 Okay so it's 10.35 so we are on 9.35 sorry I think we need to start now so welcome to this what is time diary analysis of work my name is Pierre Walterie and I will take you through this hopefully interesting journey so we have a lot to talk about and I'm going to take you through basically six things today so I will say a few words about who we are as the UK data service then I will talk a little bit about what we mean by work in the context of this type of time diary based analysis short history of time use so that we understand a little bit what this type of approach is about and then so the main meals so to speak of the main course of this presentation I will talk about time diary instruments to measure work estimates that are derived from them and a couple of research based examples so about the UK data service who are we so we are for those who don't know us the main repository for UK secondary social science data we also provide support training and guidance to users and all of this is freely accessible funded by the Economic and Social Research Council our main users historically are academic researchers and students but increasingly also now government analyst charities in the voluntary sector business consultants as well as independent research research center and all think tanks we curate various types of data our core business historically were large-scale cross-sectional UK government surveys such as for example the lead before survey but also we curate major UK longitudinal surveys where individuals are followed over time large multinational database for example from the OECD as well as international survey data as well we provide guideways and access to census data both current and history records business and some business and administrative data as well as data one of surveys or qualitative multimedia data that ESRC funded researchers have a duty to live with us and last but not least we also provide support and training to users via our help desk so any one of you using our data can ask this question on our head desk via webinars and online workshop which are either data sets based methods based or software focused we have an increasing collection of online learning materials such as our data skills modules and forthcoming data skills pathways and also we help users not only with traditional survey data but also with new forms of data social media data for example via our computational social science team okay so that was about us but now let's start with the topic of today's proper so what is work it may seem like a pedantic question but I think it's important that we actually know or make explicit what we are talking about work is a potentially huge topic work can be described as joy effort meaning conflict resources usage or via the earnings and wealth production its deep rise but obviously as you can imagine this is not we are what we're going to look at we are not looking at work from the point of view of a total social fact to use sociological jargon but we are looking at work rather and maybe more humbly through those aspects that can be measured and that can be measured in a specific way that is as the time we dedicate to it together with some contextual information now moving to the way work is formally defined I will present maybe two common definitions here so the first one is the one that is brought forward by the international labour organisation and why does it matter because it's behind the conventions that are used for most social and economic statistics used by national statistics surface so work is defined in terms of whose it is for so for own use or used by others and also divided in terms of whether it leads to producing services or goods so which leads to the these various categories you can see here on the plot our news production work employment for pay or profit unpaid training work etc etc and volunteer work another definition that is also interesting and you will see while later instead of going through this complex typology we can simply define work as anything that you may was someone else third party to you to do sorry on your behalf without losing the direct utility that you rise from it so from that perspective cooking a meal or looking at the children is work but watching a film is not because obviously the utility you derive from watching a film cannot be transferred and importantly this is work irrespective of whether someone the third party is actually paid for it okay so now how has work or how have time use been developing over time so why I'm mentioning this because basically time use research to some extent is indiscretinguishable from a willingness to understand better household or workers productivity so the prehistory of time use research or if we look at the prehistory of time use research we will see that it starts in the early 20th century where various authorities or intellectuals were interested in understanding a little bit how a peasant household in the case of russia women in poor household in the case of the fabians in london were either working or producing goods this there's been a more example with soviet economists looking at time budget how russian workers were spending that time obviously frederick taylor's scientific organization of work in the narrow sense of line assembly workers u.s. departments for agriculture studies on the time use of farm and town workers the uk mass observations all of these studies were trying to understand how through how people spend their time but also how household were producing goods and services closer to our interest here are batches studies that were sorry that started being promoted after the second world war so the founding father of modern time use studies is a hungarian sociologist who's called sander alexander salai and who in the 1960s set out to create a comparative study of urban household and all the time use of urban household in 12 countries so it was quite a feat as you can imagine in the middle of the cold war gathering research team on both sides of the iron curtain and um yes collecting survey data from the so and it's it is to be credited for the first use of time diary as it's still currently being used so time diary defined as uh uh an instrument which gathers who does what where with whom uh over 24 hour periods a second uh panel year of uh time use research is uh Jonathan Gershune at the center for time use research here in the uk set up the multinational time use study and which is still one of the main source of harmonized time use uh data after uh and slightly later uh international nomenclature of time diary data uh were uh created so the icatus uh at the united nation or the harmonized european time use study uh at the eu via euro stat so that's really a brief uh history of the field of time use research but now how do time use research uh or how do time use time there is instrument uh measure uh things and measure work so time there is basically our survey data in overwhelmingly survey data so these are surveys in which on the one hand uh as any other survey you would have uh individual questions such as the whole while you what is your job etc etc but in addition to this normal person uh level survey uh you have a time diary so a time diary is a special type of questionnaire in which people are asked to tell or to write down at various intervals but the most common one is 10 minutes time slots uh what they were doing i'll come back to this in a second so it's usually collected on two days per person basis one week day and what day at the weekend and so in the perspective of time diary the unit of observation is not the person anymore but the day so we have a sample of days so to give an example the 2015 uk time use survey collected us more than 16 000 such diary days uh from 10 000 respondents in a 4 000 household and collected data uh time diaries for all uh people aged eight at the both in uh these households so now the instrument of time diary per se uh the most common time diary instruments record activities over 24 hours uh and uh the is made of a core set of four variables so the first one is what are people what am i if i'm the rest of them doing uh or was i doing if i feel it uh after the fact uh mainly uh what was i also doing because most people often do some form of multitasking where was i like when i was doing these things and uh who else if any uh was also present when i was carrying these activities so these are the four sets of uh variables that are collected in time diary data in addition and more recently uh time diaries data have been recording some measure of uh immediate well-being enjoyment or uh the extent to which it devices were also used while carrying out activities so that's what uh a pen and paper traditional uh diary could look like so this is the instructions that were uh shared with users participants of the uk uh time diary uh study of 2015 and so you can see uh each line sorry each row of the table represents a time slot a 10 minutes time slot and then people uh filling the way uh the days the activity of their days in each one of the columns main secondary activity uh with a device was used where where they were uh and uh whether other people were present okay let's go back to work so to speak in addition to this standard time diary in which all sorts of activity but of course including work are uh recorded there's a paid work specific uh instrument that's called the work schedule and that has been collected in uh a number unfortunately not all uh main uh time use surveys so it's it's uh interesting because it collects paid work uh for a full week so it's not just two days per week uh it has a slightly closer resolution of 15 minutes uh instead of 10 but it's basically a very convenient way of mapping uh someone's paid working life through uh out of the week so that's what it looks like uh it looks like basically a series of flying that uh people rows on which people can draw a line that represents the amount of time that is spent uh on paid work so a full week could work like this and we have data uh for the UK in 2015 for example that's a really convenient uh instrument especially for researchers who are interested in looking at uh work rhythm uh atypical work schedule at etc okay so we've talked about sample we've talked about time diaries but there's also the issue or we need also to be aware of what people feel in their time diaries or how what people feel in their time diaries is actually uh coded uh and historically this has been done via standardized nomenclatures of activity uh there are several of them so uh the oldest one is probably the multinational time you study uh one so created by uh Gershuni I mentioned earlier but it gave rise or to other nomenclatures such as the HITUS one for the EU or ICATUS for the UN so these nomenclatures are basically standardized list of activities and among which uh paid work for those of you who are interested in uh different ways of conceptualizing paid work and then provide an example in a moment uh it's interesting to look also at maybe large scale or uh studies time you study national studies such as the Indian time you study which has its own nomenclature or the American time you study so this is uh a snapshot uh and of course you can imagine I can't provide a full description of nomenclature on slides but it's just a couple of snapshots from these uh nomenclatures so this particular one comes from the Indian time you survey and the Indian time you survey is interesting the activities are classified according to which uh part of the economy they contribute so your paid work will be classified according to whether it's for agriculture and what kind of contribution it makes to agriculture and you can see that given the the current importance of agriculture for Indian household there's still a level of detail that you wouldn't find anywhere else so at the other end uh that's uh the recommendation and the detailed description of work according to the harmonized European time user studies so the guideline your EU based guidelines you can see its work is defined uh not as much uh as through its contribution to the economy but as employment so it's basically work as we call it in European time you study is the time you spend uh on doing the activities as an employee and as main and secondary job the multinational time you study uh also provides uh its own uh categorization of paid work created activities so it doesn't go as far in detail as uh heaters but the logic remains the same so uh are considered paid work um oh i considered uh work primarily uh so paid work is main job at home second job or what the whole or not at home um travel is part of work breaks or other time at workplace um and of course these are these are not uh the only form of work that are recorded i'm just focusing on this here for convenience but of course other type of unpaid work are recorded such as housekeeping you can see cleaning or caring for other people i can see that some questions are being asked in the q and a but i will answer these when uh at the end of the presentation okay so maybe one last thing before we move on to uh some estimates of time diary data uh the data structure so i'm not going to go too much into detail but typically what do time diary uh files or data set look like the answer is unfortunately there's not one common or universal way of uh coding uh time uh diaries but a very common one is uh files in long format so each line of search data set records uh an episode which is a unique combination of activity location uh and co-presence and uh so these episodes embedded within uh persons uh within persons and days uh it may sound a little bit abstract so that's what it looks like uh in practice so you can see this is um some uh time diary uh of uh from a Spanish survey so the first um 19 uh lines uh in the table represent a day uh as recorded by one person and from line uh number 20 onwards we have the second uh so a second day from another person within the same household um and you can see we have um so typical typical survey identifications a household number person number uh and but here the time uh diary specific elements so we have episode number so ranging from 1 to 19 all of the different episodes recorded by the person their duration and what they were about so we are on a Sunday so there's not much uh paid work here there's some unpaid work um but so that's just to provide a an idea of what it looks like okay but now it's time to move on to uh maybe uh paid work uh estimates with time diaries so i am going to uh take you through how typically a researcher or researcher would estimate uh basic quantities of uh work uh with time diary data and maybe the most common of such quantities is duration since we have uh all these episodes and the duration uh it's relatively straightforward to compute how much people uh how much time people spend doing stuff and uh in our case stuff is paid work uh per day so in terms of how to do it uh you people usually do it by flagging uh in the diary data set relevant uh episodes or work data set defined by the coding here so i'm using data from the multinational time use study so coding from the multinational time use study and i am not including commute uh as uh part of work so once these have been uh coded and here i provide some example uh with the our software package uh the next uh thing is to uh make sure that these work episodes actually record the duration of uh the episodes and if we were to uh have summary statistics for these uh episode we would see it's extremely low uh six minutes per on average as a mean and why is it low simply because we are not looking at what we should we are looking at the mean overall mean uh uh duration of work episode instead of the total uh amount of uh work uh duration of work per day in order to be able to account for the duration of work per day we need to sum uh so to do some of uh episodes work rated episode per day and it's this sum uh whose mean that we can uh then uh look at uh and uh compare across country if that's the sort of analysis we are interested in so we can see here uh our first daily estimate of uh time use uh work rated time uh duration it's still quite low it's about 126 minutes for spain and 151 minutes for the us it's really low so uh what's going on here why why is it so low um is it because we are looking at every day in in this in an indistinguished way so what would happen if and yes sorry i forgot there's uh this data can be uh easily uh plotted obviously uh but uh maybe it's because we are not looking at uh oh we are not differentiating uh week days and weekend so if i introduce an extra distinction between weekend and week days uh in the empty west weekends and week days are uh follow the u.s convention so the week start on sunday so that's the line of code i'm using here we can see indeed that uh people report clearly uh more time on paid work uh on week days but it still remained though it's still under 200 minutes per day so it's not uh doesn't sound like the the the typical time we would spend on um uh paid work in our normal day so this uh leads us to consider maybe a choice uh most time use uh researcher have to uh take on board which is are we uh interested in uh looking at an overall mean so also looking at uh in our computation at people who did not report work on um uh the data every day or are we only looking at people who reported paid work uh well this is also known as uh participants so on the one hand if we are interested in mapping the whole day of uh server respondent we may want to indeed take everyone on board and uh take participants and non-participants alike but on the other hand if we are only focusing on paid work uh for that matter and we want to reflect or to have a sense of a typical amount of time spent on paid work then we can uh restrict the sample we compute our estimate from to those participants and what happens if we do it then uh we get this type of result so I have retained here the distinction between weekdays and weekend and we can see that now the duration uh the daily duration on average for paid work are becoming more realistic they in most cases they uh stand between 400 and 500 minutes per day which is uh it's about seven hours is uh in correspond to what they expect and from that point on what you can of course uh the conductor or look at national differences okay so that's a first instrument a second instrument um which kind of uh derives from the first one so if we're able to compute duration and select people who were participant or not then we can also compute the probability of participation the probability of reporting paid work uh on a day so which is you know the word probability that our tag tagging variable will be greater than zero and again uh in a way that's more straightforward to compute than uh duration and we can see that on a typical day people in the US for example were more likely to report paid work on weekday at least than say people in Spain or even the UK and that remains true also for weekends a third type of a common instrument that is used by time use researcher exploratory fashion or maybe for general discussion of how people spend their time is time programs time programs are basically plots or maps uh in which the probability of or the probabilities of participation uh in activities are represented at each time point uh that uh that is uh recorded in a time diary so 10 minutes time slots is the the most common one and what do they look like they look like something like this so the x axis here represent the time of the day so by convention time use surveys start collecting diaries at four in the morning and that's uh three 59 the next day and this is weekday in the UK so what we can see here is the proportion of people reporting paid work here in full-time education which is in brownish green color and its rises start rising after 7 a.m peaks at around 11 there's a short slump here for people who have a traditional lunch break and are not eating in front of their computer and then goes back up in the afternoon and go back down again afterwards from 4 30 p.m and really start declining later in the evening and concomitantly sorry other type of activities so reproductive work and shopping is being carried out by people who are not doing a proportion of people who are not doing paid work there so that's a an interesting map of activities and including paid and unpaid work and also uh what is interesting is comparing such maps between people or groups of people or even countries so this uh graph now is the same kind of data but for a sample of french people and of course you can immediately see that in France or people take a lunch break more seriously here and here there's still a better sense of clearer this section between work and non-work for clearer than it is in the UK okay so so far I've presented really basic estimates of work or paid work using time use data now I'm going to look at things that are probably closer to real work research and yes that reflects maybe the type of analysis we would do as researcher either by looking at differences in time use by groups or gender day of the week for example or when we look at contextual aspect of work or when we build really our own research in for theory informed indicators of time use so in first example is something I worked on personally a couple of years ago and it's about looking at unsocial working hours so those of you in the field are familiar with the idea of all the question is to whether having unsocial or untypical work schedule can affect your well-being on your mental health and the nice thing about time diary data is that it's relatively straightforward to build typologies of typologies of paid work based on the time of the day at which work is reported so for example using the work schedule I've talked about before we can identify paid work that happens between say 8 a.m. until 6 p.m. and those that don't those that happen at night in the evening or at the weekend and then that helps us build an indicator of how of the amount of time people spend working on such hours and the proportion of such hours so just in practice it's amounts to flagging in a data set such episodes that are good lab label as social and unsocial some the amount of time they represent and compute means by group and gender if that's what you're interested in and you can also try and this is what I did here build indicators where the proportion of such hours relative to the total amount of paid work is analyzed so that's what it looks like so this is simply differentiated by gender the mean proportion of unsocial hours by sector of activity so we have sectors sorry I'm saying silly things it's not sector it's occupation so ranging from elementary occupation to all the way down or up to professional and managers and what is interesting is that both men and women here tend to have a larger proportion of their working time that is taking place on social hours if they are working in lower skilled occupations so that maybe contradicts a little bit the idea of overworked managers who are available for work at any time of their day anyway so that's an example of mapping unsocial hours according to gender and sector of activity another type of thing you can look at and I apologize for those interested in unpaid work given the limited amount of time for this presentation most of the examples I'm providing are about paid work time diary data also enables us to look at the context of work so we can look at whether people work in isolation people working with children in the room or location work from home but also commuting so commuting and more generally traveling is something that is coded in time diary usually is part of the location variable and we can similarly compute type of commuting by gender and in this case I chose to focus on regions of the UK so these are not percentages these are mean daily minutes spent commuting and Londoners won't be surprised to discover that they are those who spent probably the longest amount of time on their daily commute at the other end people in Yorkshire or the West Midlands having shorter commutes Northern Ireland this way and in another interesting difference and which may not surprise you if you already familiar with the literature in that area is that a man's commute in time tends to be longer than women and this has to do with the division of paid and unpaid work okay so I hope this provides a little bit of initial overview of what sort of analysis can be carried out with time diary data in the context of paid work I wanted to say a quick word about advanced visualization we have more and more fancy nice instruments now that can be used to visualize data so what I've used is really basic bar plots area graph etc but there are people who are developing more advanced tools so just to give a snapshot I won't be able unfortunately to go into the detail here you can see on the top left here it's a time program that I've similar to what I've shown but you can also map transitions so what do people do for example right after right before engaging in paid work and that allows you to map transitions and such a flower type of plot can show you how to do that there's a live animation here showing a little bit like a film of these transitions of yours so this is an animated graph even if I can't show you or maps geographical maps etc so I will share the slides later and those interested in accessing these resources will be able to find them and finally if you're interested in gaining more information about time diary data time diary research I would definitely advise you to look at the center of time use research at UCL and for data either the MTUS data set or series of data sets curated by CTUR or of course the UK data service for the UK based time diary data and I provide also link to ICATUS and HITUS if you want to look at these international nomenclature so I think we are running out of time thank you very much for your interest in the topic