Exercise 11C: Comparing ordered groups in Prism

Go to parent GraphPad Prism statistical analyses

When the groups are defined by a quantitative factor e.g. dose, time so that they can be arranged in an order, you can test whether there is a trend in the data values. For instance, the data values increase when the dose increases. If you want to perform this test, you have to place the columns in the correct order, perform a one-way ANOVA and choose test for linear trend as a follow-up test.

I measured the bacterial load (expressed in log₁₀CFU) of Candida albicans in biofilms after administration of a potential fungicide. I used a 4-four dilution of the fungicide where 1 is the lowest and 256 is the highest dose.

The data are available as a csv file. The comma is used as a decimal separator.

Open the file in WordPad. The first row contains the column headers (dose of fungicide that was administered). Prism doesn't know that the first row of numbers in the csv file do not represent data but instead represent column headers. So you will have to tell Prism that.

Import the file ?
You see that the Import and Paste Special Choices window contains many tabs. In the View tab you can view the data: You can see that Prism indeed considers the headers as data values. In the Filter tab you can define the symbol that was used to denote missing values (will be used in later exercises) In the Placement tab you can define the row that contain the headers:

Is there a linear trend in the data ?

Explanation of the test for linear trend Perform One-way ANOVA on the resulting column table In the Multiple comparisons tab select Test for linear trend between column means and left-to-right column order Click OK In this test Prism performs linear regression on column mean versus column number. In the Multiple comparisons results table Prism reports: the slope of the regression line R squared: the regression coefficient the p-value for the linear trend: H0 of this test states that there is no linear trend The p-value is below the confidence level, which indicates that there is indeed a linear trend between the data and the dose. The slope is negative meaning that there is a negative linear trend: the bacterial load decreases when the dose increases. Explanation on interpreting the results of a test for a linear trend