A Chi-Square Goodness of Fit Test is used to determine whether or not a categorical variable follows a hypothesized distribution.
This tutorial explains how to perform a Chi-Square Goodness of Fit Test in Stata.
Example: Chi-Square Goodness of Fit Test in Stata
To illustrate how to perform this test, we will use a dataset called nlsw88, which contains information about labor statistics for women in the U.S. in 1988.
Use the following steps to perform a Chi-Square Goodness of Fit test to determine if the true distribution of race in this dataset is as follows: 70% White, 20% Black, 10% Other.
Step 1: Load and view the raw data.
First, we will load the data by typing in the following command:
sysuse nlsw88
We can view the raw data by typing in the following command:
br
Each line displays information for an individual including their age, race, marital status, education level, and a variety of other factors.
Step 2: Load the goodness of fit package.
To perform a Goodness of Fit Test, we will need to install the csgof package. We can do so by typing in the following command:
findit csgof
A new window will pop up. Click the link that says csgof from https://stats.idre.ucla.edu/stat/stata/ado/analysis.
Another window will pop up. Click the link that says click here to install.
The package should only take a few seconds to install.
Step 3: Perform the Goodness-of-Fit Test.
Once the package is installed, we can perform the Goodness of Fit Test on the data to determine if the true distribution of race is as follows: 70% White, 20% Black, 10% Other.
We will use the following syntax to perform the test:
csgof variable_of_interest, expperc(list_of_expected_percentages)
Here is the exact syntax we’ll use in our case:
csgof race, expperc(70, 20, 10)
Here is how to interpret the output:
Summary box: This box shows us the expected percent, expected frequency, and observed frequency for each race. For example:
- The expected percent of white individuals was 70%. This is the percentage that we specified.
- The expected frequency of white individuals was 1,572.2. This is calculated using the fact that there were 2,246 individuals in the dataset, so 70% of that number is 1,572.2.
- The observed frequency of white individuals was 1,637. This is the actual number of white individuals in the dataset.
Chisq(2): This is the Chi-Square test statistic for the Goodness of Fit Test. It turns out to be 218.13.
p: This is the p-value associated with the Chi-Square test statistic. It turns out to be 0. Since this is less than 0.05, we fail to reject the null hypothesis that the true distribution of race is 70% White, 20% Black, 10% Other. We have sufficient evidence to conclude that the true distribution of race is different from this hypothesized distribution.
Hi
Thank for sharing this great peace of knowledge, I want make small correction about null hypothesis where you say we fail to reject null hypothesis, in fact we reject null hypothesis since p-value is less than 0.05, which is other way round of what you have said.
how to do deviance goodness of fit in stata
Hi Ken…In Stata, you can perform a deviance goodness-of-fit test for a logistic regression model using the `lrtest` command. Here’s how you can do it:
1. **Fit your logistic regression model:**
First, you need to estimate your logistic regression model using the `logit` command. For example:
“`
logit outcome_var predictor_var1 predictor_var2 …
“`
2. **Obtain the fitted model:**
After estimating the model, you can use the `predict` command to obtain the fitted values. Fitted values represent the predicted probabilities from your logistic regression model.
“`
predict fitted, xb
“`
3. **Perform the deviance goodness-of-fit test:**
Use the `lrtest` command to perform the deviance goodness-of-fit test. This command compares the fit of your model to a saturated model (a model with a parameter for each observation), providing a chi-squared test statistic.
“`
lrtest
“`
This command will prompt you to specify the null hypothesis. For the deviance goodness-of-fit test, the null hypothesis is that the fitted model provides a good fit to the data.
4. **Interpret the results:**
After executing the `lrtest` command, Stata will provide the chi-squared test statistic and its associated p-value. If the p-value is small (typically less than 0.05), you may reject the null hypothesis and conclude that the fitted model does not provide a good fit to the data, indicating potential model misspecification.
Here’s an example combining all the steps:
“`stata
// Step 1: Fit logistic regression model
logit outcome_var predictor_var1 predictor_var2 …
// Step 2: Obtain fitted values
predict fitted, xb
// Step 3: Perform deviance goodness-of-fit test
lrtest
“`
Replace `outcome_var` with the name of your outcome variable and `predictor_var1`, `predictor_var2`, etc., with the names of your predictor variables. Make sure to include all relevant variables in your model.