Inferential statistics in survey analysis

Analyzing Survey Data in Python

EbunOluwa Andrew

Data Scientist

Introduction to inferential statistics in survey analysis

Associations between variables
How well sample represents larger population
Cause-and-effect relationships

Sample scenarios

Teenage girl in a car as her mother sits next to her

Happy young businesswoman

Z-score

Positive = higher than mean
Negative = lower than mean
Greater than absolute value of 3:
- Unusual value
scipy.stats.zscore()

Gaussian distribution

Survey example on demographics

print(survey)

| Gender | Age | Height | Weight |
|--------|-----|--------|--------|
| female |  20 |    163 |     48 |
| female |  19 |    163 |     58 |
| female |  20 |    176 |     67 |
| female |  20 |    170 |     59 |
| male   |  20 |    186 |     77 |

Visualizing age column

survey.Age.plot(kind='hist')

histogram of Age column

Calculating z-score on age column

survey['Age_zscore'] = stats.zscore(survey.Age)
print(survey)

| Gender | Age | Height | Weight | Age_zscore    |
|--------|-----|--------|--------|---------------|
| female |  20 |    163 |     48 | -0.1293116227 |
| female |  19 |    163 |     58 | -0.4955134449 |
| female |  20 |    176 |     67 | -0.1293116227 |
...

Calculating z-score on age column

survey[(survey.Age_zscore >= 3) |
       (survey.Age_zscore <= -3)]

| Gender | Age | Height | Weight | Age_zscore   |
|--------|-----|--------|--------|--------------|
| male   |  30 |    178 |     71 | 3.5327065994 |
| female |  30 |    160 |     47 | 3.5327065994 |
| male   |  29 |    189 |     96 | 3.1665047772 |
| male   |  29 |    183 |    111 | 3.1665047772 |
| male   |  29 |    173 |     81 | 3.1665047772 |
| female |  30 |    173 |     69 | 3.5327065994 |
| female |  29 |    170 |     58 | 3.1665047772 |
...

Z-score analysis

Business Man with a credit card - Photo by rupixen.com on Unsplash

Let's practice!

Analyzing Survey Data in Python