wendystjohn

Knowing what statistical test to perform on your data is a critical first step in any study. Below is a simple dichotomous key that will help you make decisions about the type of analysis that is required.

1aDependent variable continuous ……….go to 2
1bDependent variable categorical ……….
(e.g., yes/no, M/F, success/failure, respond/don’t respond)		go to 4
2aIndependent variable present ……….go to 3
2bindependent variable lacking …………
(i.e., second variable also dependent)		Correlation Analysis
3aIndependent variable continuous ……….Regression Analysis
3bIndependent variable categorical ……….
(e.g., present/absent, high/low)		T-test or ANOVA (Analysis of Variance)
4aIndependent variable present ……….go to 5
4bNo Independent variable ……….Distribution analysis
5aIndependent variable continuous ……….Logistic Regression
5bIndependent variable categorical ……….
(e.g., present/absent, high/low)		Contingency Table Analysis (chi-squared tests)

Practice Exercises

In this activity, you will join some of the scientists of Jurassic Park as they seek to better understand the dinosaurs under their care, and some of the factors that influence their health and growth. For all exercises, provide the following information in your lab notebook:

  • Type of variables
    • Independent or dependent?
  • Type of data
    • Independent variable(s): continuous or categorical?
    • Dependent variable(s): continuous or categorical?
  • Identify Null and Alternative hypotheses
  • Use the dichotomous key to determine the appropriate statistical test
    • Track your choices at each stage of the process
  • A screenshot of the final step in your analysis, that shows the p-value you obtained

To import a data table into JMP:

  • Copy the data from Excel
  • In JMP, “File” —> “New” —> “New Data Table”
  • “Edit” —> “Paste with Column Names”

Exercise 1:

Dr. Sattler wants to determine if the sex-ratio of adult Velociraptors on Isla Nublar is 1:1. This seems sensible if male and females have about the same chances of being born and surviving to adulthood. It is impossible to census every animal on the island, so she conducts an aerial survey of adult Velociraptors. She counts 135 males and 165 females. The sex ratio of the animals in her sample is clearly different from 1:1, but can she infer that this difference is statistically significant?

  • In JMP:
    • “Analyze” —> “Distribution”
      • Select correct variable(s) and click “Y, Columns”
    • You may want to change the orientation of the graph: “Distributions” —> “Stack”
    • Now, test the distribution
      • “Sex” —> “Test Probabilities”
    • Add Hypothesized probabilities
    • Select an appropriate alternative hypothesis

Exercise 2:

While Dr. Sattler was performing the census of Velociraptors, she noticed that the males and females do not seem to be distributed evenly across the island. She decided to do a study on this, to see if males and females really do have different habitat preferences. On the beach, she observed 73 male Velociraptors and 66 females. In the jungle in the central portion of the island, she observed 56 male Velociraptors, and 69 females. In the alley next to the cafeteria, she observed 6 male Velociraptors, and 30 females. (She made a note to mention that people coming to and from lunch should probably avoid this alley).

  • In JMP
    • “Analyze” —> “Fit Y by X”
    • Select correct dependent variable(s) and click “Y, Columns”
    • Select correct independent variable(s) and click “X, Factor”

Exercise 3:

On this tropical island, the weather can be unpredictable (chaotic, even? :D). In order to better understand the patterns of weather, Dr. Malcolm analyzed weather data from the previous year, to see if temperature and precipitation appear to be correlated.

  • In JMP
    • “Analyze” —> “Fit Y by X”
    • Select correct dependent variable(s) and click “Y, Columns”
    • Select correct independent variable(s) and click “X, Factor”

Exercise 4:

Dr. Grant is interested in learning more about the effects of different diets on dinosaur weight. So, he conducts a study in which 20 Tyrannosaurus rex specimens of equal initial weight are assigned at random to one of four diets. The Tyrannosaurs are fed their respective diets for 10 months and then weighed. Assume that the data on Tyrannosaur weights are distributed normally.

  • To perform a calculation that creates a new column in JMP:
    • “Cols” —> “New Column”
    • Give a meaningful “Column Name”
    • “Column Properties” —> “Formula”
    • Use the Formula feature to make the desired calculation (see below)
    • “OK”
  • In JMP
    • Analyze —> “Fit Y by X”
    • Select independent variable(s) and Click “X, Factor” 
    • Select dependent variable(s) and Click “Y, Response”
    • Click red triangle —> “Means, Anova”

OR

  • In JMP
    • Analyze —> Fit Model
    • Select independent variable(s) and “Add” to “Construct Model Effects.” 
    • Select dependent variable(s) and Click “Y”
    • No changes to model settings are needed
    • “Run” model
  • Exercise 5:

Generate brief descriptions of studies that would lead someone using the Statistical Key to conclude that their data should be analyzed using each of the following tests. Your examples must include the study system involved, the hypothesis being tested, and appropriate independent and dependent variable(s).

  • Correlation analysis
  • Logistic regression
  • Contingency table analysis
  • Distribution analysis
  • Regression analysis
  • ANOVA

For JUST ONE of the scenarios you have described,