Want to keep learning?

This content is taken from the Taipei Medical University's online course, Evidence-Based Medicine in Clinical Pharmacy Practice. Join the course to learn more.

Skip to 0 minutes and 14 seconds So as I mentioned, we’re going to look at several different places to find all of the data types in there. Again we mention this because you want to make sure that you have they have all the statistical tests that could analyze data. Because as you can see in this baseline, you can see that there’s a p-value for each and every one of these. You want to make sure that they’re they had a test that would be able to do it. So you can see on the right what I did was I took each one of them and I rank them as what they would be either interval/ratio, ordinal or nominal.

Skip to 0 minutes and 44 seconds So you can see from this point for the most part it looks like, as I mentioned, medicine, clinical medicine mostly is going to be interval/ratio data, but occasionally you’ll see ordinal as well as nominal. It that would be in there so ethnicity is always nominal data, because you can only be one ethnicity at least according to a scale that’s there, you’re employed yes or no. So you just need to practice in there to determine to make sure you can put them all in each one of the boxes as we put those in there. And then you’re also going to want to make sure you look at your outcomes table.

Skip to 1 minute and 18 seconds Now you can see in this one, you always want to concentrate on the primary outcome and remember there were two in this case. And so it’s very interesting that they listed vomiting episodes which should be interval/ratio data as median. Now that should put up a red flag for you. Because it means that if you listed this median, mean median mode, going back to your statistical days, hopefully you remember this from learning about basic statistics, is that if your data is normally distributed, median mean and mode should all be the same number.

Skip to 1 minute and 48 seconds So they listed this median that should give you an indication that the data is not normally distributed and you should use something other than a parametric test to analyze that. So that’s a key to look to this. Then also the VNRS which is an ordinal because remember its scales, rankings, ratings, subjective data, it should be listed as median and they listed it as median and mean. So at this point I’m getting a little concerned that maybe they don’t quite understand which test you use for which type of data. And we’ll see that as we go along. So you can see now that we have each of the three data types We have interval/ratio, we have ordinal and we have nominal.

Skip to 2 minutes and 25 seconds So we need at least two different statistical tests to cover these. And why do I say two? Well, remember that you can use ordinal data on interval/ratio. So you could actually the investigators could actually just have one of those. And it’s actually safe because you need to select the statistical tests a priori so they don’t know it’s gonna be normally distributed. So it might be a good idea to just pick an ordinal test and then also have one for nominal so they could get away with just two statistical tests in this study.

Skip to 2 minutes and 55 seconds Now, you also then need to, as you can tell from the the table that we’re gonna fill out, is you also need to go look to see what statistical test they used. Now if you look at this, it should make again some red flags. Because I told you that they probably could just use two statistical tests in this study. And just from this you can see they used one, two, three, four, five, six. Well, I counted twice. Five, that they used. Now that’s okay. Because these are the standard ones. Because one of the things you need to make sure of is that if they use a test that’s not listed in the groups that I gave you.

Skip to 3 minutes and 32 seconds They should provide you a rate ready reference for you to look at it, to make sure that it was used appropriately. Because there’s no reason to use an obscure statistical test when there’s these standards that are used. Now something else you also need to do is look here. But then you’re also going to want to look in the legends of tables charts and graphs, because often that information may not be listed here or it may be listed difference different in those charts and graphs. So you have to kind of go “Where’s Waldo” and start looking different places to make sure that you’ve found all of that information because it may be listed differently in different places.

Skip to 4 minutes and 10 seconds And that happened to be the case with this article. If this is the legend from Table two, and you can see that area that I highlighted which I mentioned that normality test they could do they did do one for listed for the KSR normality test. So they possibly but they didn’t say what data they were listed as normally distributed because remember they listed things as mean and median and they weren’t the same numbers.

Skip to 4 minutes and 36 seconds And then you look on there and they said they use different tests than what they had used in the paragraph and that will be displayed then when we make our charts to make sure that we may not be able to clearly determine what they used, if this information is different than what was in the written portion of the article.

Results of Assessing Baseline Demographics

Prof. Mary Ferrill illustrates the results of the baseline Demographics in this video.

Continuing from the previous video, we need to find the correct data types this time.

Mostly, clinical medicine analyses belong to interval/ratio data. However, you’ll see ordinal data as well as nominal occasionally.

Besides, please notice that there were two primary outcomes in this case.

An important tip is that, if the data is normally distributed, the median, mean and mode should all be the same number. In other words, this data is not normally distributed, so we should use something other than a parametric test to analyze it.

Finally, we have all three data types: interval/ratio, ordinal, and nominal, and we can use ordinal data on interval/ratio. Consequently, we can pick an ordinal test and an nominal one, so they could get away with just two statistical tests in this study.

If you have any questions or thoughts, please share them below.

Share this video:

This video is from the free online course:

Evidence-Based Medicine in Clinical Pharmacy Practice

Taipei Medical University