Reading about how a salary survey's statistics were calculated isn't fun for most people, but presenting this dry information is a necessary part of any salary survey report. So, grab that cup of coffee or highly caffeinated soda and try to get at least part way through it. We promise to try to make it as painless as possible.

In the Windows IT Pro Industry Survey 2004, we asked respondents a series of four questions designed to reveal not only their base salaries but also the overall income they expect to receive in 2004. By base salaries, we mean straight salaries—no additional income from job-related sources (e.g., bonuses, stocks). Overall income includes the base salary and any additional income from job-related sources. Here's how we gathered the necessary data and how we calculated the average additional income, average salary, and average overall income statistics in the How Does Your Overall Income Compare? chart.

First, we asked respondents to specify their expected base salary for 2004 by using the following salary ranges:

 \$300,000 or more \$80,000 to \$89,999 \$250,000 to \$299,000 \$70,000 to \$79,999 \$200,000 to \$249,000 \$60,000 to \$69,999 \$150,000 to \$199,000 \$50,000 to \$59,999 \$125,000 to \$149,999 \$40,000 to \$49,999 \$100,000 to \$124,999 \$30,000 to \$39,999 \$90,000 to \$99,999 Under \$30,000

Next, we asked respondents three questions to determine whether they expect to receive additional job-related income in the form of bonuses, stocks and stock options, or income from other sources. For each question, respondents could choose from the following options:

 \$10,000 or more \$4,000 \$9,000 \$3,000 \$8,000 \$2,000 \$7,000 \$1,000 or less \$6,000 Not applicable \$5,000

At this point, we had all the data we needed to calculate overall income. However, we soon discovered that for this calculation, it would be much easier to have a specific value representing each respondent's salary rather than a salary range. So, we decided to be conservative and use the low end of each respondent's salary range to represent that respondent's salary. So, let's say Respondent 1 selected the \$30,000 to \$39,000 range and Respondent 2 selected \$40,000 to \$49,999. Respondent 1's salary was recorded as \$30,000, whereas \$40,000 was used for Respondent 2's salary. For the Under \$30,000 range, \$29,999 was used.

After the respondents' salaries were set, it was time to determine how much, if any, additional income respondents received. For the most part, the options from which the respondents chose were already a specific number (e.g., \$8,000, \$3,000). Only two options were ranges: \$10,000 or more and \$1,000 or less. For the \$10,000 or more range, \$10,000 was recorded. For the \$1,000 or less range, \$1,000 was recorded. If a respondent selected the Not applicable option, \$0 was recorded.

Next, we added up the bonus, stock, and other-sources income for each respondent to obtain a number that represents the respondent's overall additional income. Adding three numbers together certainly isn't rocket science, but we did have to account for the fact that respondents don't answer every question. How did we do that? The following example sums up the process nicely (pun intended):

Respondent 1: \$10,000 + \$3,000 + \$2,000 = \$15,000
Respondent 2: \$10,000 + \$0 + \$2,000 = \$12,000
Respondent 3: \$10,000 + x + \$2,000 = \$12,000
Respondent 4: \$10,000 + x + \$0 = \$10,000
Respondent 5: \$0 + \$0 + \$0 = \$0
Respondent 6: x + x + x = x

In these examples, x represents a missing data point. The bold number is the number that's recorded as the respondent's overall additional income. If a respondent didn't answer any of the additional income questions, the respondent's overall additional income was recorded as missing as well. This step is important, as you'll see shortly.

Next, we calculated the average overall additional income for each job title. When the overall additional income was 0, a 0 was added to the total for the job title and that response was incorporated into the divider when calculating the average. When the overall additional income was recorded as a missing data point, no amount was added to the total for the job title and that nonresponse wasn't incorporated into the divider. For example, suppose that Respondents 1, 3, and 5 are trainers and Respondents 2, 4, and 6 are systems administrators. The average overall additional income for the trainers is

\$15,000 + \$12,000 + \$0 = \$27,000/3 = \$9,000

\$12,000 + \$10,000 + x = \$22,000/2 = \$11,000

With the average overall additional income calculated, we determined the average salary for each job title. You've probably noticed that the average salary statistic is used in many of this article's figures and tables. The respondents' base salaries were used to calculate average salaries. For example, suppose our trainers (Respondents 1, 3, and 5) earn \$30,000, \$40,000 and \$50,000, respectively. To determine the average salary for these trainers, you'd perform the following calculation:

\$30,000 + \$40,000 + \$50,000 = 120,000/3 = \$40,000

Let's say our systems administrators (respondents 2, 4, and 6) earn \$40,000, \$50,000, and \$60,000, respectively. Their average salary would be

\$40,000 + \$50,000 + \$60,000 = 150,000/3 = \$50,000

Finally, for each job title, we simply added the average salary and average additional income to come up with the average overall income. For example, the average overall income for our hypothetical trainers would be

\$40,000 + \$9,000 = \$49,000

And the average overall income for the systems administrators would be

\$50,000 + \$11,000 = \$61,000

Now, if you're still awake enough to read this line, give yourself a pat on the back.