Exercise 9B: Comparing the mean to a hypothetical value in Prism
Go to parent GraphPad Prism statistical analyses
You are the scientist who discovered that stevia plants can be used to produce sugar substitutes. To launch your product on the market, you are studying all the good sides of stevia glycosides compared to traditional sucrose. Apart from the health benefits, you notice that wasps do not seem to be attracted by stevia glycosides. You carry out an experiment in which you use a cage containing a box of sucrose and a box of stevia glycosides. You release wasps into the cage and after half an hour you count the number of wasps on each box. You want to know if the wasps have a preference for one of the sugars.
This is a typical example of a binomial test. You will compare the proportion of wasps on the box of stevia glycosides to the expected proportion, being 0.5: if there is no preference, half of the wasps will be on the sucrose box and half of them will be on the stevia box.
You do the experiment on 136 wasps and you find a total of 37 wasps on the box containing stevia glycosides.
Enter the data in Prism. Note that Prism's statistics guide advises to use actual numbers and not percentages or fractions
Which table type should you use for these data? |
---|
Data that make most sense when transformed into proportions should be enterd in a parts-of-whole table. |
Enter the data in Prism |
---|
|
Check if the wasps have a preference for a certain sugar or not.
What is the null hypothesis of this test ? |
---|
The null hypothesis is that the observed proportion equals the expected proportion: there is no preference, half of the wasps choose the stevia box. |
Do the wasps have a preference ? |
---|
In the Parameters window make the following selections:
The Result table shows that p < 0,0001 : you may reject the null hypothesis. The wasps do have a preference for sucrose.
Prism reports both one- and two-sided P-values. |