see also Vogler , 1989) .sx Following Pahl's work the focus of most research in this area to date has been on access to , and the control and management of , financial resources .sx However , we would argue that a broader definition of the resources involved is needed to analyse satisfactorily the internal workings of the household .sx These include the domestic division of labour ; the allocation of time to household members ; the allocation of consumption goods and services ; and access to consumer durables which nominally belong to all household members .sx While this broader definition provides a more comprehensive account of the internal workings of the household it also poses considerable problems when constructing a short series of survey questions .sx Most questionnaire surveys have neither the space nor time to cover the full range of items which would be needed to cover satisfactorily all the areas mentioned .sx It is in establishing the key indicators from which we can infer how and why people allocate the resources available to them , and how this process changes over the life course in relation to factors which are both internal and external to the household , that qualitative information should prove useful .sx The qualitative data already collected has produced a wealth of information on the often subtle and complex negotiations between household members concerning the allocation of money and resources .sx As one woman describes :sx I allot myself approximately about 250 ( pounds per month ) for food , which might sound a lot for three of us but we've got two big dogs that eat a lot and the two men eat like they've got hollow legs and then he ( her husband ) has a set amount which is what we sort of discuss , I mean I don't pin him down , he said a set amount which I think is extremely generous and I don't know what he does with it because he never buys clothes , but anyway , he has this money and he doesn't drink and he only smokes about five cigarettes a day and I don't actually have a set amount of pocket money but if I want something and I don't often buy things , but I really want a new dress or something and there's money in the account I'll buy it .sx But I always say to him , look , you know I've bought a new dress , I hope you don't mind .sx .sx It would be foolhardy to suggest that this depth and complexity of interpersonal negotiations can be successfully replicated in the context of a small number of survey questions .sx However , this does not imply that asking about the distribution of resources within the household in the context of a survey is a useless exercise .sx Indeed there is a definite need for a large scale data set in this area of research .sx What it does highlight however is the need for qualitative and quantitative research to be mutually informative and provide the possibility for analyses in which both types of data contribute to and illuminate the analysis as a whole .sx While the majority of social researchers would probably agree with these aims , in practice they are often not as easy to reconcile , even though the methodological divisions are by no means as clear cut in the practice of social research as the epistemological debates which posit a necessary link between theoretical perspective and method of data collection assert .sx As Bryman ( 1988 ) argues there may not be a necessary or one to one correspondence between epistemology and methods as is often asserted in the methodological literature .sx We discuss this point further in the final section .sx Problems of classification .sx The qualitative project has highlighted several problems with existing models of intra-household allocations , some of which are common to all classifications of social phenomena designed to clarify the world we live in , while others relate specifically to the needs and objectives of the panel study .sx The first point relates to the general problems of categorisation and classification and the second to the specific issues raised by the longitudinal nature of the panel study .sx When empirical work based on an already existing taxonomy begins it becomes almost immediately apparent that there are considerable numbers of cases which do not fit neatly into any one category .sx The preliminary analysis of our interviews ( Ritchie and Thomas , 1990 :sx 3 ) indicates that of the nineteen couples interviewed individually only four ( three common pool , and one independent manager ) could certainly be assigned to one of the categories of the model outlined above .sx Another two ( one independent manager , one common pool ) could be assigned if the category was broadly interpreted .sx In the remaining thirteen cases there were either significant features of a category missing or there seemed no appropriate category at all .sx As Jan Pahl observes in relation to her own data :sx In reality the different types were not mutually exclusive , and for many couples it was a case of choosing the type of allocative system which most closely resembled the way in which they organised their money .sx As so often in the social sciences , classification involved simplification .sx It is therefore justifiable to be sceptical about the adequacy of the typology .sx ( 1989 :sx 79 ) .sx Thus , by simplifying in order to clarify , the subtleties of the interactions between household members are subsumed within general categories which have a tendency to become reified .sx Instead of proceeding via a process of analytic induction , by which categorisation as hypothesis is subject to falsification , the temptation is to either squeeze recalcitrant cases into pre-existing boxes or to find reasons for omitting or ignoring them altogether .sx Of course the tendency to view existing classifications as given is shared by respondents , and may be reinforced by the design of a survey question .sx When faced with a categorisation such as that employed by the SCELI project ( which used an elaborated version of Pahl's original model ) and asked :sx " People organise their household finances in different ways .sx Which of the methods on this card comes closest to the way that you organise yours ?sx It doesn't have to fit exactly - you should choose the nearest one " , very few respondents are inclined to say " none " ( apparently zero per cent in the SCELI survey , see Vogler , 1989 :sx 46 , Table 1) .sx But the difficulty is not automatically overcome by resorting to depth interviewing since it is likely that the researcher's eventual categorisation or description of responses will correspond to her/his existing conception of what such a categorisation should look like - and what it should look like is of course something like an already existing classification .sx This is the nature of any conceptual system - it simplifies and organises , but it constrains the possibilities of thinking beyond it .sx The second major problem , relating in particular to longitudinal analysis , is that a classification into types is inherently static in its form ; people may of course move between boxes , but the boxes themselves represent states rather than processes .sx While for the purposes of analysis it may be necessary to identify 'types' , just as it is necessary to identify events or outcomes , we must remember that we may be applying relatively arbitrary delineations upon what are essentially continua .sx Following the same line of argument , a movement from one box of a classification to another for an individual , or even an entire change of classificatory typology , does not constitute an explanation .sx Too often , in quantitative research particularly , the development of a new classification or the identification of movement between states is taken as the end of analysis .sx The development of a classification and the demonstration of its usefulness is frequently not perceived in terms of the necessity to provide a theoretical justification - the identification of ordered associations is regarded as sufficient .sx Beyond this solid ground lies the dangerous and unstable terrain of the explanation of human action - the realm of interpretavism sic !sx . Here the familiar language of science - of objectivity , of rationality - must be replaced by the shifting undercurrents of far less tangible discourses involving 'meanings' and 'motivations' .sx But it is this realm which must be entered if an attempt is to be made at understanding change .sx Any simple identification of movement between the categories of a classification is bound to fail at the level of explanation unless some understanding of the processes which underlie observed outcomes is arrived at , and these processes themselves can only be fully understood with reference to actor's accounts .sx Couples , families or households ?sx These general considerations lead us to two issues of direct relevance to the BHPS .sx Both relate to problems of definition .sx Firstly there is the question of how one sets the analytical boundaries of intra-household allocations .sx It is clear that households do not exist in isolation .sx Individuals within households have relationships with individuals in other households and transfer resources of time , money or care which may materially affect how those households arrange their internal affairs .sx For example we have found that it is common for women with young children who return to work part-time to rely on their mothers to provide free childcare .sx The availability of free childcare was seen by many women as an important factor influencing their decision to return to work .sx As Brannen and Moss have observed , women engage in a " mental accounting procedure of whether 'it's worth my while working' " ( 1987 :sx 86 ) where the cost of childcare is seen as a direct deduction from their own earnings .sx As one woman told us :sx " It's funny you know , because I pay the child-minding fees and I think that I pay it because I feel that it's , the child or children are really my responsibility .sx If I go out to work then it should come out of my salary to pay someone else to look after them .sx I wouldn't dream of asking him to pay the childminding fee .sx " .sx The fact that women are able to go out to earn is , however , of considerable benefit to all household members , particularly in view of the gendered nature of spending patterns in which women's earnings are more likely to be spent on items of communal rather than purely personal benefit even where women's earnings are irregular or part-time ( see Morris , 1989) .sx Qualitative data is able to highlight the extent of inter-household transfers and wider support networks or relationships which may be vital to understanding the constraints and possibilities open to particular individuals in a given household .sx While these remain theoretically distinct from purely intra-household allocations it is clear that any analysis of the internal dynamics of the household cannot be conducted without reference to inter-household relationships .sx Secondly , the qualitative research has made us question whether it is valid to extend a model developed to categorise financial arrangements between married or cohabiting couples to the distribution of a variety of resources within households which may or may not be made up of married couples or nuclear families .sx We have already noted the importance of recognising that households exist within the framework of broader social relationships .sx In addition however , it is clear that we need to conceptualize the household not only in terms of the conjugal or 'standard' nuclear family household but to take a variety of different household compositions with varying cultural backgrounds into account .sx An approach which focuses exclusively on the way in which couples allocate resources between them , while being a useful exercise in its own right , loses important dimensions which influence the internal workings of the household .sx For example where teenage or young adult children are present their involvement or contribution to the running of the household , both in terms of money and labour within the home , is relevant .sx It is possible that the informal exchange of resources in kind such as free child-care and housework may materially affect a household's ability to survive financially .sx Equally important is the fact that households may be composed of unrelated individuals or contain an extended family of two or more generations , while ethnic minority households may operate on very different assumptions to the 'Western European' model .sx The distinction between the household and the family is one which is continually blurred in much of the literature in this area .sx