Statistical Modeling Techniques
1. Statistical Modeling Techniques
Welcome back!2. Why use statistical modeling techniques in survey analysis?
Statistical models allow us to make population predictions from survey data by learning the relationship between variables. They can include visualizations to help identify these relationships and make predictions from raw data. This is powerful because insights are unforgettable when illustrated.3. When to use statistical modeling techniques
There are three instances when we would use statistical modeling techniques: when the data being analyzed is difficult to read; when we want to decide how several variables influence each other; and when we want to predict an outcome.4. Example statistical modeling techniques
The three techniques we'll be covering are: linear regression, the two sample t-test, and the chi-square test.5. Linear regression model
A fundamental statistical modeling method is linear regression modeling. It relates the dependent variable, y, and at least one independent variable, x, taking the linear form, y=m*x+b, with m as the slope, and b as the y-intercept.6. Linear regression in survey analysis
Let's examine a survey that looked at employees' mental health in two different types of companies. Here, we want to analyze employee burn_rate and their mental fatigue score.7. Linear regression in survey analysis
If we want to use the linear regression model, we must first visualize our data to make sure it follows a linear trend. Here we use a scatter plot to do just that. Using the dot scatter function, x equals mental_fatigue_score, and y equals burn_rate. We can see both variables show a linear relationship, with a positive slope. The higher an employee scores their mental fatigue, the faster they're likely to be mentally burnt out. This is a good metric for the companies involved in this survey because they can predict a new employee's burn rate based on their mental fatigue score.8. Two-sample t-test
The two sample t-test allows us to test if there is a statistically significant difference between two population statistical means. The null hypothesis for this test proposes that both statistical means are equal while rejecting the null hypothesis proposes that they are not equal.9. Two-sample t-test in survey analysis
Let's apply a two-sample t-test to our survey on employees' mental health at two different types of companies. If we wanted to know whether the mean employee burn_rate differs by company type, we first subset the data by company types: Service, and Product. The t-test then takes the mean burn_rate of both groups to determine if there is a statistical difference between groups to accept or reject the null hypothesis.10. Chi-squared test
The chi-square test allows us to test if there is a statistically significant association between two categorical variables. The null hypothesis means there is no significant association between both variables, while rejecting the null hypothesis means there is a significant association between both variables.11. Chi-square test in survey analysis
If we wanted to test the relationship between company type and the option to work from home, we would use the chi-square test. The results will help us to determine if there is a statistically significant association between the likelihood of working from home and working for either a service or product company.12. Which technique to use? - linear regression
A quick way to know when to use any of these three statistical techniques is by looking at the types of variables being analyzed. If the two variables we're studying are numerical, and have a linear relationship, like workout time versus calories burned, we will use linear regression.13. Which technique to use? - two-sample t-test
When comparing two different populations, if one variable is categorical, and the other, numerical, like diet type versus weight lost, then we use the two-sample t-test.14. Which technique to use? - chi-square test
But if both of the variables being studied are categorical, like gender versus political party, we will use the chi-square test.15. Let's practice!
Let's practice!Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.