Website History

In 2013, ontariosunshinelist.com (the "Site") was created to provide the Ontario citizen with a comprehensive and unbiased tool to analyze the Ontario sunshine lists. It operates as a free public service, with the belief that a more informed citizenry leads to greater transparency in government.

The Site, including all data analysis and calculations, was made entirely by a single Ontario resident without support, affiliation, or endorsement from any private or public entities. The Site is regularly maintained and updated.

The Site is used by thousands of Ontarians daily and is often cited by news media, research institutions, unions and other employee groups.

Website Methodology

1. Data collection

The Site relies exclusively on the publicly available sunshine lists available for download via the Government of Ontario official website. Other than data used to access a name's assumed gender, there are no other data sources used other than the official sunshine lists released by the Government of Ontario.

In addition to the sunshine list released each year, the Government of Ontario usually includes the following data files within months or sometimes years after the initial release of a list:

1. Addendum - individuals not included in the original list but who should have been
2. Deletions - individuals on the original list but who should not have been
3. Changes - changes to an individual's record as it appeared on the original list

This Site also includes these additional data files as listed above. Furthermore, algorithms have been developed to periodically check these files, since they are sometimes updated more than once. Every effort is made to ensure the data presented on the Site is as up-to-date as possible.

The Site is updated with 24 hrs from the time the Government of Ontario releases the annual sunshine list.

2. Data processing

The Site has developed alogithms specific to the sunshine list dataset to process and clean the data. The goal of cleaning the data is to remove errors and harmonize employer names, position titles, and first and last names in the pursuit of consistency and to increase the occurance of what the Site refers to as common records.

Consider the following two definitions:

Record: a record is a single line of data from the sunshine list, ie. one individual's information for one year.
Common records: common records are two or more records which are deemed to belong to the same individual. They must meet the following criteria:

• Have identical first and last names.
• Have identical employer names.
• Be records from different years.

Note that records that meet the first two conditions but not the third condition are duplicate records. Duplicate records only occur with frequency of approximately 1 in 1250 records.

Common records provide the basis for year over year raises to be calculated, as discussed below.

The algorithms used to harmonize records are manually checked to ensure accuracy and that they are not overly aggressive.

The sunshine list dataset is ripe for data cleaning: the effect of harmonizing employer names and the stylistic versions of first and last names increases the number of common records in a given year by approximately 10-15%. In addition, over 40,000 position titles have been removed and harmonized with existing position titles.

3. Calculations

Raises:

Year over year salary increases are based on common records as discussed above. Consider the following example consisting of two individuals over two years:

Year	Person	Salary
2015	Bob Billows	$110,000
2016	Bob Billows	$121,000
2015	Sue Sampson	$180,000
2016	Sue Sampson	$216,000

Raise calculations are:

• Bob's raise: = ($121,000 / $110,000) = 1.10 = 10% raise from 2015 to 2016.
• Sue's raise: = ($216,000 / $180,000) = 1.20 = 20% raise from 2015 to 2016.

Both raises in this example are attributed to year 2016. In general, raises of (Year N / Year N-1) are attributed to year N. Raises do not include taxable benefits.

The updated table with raises included:

Year	Person	Salary	Raise
2015	Bob Billows	$110,000	N/A
2016	Bob Billows	$121,000	10%
2015	Sue Sampson	$180,000	N/A
2016	Sue Sampson	$216,000	20%

Raises are not calculated for any year where an individual first appears on the sunshine list or where an individual has no common record in the succeeding year.

If these four records were the entire sunshine list, we can say the following:

• there are four records
• there are two individuals
• there are no common records in 2015
• there are two common records in 2016.

Calculation of average raises:

The Site makes use of a metric known as the average raise. For instance, the average raise of all individuals of a particular employer in a given year, or all individuals with a particular position in a given year.

In this example, we can calculate the average raise for year 2016. This average raise must somehow incorporate the information of each of the two raises calculated in that year.

There are two approaches to calculating the average raise:

1. Head-weighted average raise
2. Money-weighted average raise

The Site uses the head-weighted approach for calculating average raises. Both methods of calculations are shown below.

1. Head-weighted average raise:

The head-weighted average raise uses a headcount based weight. Each raise receives the same weight in the calculation of the average since each raise came from 1 person. In this example, there are two people, meaning each raise receives 1 / 2 = 0.50 weight. The average raise is then calculated as:

Head-weighted average raise in 2016 = (0.50)*10% + (0.50)*20% = 15%.

This is simply an equal weighted average of all raises in year 2016.

2. Money-weighted average raise:

The money-weighted average raise can be demonstrated in two ways.

1st approach: weighted approach

Under this approach, the weights are based on the starting salaries of each component raise. The salary that grew by 10% receives a weight of 110 / (110 + 180) = 37.9% and the salary that grew by 20% receives a weight of 180 / (110 + 180) = 62.1%.

Money-weighted average raise in 2016 = [$110,000 / ($110,000 + $180,000)]*10% + [$180,000 / ($110,000 + $180,000)]*20% = 16.207%.

2nd approach: aggregate approach

Money-weighted average raise in 2016 = ($121,000 + $216,000) / ($110,000 + $180,000) = $337,000 / $290,000 = 1.16207 = 16.207% raise.

Under this approach, a single raise is calculated based on the sum of the component salaries.

From the perspective of an employer paying out salaries and determining raises, the second approach may be more useful to them. The employer paid out salaries of $337,000 in 2016 based on salaries of $290,000 in the year prior, representing an increase of 16.207% from the perspective of their total outlay of salaries of common records.

From the perspective of the employees who received the raises, the first approach may be more useful to them. Bob received a raise of 10% and Sue received a raise of 20%. On average, a raise of 15% was given to each of them.

In this example the money-weighted approach produces a higher average raise since the salary that received the higher of the two raises was, prior to the receiving of the raise, already the higher of the two salaries.