YT YouTube2Text

YouTube2Text → Transcripts → Dissertation projects: Introduction to secondary analysis for qualitative and quantitative data

Dissertation projects: Introduction to secondary analysis for qualitative and quantitative data — Transcript

by UK DATA SERVICE · 6,925 words · 1,172 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00hello welcome to this webinar on reusing
0:03data for dissertation projects my name
0:05is maureen haker
0:07and i've worked with the uk data service
0:09for about eight years doing everything
0:11from digitizing data to reuse projects
0:15and i'm here with ellie bloom who is a
0:17research associate at university of
0:19manchester
0:22all right so what we're going to do
0:23today
0:24is we're going to go over what secondary
0:27analysis is
0:28look a bit closer at some of those key
0:30methodological issues of of reuse
0:32projects for qualitative and for
0:35quantitative data
0:36but before we get knee deep in that i
0:38want to give a very brief overview of
0:40the uk data service for those of you who
0:42have not had the pleasure of exploring
0:44the archive yet
0:46i've tried to pull in some case studies
0:48throughout so you can see some of our
0:49data in action
0:51and i'll
0:52and we'll end by signposting you to some
0:54further resources
0:57which can help if you're planning a
0:58reuse project
1:00this is an introductory workshop so
1:02we've assumed most of you will probably
1:05have not used archives before
1:07um
1:09and may have only had some introductory
1:11modules on research methods more
1:12generally
1:16and before we get too far along
1:18i just want to address a simple point
1:20first
1:21what is secondary analysis
1:24in short secondary analysis is a method
1:26which asks new questions of old data
1:30it's analyzing data that you've not
1:32collected yourself
1:33usually researchers collect far more
1:36data than they actually need to answer
1:37their own research questions so think of
1:40national surveys which collect a lot of
1:42data on a representative sample
1:44or qualitative studies which will
1:46contain interviews that could last for
1:48one or two hours or sometimes more
1:51these data sets
1:53have you know lots of information can
1:56answer a lot of different questions and
1:58can be analyzed using different
2:00techniques so secondary analysis makes
2:02use of that and does just that it reuses
2:05data that's been collected by someone
2:07else
2:09but there is a complicated nuance here
2:13you may have heard some other terms used
2:16to describe this method including
2:18secondary data analysis or reuse
2:21projects and they all refer to the exact
2:24same method so don't get too confused by
2:26this but there is an ongoing debate
2:28about how to call this method so in 2007
2:322007
2:34libby bishop
2:37wrote about the primary secondary
2:39dualism and she basically was making the
2:41argument that there may be a privileging
2:44of methods where you go out and collect
2:45the data yourself
2:47and using terms like secondary analysis
2:50or secondary data reinforces that
2:52hierarchy
2:54but actually primary and secondary
2:55analysis are a lot more similar than
2:57different
2:59if you fully consider the key
3:00methodological issues of secondary
3:02analysis as a method
3:05consequently you'll find an increase of
3:07the
3:08use of the term data reuse
3:11and that's a term that we'll we'll be
3:12using throughout the workshop too and
3:14that takes into account the huge range
3:17of ways that data can be used and reused
3:19and it doesn't imply that any of those
3:21projects are secondary to the initial
3:23use of data
3:25you can use whatever term comes to mind
3:27first but just do be aware that there's
3:29a few different ways of describing this
3:31process of reusing data
3:36so now you know what secondary analysis
3:38is
3:39um
3:40but where would you find data that's
3:42already been collected
3:44this is where the uk data service which
3:46holds the largest collection of social
3:48science data in the uk comes in
3:51we are a comprehensive resource that is
3:54funded by the esrc
3:57and our main job is to be a single point
4:00of access to a wide range of secondary
4:03social science data
4:05the main purpose then is really the
4:06collection the ingest and the processing
4:09of data
4:10and then further dissemination of that
4:12data for people to use
4:14in addition to that data infrastructure
4:16core we also have a service layer which
4:19provides extensive support training and
4:21guidance
4:23who is it for
4:25um we like to think that it's for anyone
4:27who has an interest in data
4:29traditionally our main audience
4:32and the people who probably both deposit
4:34and use our data the most tend to be
4:36academic researchers and students like
4:38yourselves
4:40a lot of other groups are well
4:41represented as well um this could
4:44include government analysts charities
4:46foundations
4:47businesses research centers think tanks
4:50all give us and use our data
4:53given the importance of data and how
4:56it's used and how it's disseminated we
4:58try to reach out and support a wide
5:00range of communities
5:02what kind of data do we hold i know some
5:05of this language on this slide might
5:07might be a bit
5:08might be a bit a bit specific
5:11um but it's just to say that you know
5:14the majority of our data at least
5:16judging by the number of collections is
5:18quantitative data so we hold over eight
5:20thousand collections of which six
5:22thousand six hundred of them are
5:24quantitative collections and that
5:26includes the survey micro data
5:29um
5:30aggregate statistics uh international
5:33macro data census data micro data and we
5:36do also have
5:38a a
5:40fairly large collection of qualitative
5:42and mixed methods data as well
5:46where does it come from
5:48uh that varies depending on the data
5:50type and some of the sources that you
5:53see here
5:54which will include agencies and
5:56statistical time series those are our
5:59main sources of quantitative data most
6:01of our qualitative and mixed method data
6:04comes through individual academics so
6:06these might actually be some of your
6:07tutors
6:08when they get a research grant they may
6:10well have deposited the data at the end
6:12of that grant with us
6:15and we of course hold some originally
6:17paper-based public records and
6:18historical sources
6:21including things like the census
6:24where can you find information
6:27about it well we've got a website uk
6:29dataservice.ac.uk
6:31and that holds a lot of information and
6:33from there you can find our data catalog
6:36and we have hundreds of pages which also
6:38discuss methodological issues like
6:41gaining consent anonymizing data storing
6:44data
6:45there's also some student specific
6:47tutorials like our data skills modules
6:50we've got some exercises based on
6:52collections that we hold
6:54and there's a help page if you have
6:56specific questions about how to use our
6:59websites
7:00okay but getting back to dissertation
7:03projects
7:05what kind of research projects can you
7:07do from reusing data in an archive well
7:10we really need to go back to the
7:11beginning to answer this and think about
7:13the research
7:14process and you're hopefully familiar
7:17with this model
7:18it starts out with some kind of
7:20you know
7:21topic or general direction for research
7:24you then do some background research
7:26into the literature that's already been
7:28written on your topic from there
7:30hopefully you're inspired to ask a
7:33question which builds on that body of
7:35research and once you've got that
7:37research question you then decide the
7:39best way to answer it and design your
7:41project
7:42once you've settled on your method you
7:44then collect the data you analyze it and
7:47you begin your write-up which is what
7:49you would submit for your dissertation
7:52now you might have
7:53a few extra steps or you might swap
7:55around a couple of those steps depending
7:57on your theoretical foundation but
8:00generally speaking
8:01this is how we normally think about the
8:03research process
8:05when you're doing a project using
8:07secondary analysis however this process
8:09looks different
8:11this model that you're looking at
8:12clearly shows how the research question
8:14is built from your chosen topic area
8:16your per your preliminary search
8:19possibly the literature you find
8:21however with a secondary analysis
8:24project the research question is derived
8:27directly from the data
8:29so you start with a topic you're
8:31interested in but instead of looking for
8:33literature
8:34you look for data and start evaluating
8:36collections when you find a collection
8:39that intrigues you you then ask a
8:41question a research question of that
8:44data and from there you would then find
8:47what literature exists on that question
8:50you then would not need to collect data
8:53you just need to access it which is of
8:55course one of the key advantages of
8:57secondary analysis
8:59the data is already collected you just
9:01need to get your hands on it either by
9:03downloading it from a catalog page or
9:05you may need to actually go into the
9:07archive if it's only available in paper
9:09form
9:11once you have it you then analyze it and
9:14write up your dissertation
9:16so the key point i'm making here around
9:19reusing data for dissertation projects
9:22is about where your research question
9:23comes from
9:25would take a lot of time and inside
9:27knowledge about data within archives to
9:29be able to just come up with a research
9:32question
9:33and then search for the perfect data to
9:34answer it you might be searching for the
9:37right data for a very long time unless
9:40as i said before you actually already
9:42have a good existing knowledge about the
9:44collections
9:46for a dissertation project when your
9:48time and resources are limited you'll
9:50want to look at the data first
9:53and then from there develop a research
9:55question which gives a new take on that
9:57data
9:58you can of course do it the other way
10:00around and spend some time looking for
10:02the right data
10:03but for a dissertation project you may
10:05want to actually look and see what data
10:08exists first in your general topic area
10:11before actually nailing down what your
10:13research question is
10:17with that being said it's probably
10:19important to have some kind of idea
10:21about the kind of project you want to do
10:25while you might be exploring data
10:27without a specific question in mind
10:30you may want to give some thought as to
10:31the kind of research design your project
10:33will follow
10:35so there's four different types of reuse
10:37projects that can lend themselves well
10:39to dissertation projects
10:41and these are re-analysis a replication
10:44study a comparative study or a re-study
10:49re-analysis is probably the one that
10:51comes to mind when thinking about what
10:54secondary analysis is
10:56and this involves thinking about the
10:59wide range of approaches you can take in
11:01the analysis of a data set
11:03it usually means asking some kind of
11:05slightly different research question
11:07from what the original researchers were
11:09trying to do so for example
11:12clive seal and charters black did a
11:15study using comparative keyword analysis
11:18of illness narratives
11:20the original illness narratives had been
11:22looked at exclusively for health
11:24research so these were interviews that
11:26were meant to explore how diagnoses were
11:28made
11:29when seal and charter is black came
11:31along to do the comparative keyword
11:33analysis they were much more interested
11:36in the analysis of the discussions
11:38between patients and doctors rather than
11:40the actual health issues that came up in
11:42the interviews
11:44so the question can be very different in
11:46that kind of way
11:48or sometimes
11:49the question can be on a similar topic
11:51to the original research but have a
11:53slightly different focus so for example
11:56joanna bourneau was looking at
11:58gerontology as a topic and she found two
12:01different data sets which looked
12:02specifically at gerontology
12:05however bornet's research question was
12:07on racism
12:09which wasn't the focus of the original
12:11work but the data set was rich enough to
12:14allow her to explore that theme within
12:16the existing data
12:19if you want
12:20um
12:22to use the exact same analysis strategy
12:25that would be a replication study
12:28right now there's a real concern about
12:30reproducibility of research and
12:32replication studies can reveal the
12:34messiness that's involved in working
12:36through data
12:37so one of the most famous or
12:40perhaps really infamous
12:43examples of replication is from thomas
12:45herndon who was a postgraduate student
12:47at university of massachusetts
12:50and he was assigned an assessment to
12:52replicate results from a published study
12:55it's a fairly common exercise if you're
12:57studying quantitative methods
12:59so he chose reinhardt rogoff's 2010
13:02paper growth in the time of debt
13:05basically
13:07this paper came up with the proportion
13:09at which your national debt can be
13:12of your gdp before you see negative
13:15economic growth
13:17so thomas herndon pulled the oecd data
13:20so that it was using this sort of
13:22international data so he just pulled the
13:24same data set
13:26to re-run their analysis as the paper
13:28had laid out but he got a completely
13:30different answer
13:32the paper
13:33said that debt cannot exceed 90 of gdp
13:37however he calculated that the debt can
13:40actually exceed your gdp and even then
13:43you only see a minimum impact on growth
13:48contacting the original investigators
13:50getting the actual data set that they
13:53had worked from and not just the
13:54downloaded data set he found that there
13:56was actually a flaw where they had
13:58miscopied some cells
14:02the full story is published in 2013 in
14:04the new yorker if you're interested in
14:05reading some of those details
14:08a replication study hopefully wouldn't
14:10always find flaws in this kind of way as
14:13the original study but nonetheless this
14:15is a study design that's worth
14:17considering and can help you develop an
14:19appreciation for the research process
14:21and you can even develop a project
14:23whereby you rerun a series of studies on
14:25the same topic or perhaps you explore a
14:28complicated data set with missing data
14:30transforming variables and so on
14:34you can also do comparative work so you
14:36might be looking at an international
14:38comparison between two countries or
14:41comparing social subgroups of the
14:42population based on shared social
14:45characteristics
14:48our key data page for quantitative data
14:50sets outlines some of the large national
14:52surveys that are held at the archive any
14:55of which would allow you to do some
14:56comparative work without having to
14:58actually collect two sets of data
15:00so you could compare samples across time
15:03geographic place gender ethnicity
15:07these characteristics are usually
15:08collected as standard for these larger
15:10surveys
15:13and the final type of reuse
15:16i'm going to exemplify with a case study
15:19that i'm going to walk through with you
15:22and this this is a re-study and this is
15:24where you replicate the methods of a
15:27study for purposes of comparison so it
15:29does a bit of secondary analysis but it
15:32also allows you scope
15:34um
15:36to actually go out and collect some data
15:38too
15:39so this kind of reuse project is from
15:41the collection school leaver study
15:44the original study was conducted by ray
15:47paul in the late 70s as part of a much
15:50wider community study on the isle of
15:53and as part of that project uh paul
15:56asked
15:57teachers to set a particular type of
15:59essay just before students were due to
16:02leave school prompting them to imagine
16:04they were reaching the end of their life
16:06and something made them think back to
16:08the time they left school
16:09they were then assigned to write a short
16:11essay of what happened in their life
16:13over the next 30 to 40 years
16:16in 2009 graham crowe and don lyon
16:20and that is graham crow on the left
16:22there with ray paul on the right
16:25decided to re-analyze that data set
16:27focusing solely on student aspirations
16:31and using the very same methodology they
16:33conducted a restudy of school leavers
16:35for students on the isle of sheppy in
16:382009
16:40and the prompt supply to students
16:43in 2009 was nearly the same
16:46imagine you're at the end of your life
16:48and reflect back on what you've done
16:49since leaving school
16:51they then transcribe those essays and
16:53compare the themes from that set of
16:55essays to the set of essays that were
16:57collected by ray paul in the late 70s
17:00and you can see the wording of the
17:01prompt and a small snippet of one essay
17:03here
17:04and the findings are fascinating and
17:07they show the difference of young
17:08people's aspirations after one
17:10generation 40 years of time has passed
17:15but how exactly were they different
17:18in 1978 students expected much more
17:21grounded and arguably mundane sorts of
17:24jobs
17:25career progression was gradual and it
17:27followed on from hard work
17:29and sometimes there were talks of
17:31periods of unemployment or even death
17:33and you can see a few examples in the
17:35left column of some of the quotations
17:37from those essays such as the one on the
17:39bottom
17:40i longed for something exciting and
17:42challenging but yet again i had to
17:44settle for second best i began working
17:46in a large close factory
17:502009 2010 however showed students
17:54imagining well-paid and instantaneous
17:56jobs filled with choice but also a lot
17:59of uncertainty
18:00crowe and his research team also noted a
18:03clear influence of celebrity culture in
18:05those essays so for example you have the
18:07quote on the bottom of a girl who writes
18:10in my future i want to become either a
18:12dance teacher a hairdresser
18:14or a professional show jumper horse
18:16rider if i do become a dancer my dream
18:19would be to dance for beyonce or someone
18:21really famous
18:25this study is a larger one than what
18:27might be realistic for a dissertation
18:29project but the goal was to engage the
18:31community alongside the research and
18:33find
18:34innovative ways of including
18:36participants in the research outputs
18:38and as part of this initiative they
18:40publish the living and working on cheppy
18:42website
18:43which helps to create a shared history
18:45and memory of what living on the isle of
18:47chevy means among the community
18:51so this is quite an ambitious project
18:53for a dissertation but it's a good
18:55example of how you can combine a bit of
18:58data collection and a bit of data reuse
19:00into one project
19:02and certainly for those who might be
19:04doing phd work in future this is
19:06something that you might actually want
19:07to consider for one of your projects and
19:10for others you can design a more
19:12feasible study with a smaller sample
19:14and smaller outputs
19:18so hopefully you are now budding with
19:20ideas on what you might want to look for
19:23in the archives or what kind of projects
19:25that you might want to do reusing data
19:28and since you're not collecting data
19:30yourself you'll find that reuse projects
19:32have very few ethical considerations
19:34comparatively
19:36and hopefully you wouldn't hit too many
19:38snags with ethical review boards
19:40however
19:41this doesn't mean that there aren't any
19:43ethical considerations so there's two
19:46key points that i want to make before
19:48diving into qualitative
19:50and quantitative data so
19:52first of these starts at the point of
19:54access
19:56how do you get permission to use the
19:57data
19:58so if you're reusing data that's in an
20:01established archive like the uk data
20:03service we've taken a lot of pain out of
20:06access by negotiating licensing issues
20:08with the person who collected the data
20:10so this means that you have to sign
20:13what's called our end user license and
20:15this is a really exciting legal document
20:18which you can see on the left there and
20:20it lists just these terms of conditions
20:23things that
20:24you need to do or that you need to
20:26consider when you're reusing the data
20:28and there's two really important things
20:30that i want to stress here
20:32the first of these is that there's no
20:34onward sharing of data and that includes
20:37with your supervisor so if you need help
20:40with your analysis and your supervisor
20:42needs to see the data then he or she
20:44will need to register and download the
20:46data for themselves the end user license
20:49stipulates that you cannot under any
20:51circumstance share your data or your
20:54login details with anyone
20:58all of the data that we have has been
21:00anonymized
21:02which is again likely to be the the case
21:04if you're reusing data from an archive
21:07however just because it's anonymized
21:09doesn't mean that it's completely
21:10impossible to figure out identities of
21:12participants
21:14mark elliott has some really excellent
21:16youtube tutorials which go through
21:18anonymization theory but in short he
21:21makes the argument that no anonymization
21:23strategy will be 100 effective at all
21:26times
21:27consequently
21:28should you inadvertently uncover
21:31identities of any of the participants
21:34then this end user license says that you
21:37will not reveal that identity to anyone
21:41so those are the key issues to recognize
21:43when signing the end user license
21:47so the second point i want to make once
21:49you have sorted out the access
21:52you then need to ensure you cite the
21:53data
21:54so in short citing archive data helps
21:57data creators track the impact of their
22:00study it also supports reproducibility
22:02and it makes it easier to find the data
22:04that you used for your project
22:06and the issue is so important that the
22:08uk data service has a page on our
22:10website which goes into a little bit
22:12more detail about data citation and will
22:15help you explain why this is an
22:17important ethical issue
22:20with the uk data service
22:22we make this easy again by supplying you
22:25with the citation that you need on our
22:27catalog pages
22:29this citation you'll see underneath the
22:31box citation and copyright
22:34on our catalog page and then
22:36you just need to select what kind of
22:38citation format you want and you can
22:40just copy and paste that citation into
22:43your reference list
22:46so you've got access you've sorted out
22:49citation now comes actually doing the
22:52secondary analysis so first i'll walk
22:54you through qualitative data and a
22:56couple of key issues with getting
22:58started with qual data and then i'll
23:00pass over to ali who's going to talk
23:02about quantitative data
23:04so first i'll talk about reorienting
23:06yourself to the collection
23:09recontextualizing that data
23:11and finally just a couple of points
23:13about sampling
23:15so when you first download a data set
23:18you'll get a zipped folder which looks a
23:19bit like this
23:21we've got some folders that are stuffed
23:22with files
23:24and most qualitative data is held in the
23:27folder called rtf rtf is it is a type of
23:30word document
23:32so to get to the data you just need to
23:33go into the rtf folder
23:37and this folder when you open it will
23:39look like this
23:42there's all your data nicely organized
23:45and clicking on one of those files would
23:47open up something that looks maybe
23:49something like this
23:52and this is this is a snippet of the
23:54school leavers essays
23:55um you know what it would look like more
23:58in its uh entirety and that collection
24:01contains over 100 of those files but the
24:04files don't have to just be essays or
24:06interview transcripts
24:08you might have for example
24:10pdfs of handwritten notes like in the
24:13upper left hand corner or you might have
24:15ethnographic notes like those on the
24:17right or some collections might also
24:19have images or videos like those in the
24:21lower left hand corner
24:24more than likely
24:26you'll end up opening up
24:28interview transcripts like the one
24:30that's seen here you should have clear
24:33turn taking uh in speaker tag so you you
24:36know who's talking
24:38um
24:39and there's you know just a lot of
24:41different types of data available so
24:42have a good look through that collection
24:44first and see what's actually in there
24:48the next thing you need to do is to
24:50orient yourself to the project and i
24:52think the main point to make here
24:54is to not underestimate the amount of
24:56time that it would take to get
24:57acquainted with the data set there might
24:59be multiple levels of context to get
25:01through in order to really understand
25:03the data
25:04and what i mean by that is you
25:06you need more than just data collected
25:09at the time of the interview
25:11or whatever the data collection is but
25:14you might also need to consider basic
25:16social characteristics of the
25:17participant or the historical time
25:20period in which the data was collected
25:22or perhaps where the data was collected
25:25so the idea is that you need to
25:27understand the data set as a whole
25:29in order to really get at the root of
25:31what it can convey
25:34every collection that is archived at the
25:36uk data service
25:39will hold some documentation
25:41which is a really useful starting point
25:43for doing this
25:44and that might contain information about
25:46the methodology such as an interview
25:48schedule or a call for participants or
25:51sometimes it includes segments from
25:53publications that were arising from the
25:55original study or it might even have
25:57something like funding applications
26:00and for qualitative data sets this
26:02documentation is called the user guide
26:04or the u guide
26:06and here's an example
26:08of a user guide this one happens to have
26:09an interview guide for interviewers
26:12but there's also a consent form a sample
26:14profile and so on it's just further
26:17background information to help you
26:18understand how the data was collected
26:23but what if you want to know more about
26:24the participants themselves
26:27every qualitative study also have uh it
26:30has what's called a data listing or a
26:33new list
26:35and here's an example of one of these
26:38and it's a
26:39table that gives a brief overview of the
26:41data in the collection
26:42so each row
26:44represents a piece of data or a
26:46participant and each column lists some
26:49kind of characteristic or attribute of
26:51that interview so it's a quick way of
26:54getting to know who took part in the
26:55study
26:56basically
27:00in addition to the context
27:04of the data you may also need to
27:06consider the sample
27:07so for example if the data set is too
27:09large you may need to take a sub sample
27:12qualitative collections tend to be
27:14smaller anyways but many of the archive
27:16data sets are funded and they can
27:18collect a considerable amount of data
27:21for a small dissertation project you'll
27:23want to be realistic and decide if you
27:25need to limit the number of participants
27:27to a smaller sub-sample of those
27:30you know
27:31of those that took part within the
27:33larger collection
27:34so for example the edwardians collection
27:37which was put together by paul thompson
27:40and it's largely considered to be the
27:42first oral history of britain
27:44contains
27:45453 80 plus page interviews it's a huge
27:50collection
27:51conversely most dissertation projects
27:54probably have an expectation of
27:56you know six to ten interviews
27:59so you'd want a clear sampling strategy
28:01to help you choose which of those
28:03interviews you want to look at
28:05or you might be interested in a
28:07particular subgroup of the population so
28:09again you'd want to think about what
28:11criteria you're looking for
28:14you might also want to combine different
28:16data collections that complement each
28:18other
28:19and remember it takes time to sift
28:21through and find the pieces of different
28:23data sets to pull together but it's
28:25another possibility if you feel that all
28:28of the data speaks to the same topic and
28:30research question and you've done the
28:32work of recontextualizing
28:34to ensure that the interviews work
28:36together then there's no reason why you
28:38can't combine some of those and create
28:41your own custom data set
28:44so that is it for
28:46um qualitative i'm gonna pass over to
28:49ali now so just uh bear with me for a
28:51moment while we
28:53um
28:54yes thank you i'm just gonna switch
28:56screens um
28:59okay perfect you should all be able to
29:01see that so thanks maureen um as maureen
29:03said i'm going to give a quick rundown
29:05of the key methodological things you
29:07need to consider if you are using um
29:09quantitative secondary analysis or doing
29:12a quantitative reuse project
29:14so um i'm going to cover two main areas
29:18first the things you need to think about
29:20when you're selecting your data and
29:22secondly the things you might need to
29:23think about when you're understanding or
29:25getting to know your data
29:28so the first section is on selecting
29:30your data
29:32so some of you might already know which
29:34topic area that you want to look at or
29:36explore for your dissertation but if not
29:39i'm just going to give you a brief idea
29:40of some of the data available
29:42so the uk data service holds
29:44quantitative data on a variety of topics
29:47and if you attended jen's webinar
29:48yesterday you might know a little bit
29:50more about them but just to give you an
29:52idea now there's quantitative data
29:55available that could allow you to look
29:56at the environment
29:59workforce patterns
30:01health care
30:02family spending
30:04attitudes to the police and criminal
30:06justice system time use during the
30:09pandemic
30:10attitudes
30:12and political opinions and these are
30:13just a few examples there are many many
30:15more topics covered
30:18so once you have a general idea of your
30:20topic area you can think about the data
30:22you might use to explore it and the key
30:25thing for a quantitative reuse project
30:27is to understand what you are trying to
30:28measure so think about your topic
30:30questions topical questions think about
30:33your key concepts and importantly think
30:36about how these relate to the variables
30:38within a data set
30:41so for example let's say we were
30:43interested in looking at the
30:44relationship between fear of crime and
30:46age
30:47so we need to find some data that has
30:49variables that measure these concepts so
30:51our key concepts fear of crime and age
30:54that will allow us to formulate and or
30:56answer a research question about them
30:59now as maureen said earlier it may not
31:00be possible to find the perfect data set
31:03and you could spend a lot of time
31:04searching so instead you could start
31:06from data on a general topic and derive
31:08your key concepts and questions from the
31:10existing variables once you've gotten to
31:12know the data
31:14if you do have a question in mind
31:15already that's fine too
31:17but you may just have to be flexible and
31:19revise it based on the data and the
31:20variables that are available
31:24so if you're looking for data on a
31:26particular key topic there's a few key
31:28places you can start
31:30so you can use the theme pages on the uk
31:32data service website to look for key
31:34data sets on a particular theme
31:36you can type keywords you're interested
31:38in into the data catalog
31:41or you might want to search the variable
31:42and question bank for variables on a
31:44particular topic but please be aware
31:46that this hasn't been updated for a
31:48little bit so it might not encapsulate
31:50all the current variables
31:52you could also search hasit which is the
31:55social sciences thesaurus which lets you
31:57search data sets by key concepts
32:00and the link to all of these can be
32:01found on the find data section of the
32:03website and we will have a little
32:05practice of this in the practical as
32:07well
32:10so once you have found a data set that
32:12you think might be suitable you need to
32:13consult the catalog page
32:15and this will give you an overview of
32:17the key topics background and a brief
32:20overview of the methodology
32:22and you can also access the
32:23documentation from here including any
32:25user guides technical reports lists of
32:28the variables and any notes added by the
32:30data producers to highlight any changes
32:33re-weighting or editing to variables
32:35that might have been made since the data
32:37was first deposited
32:39and as maureen said the documentation is
32:42really
32:46importantly the list of variables can
32:48also usually be found what can be found
32:50in the documentation and this is usually
32:52in the user guide the variable list or
32:54the code book or data dictionary and
32:56this might vary between data sets
32:59and here you can find important
33:00information on what the variables
33:02measure which is how you will help
33:04relate them to your key concepts
33:08so back to our example that we were
33:09looking at earlier we might choose to
33:11look at the crime survey for england and
33:13wales
33:14it's a large important crime survey and
33:16it's a source of information about crime
33:18that is independent from police records
33:21it is an example of a repeated
33:23cross-sectional survey
33:25don't worry if you don't recognize all
33:26these terms we'll go into them a bit
33:27more in a bit
33:29it's a survey that's conducted every
33:31year and it has a large sample size of
33:3435 000 individuals aged 16 plus and 3
33:37000 age 10 to 15.
33:40so from having a look in the
33:42documentation i can see that the crime
33:44survey for england and wales has two
33:46variables which fit our key concepts so
33:48the first of these is qualify which is
33:51how much is your own quality of life
33:52affected by fear of crime on a scale
33:55from one to ten where one is no effect
33:57and ten is a total effect on your
33:58quality of life
34:01and the next is age the respondents age
34:05so once you've found your variables and
34:07it might take you a while looking
34:09through the documentation looking
34:10through the different information to
34:12find the ones that are relevant the
34:14important step is to think critically
34:16about what they measure this is
34:18particularly important for um
34:21reuse projects because you're not asking
34:23the questions um
34:25yourselves
34:26so you really need to think about what
34:28your variables um measure so for example
34:30this quality life one this call life one
34:33does it really measure fear of crime or
34:35does it perhaps instead measure how much
34:37fear of crime affects life
34:39so you need to think about what the
34:41question is really asking and to
34:43understand this you can look at the
34:44original questionnaire which can usually
34:46be found in the documentation
34:50so as well as considering your variables
34:53and your concepts you also might want to
34:55consider the kind of
34:58analysis that you want to do or that you
35:00can do with the data so for example if
35:03you were interested in looking at
35:04individuals in a particular time point
35:06you might want to use a cross-sectional
35:08survey
35:10if you want to look at
35:12data across multiple points in time you
35:14might want to use a repeated
35:16cross-sectional survey so that's like
35:17the crime survey that we were looking at
35:20if you are interested in
35:23data over a long period of time you
35:25might look at longitudinal data
35:28if you're interested in small geographic
35:30areas you could use the census aggregate
35:32data or flow data
35:33and if you want to compare countries
35:35over time you can use international time
35:37series data and you can search by all
35:39these different data types on the uk
35:41data service website
35:46so as well as this it's also important
35:49to think about your population so that's
35:51the group you want to measure
35:53so for example this might be the
35:54population of the world the uk or
35:57perhaps you want to ask questions about
35:59a particular city or local authority
36:01area
36:03and as well as this you need to consider
36:05your unit of analysis
36:07are you interested in individual people
36:09or are you interested in whole
36:11households and this will affect the data
36:13you use in the kind of analysis you can
36:15do in the questions you can ask as some
36:17data sets are only available for
36:18particular geographies or for certain
36:20units
36:23so finally it's important to remember
36:25that the process isn't linear you may
36:27need to go back and forth and realign
36:29your question with the available data
36:31you might need to compromise if the
36:33perfect data set isn't available
36:35or choose different variables and refine
36:37your question and this is all part as
36:39lauren said earlier of doing secondary
36:41or analysis or a reuse project
36:45so once you've chosen your data set
36:47there are a few final things you need to
36:49consider
36:52so as we said the documentation usually
36:54contains information on the variables
36:56but it should also have information on
36:58the questionnaire used to collect the
37:00data
37:02using data and to understand your
37:04variables it's really important to
37:06understand the questionnaire and
37:07something that's really particularly
37:09important is the routing and that's who
37:11was asked which questions
37:14and this is because many questionnaires
37:15use something called computer-aided
37:17interviewing which sends respondents
37:19through the questionnaire by different
37:20routes depending on their previous
37:22answers
37:23so therefore many questions in a survey
37:25may only be applicable to some of the
37:27samples so it's always a good idea to
37:29check the documentation to see who was
37:31asked about the variables that you're
37:32interested in
37:35so for example
37:36this is a variable called flex 10 from
37:38the labor force survey which relates to
37:40special working arrangements
37:42uh in this catalogue in this
37:44documentation entry you can see the
37:46exact wording of the question
37:47as well as a range of answers and how
37:49they're coded and importantly underneath
37:52this we can see that the question was
37:54asked if a respondent is in work and it
37:56also gives information on the variables
37:58used to identify who was in work
38:01and you'll find that the documentation
38:03looks a bit different across different
38:04data sets but this gives you a general
38:06idea of the kind of thing to expect
38:10as well as this you can also find
38:12information about how the data was
38:14processed after collection
38:16so important thing to consider are the
38:18derived variables and these are
38:19variables created from the raw data
38:24the derived variable
38:26so this shows how the flex 10 variables
38:29which we talked about on the last slide
38:30have been used to derive a new variable
38:33called flex w7
38:35and to those who responded seven on the
38:37flex 10 variable indicating they had a
38:39zero hours contract and now coded as one
38:42on the new flex w7 variable and
38:44therefore it's a new variable which
38:46measures whether someone has a zero
38:47hours contract or not now don't worry if
38:49this looks a bit complicated it can take
38:51a while to get your head around the
38:52documentation but if you are using a
38:54derived variable it's really important
38:56and you understand this
38:58and just to say that not all
38:59documentation contains these diagrams in
39:02some surveys it might just be the spss
39:05or status syntax which show you how the
39:07variable variable's been derived but
39:09although it looks complicated if you
39:11take a bit of time you should be able to
39:12get your head around it and understand
39:14where the derived variables have come
39:15from
39:17the final thing you need to think about
39:19as with qualitative analysis is sampling
39:22so surveys and similar quantitative data
39:24sources are almost always based on
39:26samples
39:28and one important question you need to
39:30ask is is your sample representative
39:33so you need to know
39:35who is included in the sample
39:37is it adults is it only those who live
39:39in private addresses is it only those
39:42who attend school
39:45who was the what was the response rate
39:47and is there any information about
39:49different response across the population
39:51and this will tell you potential bias
39:55and you also need to know do you need to
39:57apply a weight to make the data
39:58representative
40:00and you'll once again find all this
40:01information in the documentation that
40:03comes along with the data so in the user
40:05guides and the technical reports
40:08finally you need to think about whether
40:10you have enough cases to make a precise
40:12estimate
40:13so for example the crime survey for
40:15england and wales which we looked at had
40:16a very large sample size which should
40:18allow you to make precise estimates
40:21however with smaller samples an analysis
40:24of smaller sub-populations so for
40:26example say you're cutting your data set
40:27down to study only a particular group of
40:30people
40:31you might not have enough cases in your
40:32sample to make precise estimates
40:35now i've just given a very brief
40:37overview of sampling considerations
40:39and there are more resources available
40:41on the uk data service website so we
40:43have a guide on survey weights and
40:45complex sampling so do look those up
40:47when you come to that part of your
40:49dissertation to help you understand
40:50those more
40:52so finally just to summarize for uh
40:54quantitative projects you need to think
40:56about your key concepts what you're
40:58trying to measure and how these relate
41:00to the variables in a day
41:02important that you check the catalog and
41:04the documentation to help you understand
41:06your data and the variables and where
41:07they came from and you need to consider
41:09issues to do with sampling
41:14okay so maureen did you just want to
41:16talk through some of these additional
41:18resources
41:20yeah so hopefully you know you've got
41:22some inspiration now for a project you
41:24could do
41:26and um you know this workshop is really
41:28just a quick overview and i know there's
41:29a lot of information being thrown at you
41:32so we just wanted to give you some
41:33further resources which will help you
41:36navigate your way through a reuse
41:38project
41:39so there's a few publications out which
41:41would be useful if you're doing
41:43quantitative work
41:44then emma smith's book using secondary
41:47data in educational and social research
41:50might be useful there's also a chapter
41:52in sages handbook of social research
41:55methods
41:56on secondary analysis of quantitative
41:58data sources
41:59and we also have some data skills
42:01modules uh which will help guide you
42:04through you know steps of the of
42:05statistical analysis
42:08for qualitative projects there's a
42:10chapter in silverman's qualitative
42:12research textbook on secondary analysis
42:15and there's also a working paper series
42:17available through the timescapes archive
42:20karen hughes and aaron tarrant have also
42:22recently published a handbook of
42:25qualitative secondary analysis
42:27and finally if you need more help with
42:29finding accessing your downloading data
42:32we have video tutorials which will walk
42:34you through the process with
42:35step-by-step procedures

About this transcript

This page contains the full transcript of Dissertation projects: Introduction to secondary analysis for qualitative and quantitative data by UK DATA SERVICE, generated from the public captions YouTube serves with the video. The transcript has 6,925 words across 1,172 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.