Demographics of survey respondents |

Generally, when distributions are skewed, the median gives a better "central tendency" of the data than the arithmetic mean. This is because the median is the middle (50th percentile) observation in the data and would not be influenced even if someone reported being a million years old. The three questions related to how many hours per week on average an individual does programming as paid work, as part of education, or as part of unpaid work, the mean and the median give quite different results.

For paid work, the median is 37 hours per week and the most frequently occurring response is 40 hours per week. Time programming in education or courses comes out at about 4 hours whereas unpaid work at 5 hours.

About 50% of the paid time goes into actually doing programming with a standard deviation of about 30 percentage points. This implies that under a normal distribution assumption, almost 70% of respondents spend between 20 and 80 percent of their time doing programming.

The total months of programming experience have some really extreme values. These outliers were removed from analysis using Grubbs' test. The median amount of total programming experience is about 4.5 years with the mean of about 6 years. The figure below shows the distribution for this variable.

## No comments:

## Post a Comment