Skip main navigation

New offer! Get 30% off your first 2 months of Unlimited Monthly. Start your subscription for just £35.99 £24.99. New subscribers only T&Cs apply

Find out more

Steps to Fill Out the Statistical Table

Steps to Fill Out the Statistical Table
All right. So here I just took a little bit snippet of that chart, so you can see how we’re going to use this and then fill out that big table that we had. So you can see I just just took a clump from the the interval/ratio and the ordinal section to help show you what I’m talking about. So the first thing you’re going to do remember is look to see if the tests used independent or dependent design, either parallel, versus crossover, pretest-post-test is another form of dependent. And then again remember in one in one study you can have both. But in this case it’s just independent for the example that we that I provided to you.
Then you want to look at the data type, was it two groups or two or more groups? And in this case it was just two groups. And then with this one then the third one you’re looking at the data types that they had. Now, remember I told you that if they have interval/ratio, you can also use ordinal test. But this in this example I’m just giving you the fact that we’re going to look at the ordinal data, so we can use Mann-Whitney or Wilcoxon rank-sum. But remember I also told you that if a study could handle three groups that also could handle two.
So when you’re making your chart you need to make sure in the section what could be used, because you’re going to want to check what could be used versus what was used to make sure that they’re correct. So you’d want to put both all three of those tests in the what could be used chart and then if it was dependent, you’d pick it just from the other side of the menu. So you can see for this that all three of those statistical tests for ordinal or interval/ratio data would be appropriate for two groups. All right. So here’s the big one of these then let’s go through and fill this out for the sample that we had.
So you can see at the very top it says N greater than 30 in each group and you might remember that the reason we’re concerned about that is we want to make sure that we can select a parametric statistical test or not and we know they also did a test for normality. So yes, they did have more than 30 for each group but we’re a little concerned about the normality since they were listing their data as both median and mean. And then you can see we already talked about it was an independent test, the number of groups was 2 and then remember a number of comparisons why are we concerned about that?
Well, remember if they use the t test, it can’t be it can only be used for one I/R comparison. And if they use it for more than one our I/R there is a Bonferroni rule that you can adjust that you could maybe still use the t test but you can’t use it for more than six. So when I fill out this I usually say either grade at less than six or greater than six and then if they don’t use a Bonferroni you would have to then calculate that yourself and there are resources that can show you how to do that.
So then you’re going to go through and take all of those data points that you determine what they were and put them in the correct box. And then you’re going to you can either list the test that they used, sometimes I like to list the all possible tests before I start getting talked into what they used but whichever way you want to do it is fine.
So if you want to list what they used, here’s what they said, and again if you look at the the paragraph in the methods section under the statistical test, that it’s a little bit different than what you saw in the legend so you’re gonna have to list them both or take it with a greater Sall to figure out what they did. So they used it as remember in the in the chart you saw that they had mean mean and median and you can see what they used for those. And then you can see what they use they said they were going to use for the nominal. So we’re going to list all possible.
Remember I showed you how to use that chart and then figure out what could possibly be used. And you just need to make sure that the test that they use is in your possible list. And same thing true for the ordinal. This is what we already went through that example that these are the three that you could use and then this one would be for the nonparametric. So you can see it’s fairly easy at the bottom if you fill the chart out correctly were the correct statistical test used for the parametric. Well, yes and no because we know t test isn’t a good choice because they’re more than six comparisons.
Even if we use that Bonferroni correction it wouldn’t have been good. Now unfortunately, even though they should, statisticians and articles don’t tell you which tests were often used exactly in which data they did for a couple of them. So if they say they use the t test, you have to assume potentially they used it incorrectly for all of them unless they differentiated. So then that would negate some of those p-values. And you can see that they used parametric statistics for ordinal data which is not considered correct, so that box should definitely hit no that wasn’t correct and then for nominal you see that what they did was correct. Now remember the primary outcomes were vomiting episodes and VNRS.
So you can see we potentially got a problem here, because they possibly use the incorrect statistics for those.

Prof. Mary Ferrill demonstrates the steps of statistical analysis.

There are three main steps to use the chart.

Firstly, we have to determine the test is dependent or not. Then, we can check the number of groups. Finally, find the data types are interval/ratio or ordinal.After that, we can fill out the table.

Check whether N is greater than 30, make sure its number of groups and comparisons, and list all study outcomes. Following that, we should list statistical tests used in the study, and all possible appropriate tests.

Ultimately, determine whether the statistics used in this study is appropriate or not.

Which kind of conclusions can be drawn from this process? Are those statistics appropriate?

Please share your answers and thoughts below.

This article is from the free online

Evidence-Based Medicine in Clinical Pharmacy Practice

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now