Research Methods, Limitations and Cautionary Notes

The analysis and use of data pertaining to political donations is plagued with difficulties and limitations. This is largely the result of systemic failures arising from imprecise and inadequate legislation that leads to a lack of transparency and accountability. Consequently, this material comes with a warning to all users of the limitations in the data set. A set of notes that identify key limitations and research methods are outlined below:

Research Methods

1. There are two primary sources of political donations data for Western Australia, namely the annual financial disclosure returns of the Australian Electoral Commission (AEC) and the West Australian Electoral Commission (WAEC). As the WAEC accepts AEC disclosures, it is only necessary to focus on one data source since they are effectively a duplicate of each other. To reduce the risk of human error, the primary data source (i.e. a hard copy of the original disclosure submitted by each political party) submitted to the WAEC was favoured over the electronic AEC version. Since the AEC manually enters the data provided by each political party, their data set can be considered a secondary information source.
2. The donors have been grouped into 22 categories. To do this, the database adopted the Australian and New Zealand Standard Industrial Classification (ANZSIC) method which applies 17 categories including:
   1. Agriculture, forestry and fishing
   2. Mining
   3. Manufacturing
   4. Electricity, gas and water supply
   5. Construction
   6. Wholesale trade
   7. Retail trade
   8. Accommodation, cafes and restaurants
   9. Transport and storage
   10. Communication services
   11. Finance and Insurance
   12. Property and business services
   13. Government admnistration and defence
   14. Education
   15. Health and Community services
   16. Cultural and recreational services
   17. Personal and other services

   A detailed description of each ANZSIC category can be found at arc.gov.au. It was necessary to add to this list an additional five classifications to properly cover the range of organisations represented in the dataset. The additional classifications include:

   18. Personal
   19. Political party
   20. Political party - associated entity
   21. Professional organisation
   22. Unknown
3. An entity in the dataset was classified as "Non-Corporate" if the following ANZSIC categories applied:
   1. Government administration and defence
   2. Education
   3. Health and community services
   4. Personal
   5. Political party
   6. Political party - associated entity
   7. Professional organisation
   All other categories were classified as "Corporate".
4. The following WA political parties are represented in the dataset:
   1. The Liberal Party
   2. The Australian Labor Party
   3. The National Party
   4. The Australian Democrats
   5. The Greens
   6. The Christian Democratic Party
   7. The Pauline Hanson One Nation Party
   8. Family First
   9. Australia First
   10. The Home Party

Limitations

1. The first limitation that must be identified is the standard one - the risk of human error in data entry. Whilst every effort is made to minimise this introduced risk, it cannot be completely avoided.
2. Several problems arise from the ambiguous and inconsistent classification of donations and non-donations, loans, gifts, waivers and other monies. The AEC requires that all funds received by political parties be classified as either a "donation" or "non-donation", however, there is great concern that "non-donations" are donations in disguise. For example, "non-donations" might include such entries as Australian Taxation Office (ATO) tax return receipts, internal funding from a party's National Executive or funds raised from $2000 a head dinners. Clearly the funds received from the ATO cannot be considered a donation, however the two latter receipts, could certainly be classified as such. It is for this reason that both "donation" and "non-donation" receipts (excluding such obvious non-donations as those from the ATO etc…) are included in the database.
3. Since both associated entities and their respective political parties report the funds raised in each fiscal year (associated entities disclose the funds they raise and from whom, whilst their related political parties disclose the amounts in total received from their associated entities) it is only necessary to input data from a single source. For consistency, the data disclosed by political parties was selected for entry. Notwithstanding this, the disclosed values reported by associated entities provide a valuable insight into how political donations can be hidden by means of an intermediary. For instance, an individual may donate to the 500 Club knowing that these funds will be forwarded on to the Liberal Party, but only the Club details will appear in the database, not the individual. An argument can be made for expanding the associated entity donation that appears on the party's annual return. This is clearly a limitation of the dataset since a large number of donors are consolidated into single 'associated entity' donations, thus reducing transparency and accountability.
4. It must be noted that the dataset only includes reportable political donations. That is, donations over the disclosure limit of $1500. This complicates matters since:
   1. It can be expected that a large number of donations below the disclosure limit are missed.
   2. Overall, the Liberal party has received approximately a third more reportable funds that the Labor party. This may be due to the fact that the Labor party receives a greater proportion of its funding from donations below the $1500 disclosure threshold. If so, the results will be heavily distorted.
   3. Analysis and deduction from the data and trends only represent donors who are donating in excess of $1500. Thus, the analysis is skewed towards focusing on the extreme (in statistical terms), rather than the median donor. This will have statistical and analytical consequences.
   4. The sample under analysis is likely to be significantly smaller than the overall donor population. Moreover, the sample is non-random and concentrated due to the disclosure threshold. Together these imply that the results risk being non-representative of the true donor population.
5. As stated in Research Methods, the database adopted the Australian and New Zealand Standard Industrial Classification (ANZSIC) method which has been used to sort donors into 17 categories (with an additional 5 categories added out of necessity - see Research Methods for more details). The ANZSIC categories are necessarily broad and consequently a limitation of this classification method is the risk of misclassification. This may result from the categories being too general or from the entity being classified being too general. For example, Wesfarmers - best described as a conglomerate, can be classified as a mining, manufacturing, transport, finance and insurance or retail trade company. Since a single category could only be applied in Wesfarmers case, its most consistent and biggest revenue earning business segment was selected - retail trade. Moreover, an added risk is the change in business focus that may have occurred during the time period in question. Since the database only applies a single classification over the entire timeframe, misclassification can once again result.
6. Another issue that stems from the use of applying a classification system is the inability to apply a classification for lack of information. This is a serious problem as evidenced by the fact that "Unknown" ranks as the 7th highest industry classification with in excess of A$1.6m. Thus, it is apparent that several entities have not been adequately identified and classified. Moreover, several of these entities have made considerable donations. For example, little is known about Matfield Holdings, the fourth largest overall corporate donor and largest corporate donor to the Labor Party in WA.
7. The task of comprehensively consolidating the donor list was abandoned due to the degree of difficulty involved. Whilst a consolidated donor list would be infinitely more valuable to determine the true identity of key donors, it is a complex task of tracing ownership stakes (which can fluctuate). Moreover, there is a very grey area between ownership and control at the consolidation level. That is, an entity may have a minority (e.g. 30%) stake in a firm, yet due to the structure of the remaining shareholders, it may still exert a controlling influence. Hence, it should be noted that a limitation contained in the data is the fact that the values reported are not consolidated. For example, Wesfarmers and Bunnings are reported separately even though Bunnings is a wholly owned subsidiary of Wesfarmers.

Document last updated on: 10:30 3rd Apr 2006.