Chi-Square Test
A Step by Step Guide to the Chi-Square Test
Chi-Square Test
The chi-squared test is a statistical hypothesis test that provides a quantitative method to compare observed frequencies with expected frequencies. For example, we could use a chi-squared test to evaluate if a coin is likely biased.
The general strategy to perform a chi-squared test is to calculate the chi-squared value (or chi-squared statistic), and then use a table of chi-squared values to estimate the probability that our result is due to chance alone. This probability is called the p-value. The null hypothesis, i.e., that the difference between expected and observed frequencies isn’t significant, is typically accepted for p-values greater than .05.
Goodness of fit
We’ll use a chi-squared test to determine if a certain coin is biased. Suppose we flip a coin 100 times and the coin lands on “heads” 42 times. We would expect that an unbiased coin would land on heads half of the time. Using a chi-squared test gives us a concrete way to compare what we observe with what we expect.
Here’s how you can use Plotly to calculate the chi-squared statistic and p-value for you.
Step 1: Set up the grid
We’ve already entered the coin flip data into Plotly. Click the link to open the data in your workspace. |
Step 2: Chi-squared test
Once the data has opened in your workspace, select Chi-squared test from the ANALYSIS menu. | |
Select choose as obs in the column labeled “Observed”, and choose as exp in the column labeled “Expected”. Click on the blue Perform Chi-squared Test button in the sidebar. | |
Plotly puts the results in the first two columns to the right of our data. In this case, the chi-squared statistic is 2.56, and the P-value is 0.1096. By convention, because 0.1096 is greater than 0.05, we decide that the difference is not statistically significant. In other words, even though we didn’t observe exactly 50 heads and 50 tails in our coin toss, the distribution that we did see is likely due to chance. |