If you want to analyze your own data or need to work in a research team that uses IBM's SPSS software, then this course is for you.
From the import of data, through descriptive statistics, data visualization, correlation, the comparison of means, and the analysis of categorical variables, this course will leave you familiar with the user interface and able to conduct all of the most common statistical tests.
A warm welcome from me to this course on using SPSS for healthcare and life science statistics. I hope to empower you to do your own statistical analysis or be comfortable working in a team that uses SPSS, still the most widely used statistical software package for healthcare in the world.
In this section we will take a look at importing data and we will conduct a variety of descriptive statistical analysis.
In the analysis of data, the first step is to calculate measures of central tendency and dispersion. The mean, median, and mode are common point estimates (measures of central tendency). Standard deviation, variance, range, and quantiles are common measures of dispersion.
These values summarize our data and makes it easier to interpret.
In this section we calculate common measures of central tendency and dispersion. These include, mean, median, and standard deviation among others.
This section is all about learning the details of your dataset through visualization.
The visualization of data is another important first step in analyzing data. As humans, we get a better understanding of our data when it is visualized. It is indeed much better than staring at columns of numbers.
SPSS can generate almost all of the graphs, plots, and figures that you will ever need. While the default settings are not the most pleasing, rest assured that you can change all the aspect of your graphs.
In this video you will see the dataset that will be used in this course. It should easily generalize to your unique research area and contains a good mix of nominal, ordinal, and numerical variables.
Histograms are frequency charts. It divides the range of data points values for a numerical variable into equally-sized bins and counts the occurrence of values in each of these bins.
If you want to fit a histogram of a dependent variable for more than one independent variable, a stacked histogram is the way to go.
Another way of comparing the histograms for a variable between two groups, is the population pyramid. It creates two horizontal histograms.
A bar chart can be used to count the instances of unique data point values for a categorical variable.
Box-and-whisker charts are the most commonly used figures in the healthcare literature. They convey useful information.
A scatter chart plots pairs of values. It is the preferred method at investigation the correlation between two variables.
SPSS can create scatter plots for a pair of variables for more than one independent group.
Instead of pie charts, I recommend frequency tables. In this video I show you how to create them.
Are two numerical variables from each research subject somehow connected? Does a change in one lead to a change in the other? This section looks at correlation.
I quick refresher on correlation.
The main assumptions for the use of Pearson's coefficient correlation include:
In this video we look at how to conduct a correlation test when the assumptions in the previous video are met.
The meat and potatoes of the course as my grandmother would have said. The various t-tests, including Student's t-test and analysis of variance are the most common tests that you will conduct and see in the literature.
In this video, we discuss the comparison of mean values between two or more groups.
Student's t-test is one of the most common statistical tests in the literature. The t-tests compares the mean values for a numerical variable between two groups.
A variety of assumptions must be met before we can use these tests.
In this video we take a look at how to test for the assumptions for the use of t-tests and how to compare them.
The paired samples t-test compares the means for the same variable, measured twice in the sample sample subject.
In this video we take a look at how to calculate a new variable that measures the difference between each pair of data point values. We then look at the assumptions for this test before actually conducting it.
Fisher's test allows us to calculate the difference in means between more than two groups. It has similar assumptions at the t-tests.
In this video we use analysis of variance two compare the means of more than two groups.
This section leaves numerical variables behind and looks at the common tests for categorical variables.
Now that we have seen the most common tests for numerical variables, let's take a look at the tests for categorical variables.
Let's have a look at conducting the chi-squared tests for independence and Fisher's exact test in SPSS.
Thanks! I really enjoyed putting this course together. Not only is it available to everyone in the world, but it is also for official use in my own Department.
I am a Senior Lecturer in Surgery and the Head of both Postgraduate Surgical Research and Surgical Education at the University of Cape Town, South Africa. My academic interests extend to online education and I am the recipient of the Open Education Consortium Educator of the Year Award in 2014. My course on Healthcare Statistics is also the first course from a University in Africa on the massive open online Coursera platform.