[0:13]So in this last module, we're going to talk about writing the abstract. This is going to be kind of a short module because it should be fairly easy to write your abstract after you've written your paper. In fact, the abstract, what it means that word means, is literally to pull out. So what you're going to do to write your abstract is to pull out little sections from the rest of your paper.
[0:41]So it's just going to kind of pull out a little bit from your introduction, pull out a little bit from your results, pull out a little bit from your methods, etc. So if you're doing the abstract if you're writing it last, it actually should be fairly easy to do. So what does the abstract do? It gives you an overview of the main story of your paper, obviously it's what a lot of people will read, and so you need to get all the key points, as many as the key points as you can in there. It uh highlights something from each section of the paper. And it's usually of a limited length, sometimes 100 to 300 words, so this is a great place where you can use all those skills about cutting clutter to get into that small word count, the really key important information and cut out all the words that are really unnecessary. The abstract, like the tables and figures, has to stand on its own, so people often times will only have access to the abstract. So it has to be its own unit, you can't expect that the readers will be able to read any of the full text. So it has to stand on its own, and it's often used along with the title for searching, like for searching PubMeds and things PubMeds and things. So make sure you're getting all the keywords in between the title and the abstract and the keyword list, which they'll uh journals usually allow you to give. Make sure you're getting all the keywords that you want so that people who are searching in your particular topic will find your paper. And unfortunately, most often the abstract is the only part that people read. That means that you need to convey the main finding of your paper, the main take-home message of your paper, and and also why your paper is important, you have to convey all that in a, you know, a very small space. It but that has to be in there because sometimes that's all people will read. I'm often scanning abstracts, I will scan PubMed for things that I want to write about for a general audience and I'm often scanning abstracts. And if I get to an abstract and it does doesn't look like it's that important or interesting, I might not end up writing about it. So you got to say in the abstract why your research is important and interesting, why should people care? So, um Mimi Zeiger in her book, uh, I've mentioned her book before, kind of has a nice uh little summary of what should go in the abstract. Again, it's essentially just pulling out little bits from all the rest of your paper. So there's it it's good to have uh a one sentence statement of some background. Just to give a little bit of context because again if somebody's just jumping into your abstract, you have to give them a little bit of context about where you're coming from, why is this research question even important. So something to ground them in what the research question is. And then you state explicitly that research question, just like you did at the end of the introduction section, you're going to use that exact kind of phraseology. We asked whether, we hypothesized that, we speculated that. So you're going to very clearly state what was the question that you were trying to answer or the aim or the hypothesis of your study and make it easier for the reader to find. So those two, the background and that question statement come essentially right out of the introduction. Then you're going to give a quick summary of the experiments that you did. This is obviously you're not going to have room for a lot of details but all the main, you know, key points about what major experiment uh, you know, what the key points of your experiment are, so uh without a lot of details. But you need to say some of the key things like was it a randomized trial or an observational study, you know, what if you did a study in animals, what type of animals, things like that. So quick statement of the experiments run, and then uh that's just you're pulling that a little bit out of the method section, and then you're pull something out of the result section and then it's going to be very uh minimal. You're going to give a couple of key results and you're not going to be able to cram a lot of numbers in there. So summarize everything, maybe give one or two key numbers like the main effect size or the main P value. And then you're going to have a often usually a one statement conclusion. You're going to answer that the question that you asked or the hypothesis that you were testing in your study, what did you find? Was your hypothesis proven or not? What was the answer to your question? So give an answer to that question and some kind of take home message so that you're actually saying, well what did your data show relative to that question? So some kind of take home message about whether or not your hypothesis was proven, is does breast cancer, is it linked with smoking in in your study? And then uh, there's one more thing that I think is very important that you put in. Uh, you need to have, as Mimi Zeiger puts it, some kind of implication, speculation, or recommendation. So what that means is that there's one sentence at the end of your abstract where you're going a little bit beyond and to giving people that, why should I care again? What's the implication? Why does this research matter? So give something a little bit more big picture. Again, I'm scanning abstracts a lot of times and I need that to really be able to say, well why should I care about this paper? Have that little last thing in there for the reader to really understand why your research is important. Now abstracts come in basically two forms. They may either be structured where the journal says, hey, we want you to have something for each of these pieces.
[5:46]Uh with subheadings like introduction, results, methods, or the subheadings are fairly obvious. Uh some journals will have more subheadings and more categories than others. Or you may have a free form abstract. So you can see both kinds. The structured ones are obviously a lot easier to write. So here's an example of an abstract. This was uh looking at influenza virus and uh a potential vaccine, they're trying to create a vaccine for um avian flu. And so uh so here's the background. So we know that avian flu is a bad thing and people are worried about it and they're trying to create um some kind of vaccine. And we don't really have a good vaccine yet. And then notice here's in our we get so we kind of got the background and the context and now we get the question asked. That's that very explicit statement, we aimed to develop an influenza vaccine and assess its immunogenicity and efficacy to confer protection in mice. So this is a mouse study. Then they give a quick summary of the methods that they did to to create the vaccine and test it in mice. And then they give a quick summary of the results. So they give you a couple of key pieces here. So first of all, the the vaccine seemed to work at least in the mice, it provided um effective protection from disease and death and primary viral replication and notice a highly significant P value there. Uh they they just put the P value, they didn't give you any other numbers, but it's a, you know, that's probably uh enough. And then they, you know, tell you something, uh well then they get to and they tell you that the vaccine induced a three-fold to eight-fold increase in um CD8 cells, so in your immune cells, so they actually give you an effect size for that. This is a big effect on your immune cells and then they give you one more P value. So that's the main results, notice they gave you only a couple of numbers in there. Then they give you the answer to the question that you asked, this is the conclusion. Our findings highlight the potential of this particular vaccine delivery system. Um to so it highlights the potential of it, it's not saying it works because we haven't tested it in humans, but it seems to work in mice, so it has some potential. The wider implication is that, hey, it offers stockpiling options, so in case there's a pandemic flu outbreak, uh this is a type of vaccine that we could actually stockpile. So that has some nice public health implications, so they snuck that in at the end to give you that, why is this study important to make sure they got that in there.
[8:44]So then finally, I'm just going to show you one example of an unstructured outline uh abstract. The um it's obviously easier to write the structured ones, uh some journals uh more most journals have the structured kind but some journals like Science actually have unstructured. So you got to be, you know, you have to figure it out on your own, it's not quite so easy. Um but reading through this one from Science, empirical research with non-human primates appears to support the view that causal reasoning is a key cognitive faculty that divides humans from animals. The claim is that animals approximate causal learning using associative processes. The present results cast doubt on that conclusion. So that's a nice, right, that's very intriguing, it's like, oh wow, that these results cast doubt on that. So they don't give you any of the details of that but they kind of give you that punchline right there and it makes you want to read on and find out why. So then uh then they give you the details of both the experiment and the results kind of rolled into one. Sometimes you can roll results and experiments results and methods together like this. Rats made causal inferences in a basic task that taps into core features of causal reasoning without requiring complex physical knowledge. And um they derive predictions of the outcomes, so they're giving you some background, they more details about the study, they derive predictions of the outcomes of interventions after passive observational learning of different kinds of causal models. And these competencies cannot be explained by current associative theories but are consistent with causal Bayes net theories. They don't tell you what that is, but uh assuming that uh you might know what that is. So you can see that they um have kind of give the big implication and uh the the why you should care kind of in the middle of this one, but it is there. And so they got all kind of the key points in a very short space. Now, the one thing I want to add on the abstract, so you have the uh besides the fact that you have structured and unstructured, um one thing that I'd recommend on the abstract, as I I've said, I'm recommending that you write it last. So you write your research paper and then it's very easy to write the abstract because you can literally kind of just pull sentences out of the rest of the paper. Of course, you're going to have to rewrite them so they're not exactly what appears elsewhere in the paper, but you can start by just kind of pulling information out of the rest of the paper. Um that's a very good way to write the abstract. What often happens with the abstract, the one caution I'll give you is that there's a lot of times when people write an abstract prior to writing the paper because they want to present the abstract at a conference. And you know, a lot of times you've just gotten data and you kind of rummage it together quickly and throw together some abstracts so that you can present it at a conference or poster presentation. So people will sometimes do that. And the one mistake they make is that months later when they actually go to write up the paper, they're like, oh, I already wrote the abstract. Great, I can just plug that abstract in and I'm going to start my whole paper there. So the problem with that is when you threw that abstract together, you probably hadn't nailed down all of the uh data analysis and the tables and figures. And it it was probably very preliminary. And if you start with what the abstract that you've already written, it's going to be hard to get it right. Because um, you know, that that was in the past, that was in a prior analysis before you really looked carefully at the data and things. So my idea is just throw out that old abstract. Don't, I mean it feels like, oh, I've already got it written, I should use it. No, just throw it out, write your paper and then just rewrite the abstract, start over on the abstract using the material in your paper and again it's really easy to write an abstract after you've written the paper, so I would just do it uh in that order.
[13:58]The preceding program is copyrighted by the Board of Trustees of the Leland Stanford Junior University. Please visit us at med.stanford.edu.
