Welcome back today we're going to have a chat about qualitative versus quantitative data okay and this is the one of the main dichotomies in research between two types of research and two types of data. And you might find that researchers prefer one to the other. Is that the case for you guys? Do some of you prefer quantity and some prefer quality? Always quantity. Always quantity with you. Okay. Who here prefers qualitative? All about the words. All about the words. And I do mixed methods. And you do mixed methods. So you do a bit of both. Marvellous. Okay. I'm a bit more towards the qualitative because that's my dissertation is on qualitative analysis. So I'm a little bit more that way. But...
Hopefully, by the end of the day, we'll see the approach for both of them. Okey-doke. Short one today, just to be able to describe examples of different types of research and their strengths and weaknesses. That sounds reasonable, doesn't it? Brilliant. And as ever, we're a nice, friendly environment. Please feel free to put your hand up and ask a question or make a point or disagree with me. I don't mind. Brilliant. So what is the main distinction between quantitative and qualitative data? You've kind of already discussed it, haven't you?
numbers versus words brilliant absolutely so as we've got a kind of a pick and mix theme here is some pick and mix okay so if we were to approach this from a quantitative viewpoint how would we approach it what sort of data would we be interested in what sort of stuff would we want to measure and what sort of research would we do you know I'm going to perform some quantitative research on pick and mix sorry so how many of each suites there are yeah brilliant yeah Colours, brilliant, yeah. So what colours of sweets there are there? Yeah, so what's actually sold, what isn't sold, brilliant, yeah. That was going to be yours, was it, Louis? Okay, so people who prefer quantitative, what else would you, you know, this is your domain, what would you want to do? Size of the sweets, brilliant, yeah. Colours, yeah. Weight of the sweets, yeah. What else? The healthiest of calorie content, yeah. Good.
Any chemographics who choose the sweets? Brilliant, yeah. Who buys what sweets? Brilliant. Who has diabetes? Brilliant. We're looking at it at a clinical point of view. So who has diabetes who buys these sweets? Brilliant. So if we're thinking about it more clinically then from maybe a research nurse point of view with these sweets, what would you want to do? Do you buy wine gums more often? Again with the wine gums? Yep. What sort of sweets do researchers make and put in any of these?
Yeah. So what suites do research nurses go? And you might compare that to non-research nurses. So, you know, educators, the general public, etc. Brilliant. How do you control it? So you can have a control, can't you? So you might want to think about how can I have a control with this quantitative data? So, you know, that might be the general public. It might be people who don't take pick and mix. So you could then have research nurses who use pick and mix and their health outcomes versus those who don't use pick and mix and their health outcomes. Yeah, it's good. Okay.
And then what are we going to be able to do with that data when we get it? It's going to be in the form of numbers, isn't it? Quantity of certain tweets isn't their favourite. Yeah, absolutely. They don't overstock on certain ones that they have. Yeah, so like Megan said, we can do some statistics with it brilliantly. And then how can we present our data? It's people. Yeah, some form like this, can't we? So quantitative data is in numbers, it's in categories, it's measurements, it's things that we can rank. We can present it in one of these sort of formats in a nice clear graph. Somebody can stand up in a meeting and say, we are selling more tongues to research nurses. Therefore, when, you know, we should sell, we should make sure we stock up more on tongues in the pick and mix of places that are near the research nurses, okay? And like Lucy said, it can guide how we're doing it, okay?
They're about tongues. They're sweets. They're teeth. Are they teeth, are they? Sorry, teeth, not tongues, sorry. Maybe, sorry. This is why you keep aspirating watermelon. This is why I keep aspirating watermelon. Okay, what about qualitative? Yeah, okay.
What does it taste like? What does it taste like? Yes. Does your pick and mix change on your mood? Does your pick and mix change on your mood? Yes, absolutely. Why have you made that choice that you've made? Why is that today? Would it be different on another day? Yeah, brilliant. Okay. Anything else?
Yeah, brilliant. Yeah, absolutely. So why pick and mix and not something else? Why not something else? Brilliant. Yeah, absolutely. So we might, for example, use these sort of graphs to say, oh, this is, you know, we're selling a certain flavor of ice cream more than another. But for this lady right now having this ice cream, you know, why has she chosen that flavor? Is it because it brings back memories of her childhood? Is it because it was whatever was on sale that day? Is it because she's on a diet and this is the lower fat option? Et cetera, et cetera. Okay.
The reasons why she bought that ice cream would be less easy to put in a graphical data. How are we going to find out why she bought that particular flavour of ice cream? What sort of research model are we going to use? Questionnaire, yeah? What sort of questionnaire? Open-ended. Open-ended questionnaire, yeah, because I can give you a questionnaire that's got five choices, and that can in theory be quantitative, because I can put that into categories. But yeah, an open-ended, here's a one side of A4, please tell me why you bought this flavour of ice cream, yeah?
interview brilliant absolutely yeah sit down with her and you can do a recorded interview or you can transcribe it while you're talking to her yeah yeah you can do it retrospectively again you want to be knowing in her words as much as possible and that's key isn't it you want the subject's viewpoint as much as possible yeah in what way yeah there's more of an observational element there's more of a longer time period element there isn't there with with qualitative research brilliant yeah okay so the viewpoint of the world with quantitative research is very fixed there is a truth that is out there that is universal I will find it and I will present it to you qualitative that's much more dynamic and it's much more dependent on people and the settings okay so you can see that this much more favors itself much more towards social sciences okay people and their opinions their perceptions and their beliefs okay cool
right then and then how are you going to present qualitative data once you've got it yeah absolutely so you're going to have a look through aren't you okay and you're going to identify themes okay I've interviewed 10 people about what ice cream choices they've made what are the themes that have come out when I've read what they've written or what they've said when that's been transcribed brilliant yep anything else in other ways we can present qualitative data yeah absolutely yeah yeah good yeah I mean we try and do as little as possible to qualitative data because we want to show it in their words okay because if we paraphrase and put our own meaning on it we can alter the the findings but it's quantitative we do a hell of a lot of things to it we do all sorts of statistical analysis before we present it cool so team quantitative and team qualitative so you've got a piece of paper in front of you what I'd like you in your groups and Lucy's going to help is I want you to think about the strengths the weaknesses how you might design a trial
Also think about the role of the investigator to this research. How distant, how embedded are they? Also have a think about the goal of that research, the sample size, are variables important to that research? How relevant is it to other people if they're reading your paper? How generalizable is it? How valid and reliable is it? Does that make sense? Cool. So you've got strengths, weaknesses, the trial design, the role of the investigator, the goal of that research, sample size, variables, and others.
Nicely done. Finished? Sort of, yeah, cool. So quantitative leap. So would you like to take us through what you found out then, what you've come up with? So strengths, it can produce strong evidence. So strengths, strong evidence, yep. They can be more powerful evidence. Yep, why is that? Because there's more proof, there's more like numbery proof that it works rather than a hearsay proof that it works. Sure, okay, yeah. It's objective about a certain...
subject, it doesn't have any confounding factors. And for weaknesses, it's only got depth about one subject, it doesn't look at the breadth of that subject. It can be expensive. You potentially need a large sample size to make it statistically significant. Variables are, well, everything. It's all about the variables. So we try and control as many of these as possible, but there's always going to be extraneous variables. Your goal is to answer slash prove your hypothesis or your null type hypothesis.
The sample size is potentially big to create statistical significance. Generalizability depends on your sample size, so it can be extrapolated if you have a good, well-rounded sample. However, it can be too tight a sample and therefore not. generaliser. Sure. Trials design is an RCT platform. Yep, so randomised control trial is going to be your gold standard here, isn't it? Yeah, absolutely. And investigators have the ability to data mine or corrupt the data when they're doing statistical significant stuff. Yep. But we have less contact with the actual individual, so less spinny on the data. Yeah, brilliant. Yeah. And what also can we do to the investigators? Blind them. Blind them. Brilliant. Yeah, absolutely.
Sorry, poke out their eyes, hopefully not literally. Okay, so when you guys are doing your, you guys take part in randomized control trials, how involved do you feel with that data? With the data, with the patient, with the subject. Yeah. Less so. More numbers. And you get to talk to them. Depends. Yeah. Yeah, absolutely.
Yeah, so in quantitative research people are broken down into just numbers really in categories, aren't they? You know, are you male, are you female, what's your blood pressure, etc. And you can be the investigator with just a piece of paper with some numbers on it. You could be triple-blinded so you don't even know, you know, are these patients who got the placebo, are these patients who got the tranexamic acid, for example, for one of the trials. So you don't know, you're very far away. You could be doing research on data collected from Germany that's emailed over to you, sat here in Nottingham and doing it and you are distant from it.
you're going to need a large sample because small quantitative studies are not as useful if you need a large one okay but like Megan said it's a very narrow focus okay only on one thing so lots of people but you're only interested in one thing hello yeah so what is a platform trial Sure Super large Okay, so because usually yeah
Okay, because the variables are all important in quantitative research, and whether that's only one variable or four variables, it's all about it in quantitative research, and you're going to want to think about having controls as well as that. And because of the amount of people, because you might be using more sites, it is going to be more expensive, it is going to involve more collaboration. But the data you've got... is potentially that bit more useful because it goes through statistical analysis. So somebody over in America can look at it and go, do you know what? This is probably more generalizable. I think it's probably more representative of my population as well. So that's good.
I think the downside of this, I would say, is that statistics aren't always open to everybody. I think this is something that puts people off, and I think that's something that will put people off quantitative research as well. I counter that with it is easier now to do it because you can buy software that you can put your data through, and it will tell you the standard deviation. It will tell you whatever you need to know about it, and confidence intervals, et cetera, et cetera. So it is more open to that. Good.
And as I said, yeah, others are going to be more likely to be able to have a look at it and be able to use it as well. A downside I would say from the distance point of view is that I'd say that you need to know the context of data to get its true meaning. So you could look at our four-hour target in December and January and go, wow, that was a really bad A&E. Your guys' four-hour target was bad. But unless you were actually on the ground floor and seeing what it was like in our department between December and January, you're not going to get that idea, are you? And you'd be able to, you know, you might just publish data going, yeah, Queen's A&E is really bad in December and January. Don't go there, exactly. Whereas actually, unless you're in the ground floor and seeing it, you're not going to be able to do that, are you? Good, cool.
Quality Street. Really good name, I like that name. So, false friends, we got that qualitative data was reliable and it's also quite cheap to run because usually you only need pen and pencil and paper. Or a dictaphone maybe. And also that you're getting the opinions of real people on a sheet basically. Yep, good.
Weaknesses, we got that it was quite subjective. Yep, very. It's not very generalizable because it depends who you get on today. Yep. The data's harder to prevent. Yes. As well. And also, depending on the way that the questions are written slash asked, they could be leading questions or they could be easier to have an element of bias involved. True. For the trial design, we said it was observational. Yep. Trial.
Do you want to come in and do some research? Okay, friend. For qualitative data, it's quite common to have focus groups, so people where everyone out of ideas are. It is indeed. The investigator would be very involved in potentially carrying out the questionnaires and seeing the participants face to face, perhaps more than once. The goal tends to be to gain opinions of people. Some guys can be a smaller number than people using quantitative data.
because you don't need as many people to make it. It's probably not going to be as statistically reliable. So your variable depends who you get on the day. Basically, you're going to have far more uncontrollable variables. Of course you are, yeah, absolutely. And others. Anything. There's no others. There's no others. There are no others. Okay, cool. Okay, so qualitative data is going to involve a smaller sample size.
Because you could do a bigger sample, but it's going to take you forever. You are, like you said, you're going to be embedded in that culture, in that environment, because we're interested in the subjects in their natural setting. We want to see those research nurses at work eating the sweets and talking to them then. We don't want to just be following something off the screen just a few miles away, etc. Okay, good. Variables, yeah, like you said, could be depending on what it is really. And you're not, hello.
Sorry, just another bit. No, go for it, Lucy. Yeah. So if you're doing a question interview study out to the Home Trust because you want to find out what their thoughts are on a certain structure. So how main interview is and what to do. You'll get a certain type of person that will be really eager to go to and be a part of the study. You may not then meet everybody that has thoughts and opinions because there'll only be a certain type of person who will sign up for a study and give their time. You can't really say...
this will reflect the opinions of everybody working here because it's not so emotional, like dysfunctional people, people quite sort of So you have to think about that in all of it, don't you? And it does come into your sampling. Do you use purposive sampling? So I want the opinion only of these people, or is it anybody who writes in, who, like you said, maybe somebody who has an axe to grind, maybe somebody with certain opinions, etc. It might be a certain demographic of the population who are more likely to write in and want to talk to you rather than being truly representative. Absolutely.
And as you said, in the questions that you're asking, you do have to be very, very careful. And you do have to have an evidence base as to why I'm asking these questions. You can use semi-structured interviews. So I do have some key questions, but I'm going to deviate and allow you to elaborate on any points that you've got coming along. But yeah, variables are very difficult to predict. You can do a bit of a literature review and say these factors have been shown previously to affect opinions. But ultimately, it's going to be whatever comes up in the text and in the course of the interviews. Good. And as I already said, you're into the natural setting, so you get that really good involvement. But it takes a lot of time. And it does take a lot of time to do it properly.
Okay. Well, exactly. That's different techniques, isn't it? They may be better or worse interviewers. Yeah. Okay. And if you're doing a focus group, somebody very early on might go, oh, I really like this product. And then everybody else feels they have to say the same thing because that person said it. So if that's something else, you've got to think about some focus groups. Good. Is it easy to do qualitative analysis?
depends to what level but yeah I think it's harder than people think I think if you truly want to do proper analysis sometimes you have to be incredibly sub-specialised you know you might have to send your data off to somebody over at the school of English who's really involved in linguistics to go to break it down into this is the language that's being used you know that somebody from a more medical background wouldn't be able to do so that's something else you have to think about that's another process that you're adding in another human being that may take more time I haven't no no I haven't no I played with nudist nudist nudist That's all how y'all think about it.
But our sample size is going to be big. Our sample size is going to be small, sorry, but our focus is going to be very broad and we want to know everything. Tell me everything behind why you've chosen these suites, why you've chosen this choice of ice cream. Brilliant. Presenting that data is going to be harder, as we said, and then it is going to be harder for somebody else in another site to go, this is representative of my population. How do you spell Nudist? Nudist qualitative data analysis. It is called the Nudist. Nvivo is another one.
But it does give us that greater overall idea of what's going on and you may want to use a mixed method approach because you want to maybe see what that four hour target was like in January and December, but you're also going to want to know what people felt like at the time, what was it actually like, you might want to interview people. Brilliant. And as I said, the researcher is right in there. They have to be in that environment. The other limitation I would say from an educator point of view is that can you always trust what your subject is saying? So I will put on a teaching session for students and they will always say afterwards, I feel more confident in X. You know, I'm pleased that this happened. And I can put that down. Look at this great piece of teaching I did. But does it actually mean they've learned anything? You know, so we have to be very careful with confidence. These are bad, you know, confidence, attitude, et cetera. It's very, very difficult to prove when you're doing these things. And you have to bear in mind, you know,
Somebody might say, oh, I buy this ice cream because, you know, of blur. Is their opinion actually true? These are all things that you hold yourself open to. Cool. Well done, guys. As ever, this will be available through TakeOri.com and SoundCloud and iTunes. Thank you very much. Next month, we're going to be looking at sampling. Who is the research nurse for next month? High fives, Cecilia. Go on, then. So... So...
Again, as I said at the beginning, some people are going to prefer qualitative, some people are going to prefer quantitative. Hopefully we've got a bit more of an idea of both of them and a bit of appreciation because it's important not to narrow ourselves down when it comes to research. So hopefully we'll be able to sample more than qualitative research and not always drawing the line at quantitative research. It is very hard to make a pun out of qualitative and quantitative research. Thank you very much for coming, guys. You've been a great audience as ever. Thank you.
Thank you. That was the live recording of Unblinding Research, qualitative versus quantitative data. You can find the full blog entry, including the tape visually for this podcast at TakeHority.com. Remember, you can subscribe to TakeHority on both SoundCloud and iTunes, and find TakeHority on both Facebook and Twitter. For more information about research and education opportunities for the emergency medicine and acute medicine and major trauma, you can find NUH Dream on both Facebook and Twitter.