You are not currently logged in. Please create an account or log in to view the full course.
Why Does Sociology Need Numbers?
Generating Lecture Summary...
Generating Lecture Summary...
Generating Vocabulary List...
Generating Questions...
Generating Questions...
- About
- Transcript
- Cite
About the lecture
In this lecture, we think about the use of numbers in social research, focusing in particular on: (i) the importance of numerical data in identifying patterns of stratification and differentiation, both in education and in society more broadly; (ii) the distinction between qualitative and quantitative in social research, and whether this binary is helpful; (iii) some important factors to consider when evaluating the quality of research, whether “qualitative” or “quantitative”.
About the lecturer
Professor Stephen Gorard is Professor of Education and Public Policy and Director of the Evidence Centre for Education at Durham University. He has written very widely on educational inequalities, education policy, and research methods. His recent publications include How to Make Sense of Statistics (2021), How Can We Get Educators to Use Research Evidence? (2019, co-author), and Education Policy: Evidence of Equity and Effectiveness (2018).
Hello. My name is Steven Gerrard.
00:00:05I'm a professor of education and public policy at Durham University.
00:00:08I also direct and evidence centre for Education.
00:00:11So starting with the first issue,
00:00:14which is why Why are we worried about numbers in sociology at all?
00:00:17I suppose I would argue that everybody needs numbers in real life.
00:00:22Let's start with real life.
00:00:26Currently, we are still under covid 19 restrictions in the UK
00:00:27The news is still full of things about the numbers of
00:00:34people who had covid or who have been hospitalised in covid,
00:00:37or the success rate for transfer or transmission of covid
00:00:40with different vaccines and the side effects of vaccines.
00:00:45And all of these are based on numbers.
00:00:48And as a population we are generally becoming more familiar with
00:00:50these with risk rates and all these kinds of things.
00:00:53And what we do in sociology with numbers is no more complicated than that.
00:00:58That's the kind of thing that we see on the news every day, employment rates
00:01:03and in your own life, your past, on life.
00:01:08If you were planning a journey or renting or buying some accommodation
00:01:10or checking the sports scores or looking at the weather forecasts,
00:01:14you would be using numbers and you'd be probably using them completely,
00:01:17openly, reasonably, without concern
00:01:23and so on.
00:01:26And it's the same in research.
00:01:26As soon as you put your kind of research hat on,
00:01:28I'm going to look at or do some research in sociology or the sociology of education.
00:01:31Um,
00:01:37you're still going to be dealing with numbers,
00:01:38too.
00:01:40Obvious arguments would be first of all,
00:01:40so much of sociology is based on stratification
00:01:43and on the potential and fairness of life,
00:01:46chances and opportunities.
00:01:48And these can be very, very succinctly expressed as numerical expressions.
00:01:50And the other one, of course,
00:01:56is if you're going to read the research other
00:01:57people have done and try and summarise it,
00:01:59maybe you're doing a little dissertation or project,
00:02:01and you want to look at what people have done in a particular area.
00:02:04Some of those people, at least, will have used numbers.
00:02:07So if you're going to look fairly at all the research in an area, you must be prepared
00:02:10to look at research involving numbers.
00:02:15The reason I'm making a fuss about this in the first session is just this idea,
00:02:20which you've come across on the syllabus
00:02:24that you can divide research or research evidence
00:02:26into things which are called qualitative and quantitative.
00:02:29Now I find those labels as a researcher and as a successful researcher.
00:02:34I find them both confusing and unrealistic.
00:02:38I'm not really sure. Why is this this emphasis on them?
00:02:43They kind of mean the quantitative is about using numbers,
00:02:46and the qualitative is about using any other form of data,
00:02:50and I think it's unrealistic.
00:02:54So if you imagine again in real life, perhaps in your individual life,
00:02:56you or your parents or whatever, decided to buy a house.
00:03:01So someone's going through the process of buying a house, what would they do?
00:03:05I don't think they think about the philosophical ideas.
00:03:09They wouldn't worry about whether the house was real or permanent or not.
00:03:12They're very unlikely to say to the estate agent, No, don't tell me the price.
00:03:16I'm not interested Numbers? I don't know.
00:03:22I don't want to know how many rooms there are.
00:03:24Don't even tell me the address because that's got a number in it.
00:03:26Nor are they very likely to say I don't want to visit the house.
00:03:29You know what you would do is you'd go along
00:03:33and you'd compare for the cost of the house.
00:03:35You'd look at things like,
00:03:38you know, the quality of the neighbourhood, the noise from the pub,
00:03:41the smell of the damp in the back bedroom and so on.
00:03:45And you put all of these factors together.
00:03:47You know, the size of the garden and synthesise that material and come to a decision
00:03:50that you did or you didn't want to make an offer to buy that house.
00:03:55And if you think of any skilled task or any task
00:03:59you undertake in real life where you care about the results,
00:04:01the safety of your loved ones or something,
00:04:05that is how you would behave, nobody would say, Oh, no, I'm a quantitative research.
00:04:07And therefore, I'm not going to look at this or I'm a qualitative researcher.
00:04:11I'm not going to look at the other thing.
00:04:15All research, all successful research actually uses data of any kind.
00:04:17You know, we go out there were hungry for, um,
00:04:23evidence that we can collect and we would never turn anything away.
00:04:26So, for example, if you were writing a report, a few a level A s level studies
00:04:31and you decided to add a table of figures into that report.
00:04:37That wouldn't mean you'd have to,
00:04:41um, you know,
00:04:43forgo a particular paradigm you're working in or mean that you've
00:04:43entered some new universe or giving you some new philosophical commitment.
00:04:47You could just be showing how many people
00:04:51are unemployed in a particular area or something,
00:04:53and that's just natural.
00:04:55And I really can't understand why we have this divisive idea.
00:04:56That kind of then gives people the reasons that all I
00:05:03don't do that or I'm not going to look at this.
00:05:05All evidence is relevant.
00:05:07And if you care about society,
00:05:09if you compare about as I do as trying to fight unfairness and injustice,
00:05:11then numbers will be a key part, but only a part of what it is you do.
00:05:16One of the arguments I've heard is that well, you know,
00:05:21quantitative research is objective and qualitative is subjective.
00:05:24But again, that doesn't make any sense.
00:05:30It doesn't stand up to even superficial scrutiny.
00:05:31Imagine you were you were catering for a party may
00:05:35be a wedding or some kind of formal party,
00:05:38and you knew how much it would cost for the
00:05:41drinks or or the food for each person who was coming
00:05:45and you knew how many invitations you'd sent out.
00:05:48But you didn't know how many people were coming,
00:05:51and you were trying to work out the cost of the party.
00:05:53You could make an estimate of how many people you
00:05:55thought were coming and multiply it by the number,
00:05:58which is how much the party would cost per person.
00:06:01Now that would be like an objective figure. It would be a true figure.
00:06:04You know how much it would cost.
00:06:07Your estimate would be subjective,
00:06:09but you could easily multiply them together to come up
00:06:11with an estimate of how much the party would cost.
00:06:13And again, there are no sort of philosophical commitment.
00:06:16We do that all the time in real life, and we should do that in our research.
00:06:18In the measurements are largely of qualities. That's what you're measuring
00:06:22someone's self esteem or how attractive houses
00:06:27or whatever it is you're trying to measure
00:06:31and
00:06:33analysis of things like interviews,
00:06:33nearly always comes up with things like, Well,
00:06:37most respondents said this or a few people said that or somebody said that
00:06:39and really What we're all trying to get at for
00:06:44any piece of evidence is how big is a difference,
00:06:47or how strong is a pattern or a trend, or how valid is a particular exception,
00:06:51a unique example we're going to talk about.
00:06:57So I you know you could write about this,
00:07:02but I really think you should drop any idea that there are these different, you know,
00:07:05silos or schisms of research and just focus on doing research,
00:07:10which must inevitably, if it's any good to you, use numbers.
00:07:15It's quite clear that when you design a project, the design is independent of
00:07:21the methods of data collection or analysis. So, for example,
00:07:27if you were looking at a longitudinal study,
00:07:32a longitudinal study is one where evidence is collected
00:07:34from the same group. Maybe they're people.
00:07:38There could be institutions or whatever,
00:07:41but imagine their people and collected from the same people
00:07:43over a number of times.
00:07:46So maybe over three years,
00:07:48every year you would contact this group
00:07:50of people and collect information from them,
00:07:52like in the British Cohort studies.
00:07:55But
00:07:59if you collected information like that
00:08:00and you interviewed the people every year,
00:08:03the study will belong to general. If you surveyed the people every year,
00:08:06the study would still belong to terminal.
00:08:11If you ask to see their pace lit every year,
00:08:13the study would belong to digital and so on.
00:08:16If you did all of the above, it would still belong to digital.
00:08:18The design is nothing to do with the data you collect.
00:08:21It's completely independent of it
00:08:25and other issues like, you know, how many cases would you need to conduct a survey?
00:08:28It doesn't matter if the survey is face to face online
00:08:33a postal, whatever it is.
00:08:37The answer to the number of cases you'd want is
00:08:40independent of the method of delivery of the survey.
00:08:42There are no particular quantitative or qualitative designs.
00:08:47It's important when you're doing a systematic review.
00:08:51If you are going to review all of the literature
00:08:55in a particular area to come to an answer on,
00:08:57what do we currently know in social science about a particular topic
00:09:00that you search for all of the data you can't cherry pick?
00:09:05What you have to do is to look at all of it and
00:09:09then wait it by the quality that you perceive that study to be,
00:09:12In which case you'll need some rules for how you would judge the quality of
00:09:18a piece of research so you'll see on the screen now a table that suggests four
00:09:21factors that you might like to consider at the beginning,
00:09:26when you're judging how much you should trust a piece of research evidence.
00:09:29So the first thing was, Is the design appropriate
00:09:33for the question being asked?
00:09:36And it's often the case that's not so.
00:09:38I've seen many studies where people are looking at changes
00:09:40over time and saying things have changed over time.
00:09:43But they've only got data about today,
00:09:45or that it's different between two groups,
00:09:48but they've only got data from one of the group's
00:09:50education for homeless people is worse than for non homeless people.
00:09:52But we've only got data on homeless people,
00:09:55and I'm not disputing that the education might be worth.
00:09:57What I'm saying is for research.
00:10:00You do need the evidence from both groups,
00:10:02so you'd want to design that would match.
00:10:05The questions are being asked if the question is comparative,
00:10:07design must be comparative with appropriate groups.
00:10:10The thing you know,
00:10:14the second column is about how large would you want to study to be here?
00:10:15The key issue is the size of the smallest cell in any comparison.
00:10:19So if you're just making a descriptive claim about one group,
00:10:24that's the size of the group.
00:10:28But if you're comparing two or more groups,
00:10:29then the strength of the study depends on the size of the smallest group.
00:10:31Just to convey in a silly example, what I mean by that. Imagine
00:10:35there was a sociology research project,
00:10:40and there were a million people took part in the study.
00:10:42You think that's a lot of people? That's a strong study.
00:10:45But imagine they were trying to compare two groups
00:10:48among those and 999,999 people were in one group,
00:10:50and there was one person in the other
00:10:55than any comparison would be very,
00:10:57very weak because there's only one person in the other group.
00:10:59It doesn't matter. You got a million cases. It's not about the overall scale.
00:11:02It's about how small is the smallest group.
00:11:06The third column is about missing data now.
00:11:09Missing data can occur at any stage in a study
00:11:12and is often very badly reported
00:11:15So if you set out to try and interview some people or to do a survey of people,
00:11:17you need to get a sense of, Well, how many people have said No,
00:11:22you've got what's called non response.
00:11:25Then, after cases that agree to take part,
00:11:27some of them will not respond to some of the items.
00:11:29You know, If it's a survey, you might have 20 questions,
00:11:32and somebody might miss out the 19th questions.
00:11:35And then, if you're going to do something like a longitudinal study,
00:11:38where you're collecting data again and again from the same people,
00:11:42you'll find that some of the people will drop out
00:11:45and you'll get what's technically called attrition in your study
00:11:47so that in the end you'll have missing data from a whole range of sources,
00:11:50and none of it will be random. It will all be patent.
00:11:55You know, the people who don't respond.
00:11:58The people who drop out will do so for a reason,
00:11:59and therefore you can't just assume that the dropout doesn't matter.
00:12:02It does so the more drop out there is,
00:12:06the week of the study is,
00:12:08and the final
00:12:10column I've got there won't spend long on.
00:12:11It's about how good the measurements are that you're using in your study.
00:12:13So, for example,
00:12:17it's easier to count how many people around the room and come to an accurate answer,
00:12:19which you can check with other people than it
00:12:23is to measure the self esteem of an individual.
00:12:25For example, you're looking at a latent or hidden factor,
00:12:29so the measurement is necessarily going to be weaker.
00:12:32That doesn't mean we shouldn't research those areas.
00:12:35It means that when we're looking at studies that have
00:12:38done that or studies that we do in that area,
00:12:40we must be more humble about the quality of the measurements we have.
00:12:42And I'm just proposing on that table because you start in the first column,
00:12:47find the best description that suits the research study you're looking at,
00:12:51and then move to the next column and the same row.
00:12:54Don't move up, move down or stay there
00:12:58and keep doing that with each column
00:13:01until you've got to the end.
00:13:03And then that padlock image is how security and of course,
00:13:05having five categories is purely arbitrary.
00:13:09But for simplicity, you could use a scheme like this
00:13:12and judge it and then what it means is, if you find three padlock,
00:13:16four padlock kind of studies,
00:13:20we should trust them a lot more than the ones that are not or one or even two.
00:13:22Of course, there are other factors you want to look at,
00:13:28but normally what happens is if you've done those four, you've set the bar.
00:13:30Nothing else really will alter how good the research is.
00:13:34It's very unlikely that the piece of research is brilliant in all other aspects,
00:13:37but is very small, has high dropout, very poor measurements and so on.
00:13:42What I'm trying to lead to is the idea that, first of all,
00:13:48a you must use numbers you can't simply ignore.
00:13:51Work that does involve numbers, in fact,
00:13:54should celebrate the use of numbers because it's very powerful technique,
00:13:56As I hope I'll demonstrate in future sessions.
00:13:59But secondly,
00:14:04when you refer to a piece of work or the work of a previous research or sociologist,
00:14:05you must wait.
00:14:12Your citation by some might give some idea of how much you trust that work.
00:14:14If you cite it without any comment,
00:14:20the assumption of the reader is that you think that's
00:14:23ideal, perfect, wonderful piece of research.
00:14:25And of course, that doesn't happen in real life. We all have compromises.
00:14:29We all have missing data and so on.
00:14:33So what I'd like to see is more people being a bit more tentative.
00:14:34And so, if you like grading the tentativeness,
00:14:39say they're making their statements with according to how good
00:14:42the research basis is for what they're trying to claim.
00:14:45
Cite this Lecture
APA style
Gorard, S. (2021, September 30). Social and Cultural Continuity and Change - Why Does Sociology Need Numbers? [Video]. MASSOLIT. https://massolit.io/options/social-and-cultural-continuity-and-change?auth=0&lesson=4052&option=16784&type=lesson
MLA style
Gorard, S. "Social and Cultural Continuity and Change – Why Does Sociology Need Numbers?." MASSOLIT, uploaded by MASSOLIT, 30 Sep 2021, https://massolit.io/options/social-and-cultural-continuity-and-change?auth=0&lesson=4052&option=16784&type=lesson