The white shark study, and the file it came from
A secondary-data investigation, followed from start to finish: no site visit, no equipment, no fieldwork. One published file, every animal caught on Queensland’s shark nets and drumlines since 1996, and every step needed before it could answer a question.
Every number below came out of the real file. So did every problem with it.
Since 1962, Queensland has set shark nets and drumlines (baited hooks on buoys) off popular beaches on its coast, in Australia. They catch more than the sharks they are there for: the programme’s own file records 1,456 turtles and 590 marine mammals. They also catch white sharks, a protected species. Which gear to set, nets or drumlines, is the programme’s own decision, and it records every animal each one catches. So I could test, not just claim, whether the two gears differ in what they do to one protected shark.
The question, in three drafts
Most first questions are too big to answer or too vague to measure. The dangerous draft is the second one, because it looks finished.
“Do shark nets work?”
No indicator, no units and no variation to explain. And "work" means bather safety, which this file does not measure.
“Do nets catch a larger share of white sharks than drumlines across Queensland?”
Measurable and dated, and asking the wrong question of this file. Drumlines are also set far north, from Cairns to Bundaberg, where white sharks do not swim, so a statewide comparison compares places as well as gear.
“A comparison of the share of the sharks caught that were white sharks, between nets and drumlines, in the four areas of Queensland's Shark Control Program that set both (Gold Coast, Sunshine Coast North, Sunshine Coast South and Rainbow Beach), 1996 to 2023.”
Restricted to the four areas that set both gears, so the two are compared in the same places.
Independent variable: the gear, Net or Drum. Dependent variable: white sharks as a percentage of all the sharks that gear caught.
“Nets will catch the larger share of white sharks, because a net takes whatever swims into it, while a baited hook takes what comes to the bait.”
Students often think a hypothesis the data refutes is a failed investigation. It is not. This one was refuted in the four areas, where the drumlines had the larger share, although the statewide figure would have seemed to support it. I kept it, because working out why it was wrong is most of the analysis.
The system, and the three lines
Every investigation in this guide fills three lines, running one way. Here the third line is also the first: the plan is the gear.
The gear: a net, or a drumline (a baited hook on a buoy). The programme chooses which to set, and the file records it in one column, Gear.
White sharks, a protected species, as a share of all the sharks each gear caught. Worked out from CommonName and NumberCaught.
Queensland's Shark Control Program. The gear is the plan, so the real-world strategy and the independent variable are the same thing.
It is tempting to read a catch file as a count of sharks. It is not. The file records a flow: the animals the gear takes out of the coastal water each month. It says nothing about the storage, how many sharks live along the coast. White sharks pass in winter and spring, so how many are caught depends on when they pass, where the gear is and how much of it is in the water, and only partly on how many there are. That is why this study can compare the gears, and can never say whether white sharks are declining.
The strategy, and the argument about it
Criterion B (Strategy) needs one real-world strategy and one explained tension. Here the strategy is not beside the data: it is the independent variable, as close a fit as a study can have.
Who: The Queensland Government, through the Department of Primary Industries: nets and drumlines off about 85 beaches, serviced by contractors.
How it bears on the question: Directly: the gear is the programme. Which to set, nets or drumlines, is its own decision, and my question asks what each one does to one protected shark.
The tension: Swimmers' safety and a beach economy against wildlife: the file itself records 1,456 turtles and 590 marine mammals caught. The government's own KPMG review advised moving away from nets and drumlines; the 2025 plan expanded them, and scientists said culling does not make beaches safer.
See the sourceWho: Humane Society International, with the Environmental Defenders Office, took the programme to the Administrative Appeals Tribunal. Queensland now runs catch alert drumlines there, which send a satellite alert so a team can release the shark.
How it bears on the question: Indirectly: it is the same argument, taken to a tribunal and then a court, and it shows what the tension does.
The tension: The tribunal found the evidence "overwhelming" that killing sharks does not reduce the risk of bites, against a state government that appealed the ruling and lost.
See the sourceStudents usually look for the disagreement in the data. It is not there. It lives in the government’s plan and the programme’s own pages, in a review the government commissioned (KPMG), in a tribunal decision, and in the news reporting of both sides.
What the tension does: resistance and delay (a lawsuit, then an appeal), and uneven costs. Bathers and the beach economy get the claimed safety; the wildlife, and the people who value it, bear the catch. For the environment, more gear in the water means more animals caught, protected species included. My question sits on the choice the plan makes: what each of its two gears does to one protected shark.
The dataset, and the protocol
On this route, the method section tells a reader how to get exactly the same file. This table is the model. Everything in it was recorded from the screen at the moment of download. None of it could have been worked out a month later.
Open the dataset page and download the file yourself
Click the file’s link in a browser. The portal’s bot protection answers a script with an empty file, so “the download worked” and “I have the data” are not the same thing. Open it and check the rows are there before anything else.
The usual assumption is that one row is one animal. It is not. One row is one kind of animal caught on one kind of gear at one beach in one month, with how many in NumberCaught. In the rows I analysed, 591 of 4,299 record more than one. And a month in which no white shark was caught has no white shark row at all, not a row with 0. Adding up NumberCaught handles both. Counting rows handles neither.
What the file actually was
A government file looks trustworthy, and this one is the programme’s own record. It still says less than it seems to.
All sharks caught ran at about 500 to 850 a year until 2022, then 948, 1,497 and 3,430 in 2023 to 2025. From 2024 the programme brought in new locations, daily servicing and new drumlines, and because the file records the programme’s own catch, more gear looks exactly like more sharks. A study that runs into those years measures the programme as much as the coast. And 2026 stops in May.
White sharks are rare in it. In the four areas, 1996 to 2023, they come eighth among the sharks caught:
And they are almost all caught south of Bundaberg. The drumlines are also set from Cairns to Bundaberg, where the sharks are almost all tiger sharks, whalers and hammerheads. That one fact is why the second draft of the question had to go.
The whole calculation
Raw counts cannot be compared between the gears. Nets caught 3,296 sharks in the four areas and drumlines 1,932, so any count of white sharks would mostly compare how much each gear caught. So each gear is measured against itself: white sharks as a percentage of its own shark catch. That is the normalising control. The file does not contain that number; I made it.
The restricting control is the inclusion rule. I fixed it before I looked at a single white shark:
- Year 1996 to 2023
- SpeciesGroup SHARK
- Gear Net or Drum
- Area Gold Coast, Sunshine Coast North, Sunshine Coast South or Rainbow Beach
Two more decisions, also made before looking. Gear “Other” is 5 rows, holding one white shark: left out. UNKNOWN SHARK (18 animals) and the starred groups stay in “other sharks”: they are sharks, and none is recorded as a white shark. What survives is 66 beaches (16 with nets only, 35 with drumlines only, 15 with both): 4,299 rows holding 5,228 sharks.
Filter to the rule and copy what is left into a new tab. Columns as downloaded: F is Gear, J is CommonName, O is NumberCaught. P is the one column I added.
Why a new tab: SUMIFS adds up every row in the column, including the rows a filter has only hidden. Run these on the filtered original and they count every year, every area and every species, turtles included. The check: 73 + 3,223 = 3,296 and 74 + 1,858 = 1,932. If yours differ, a filter is wrong.
One helper column, four SUMIFS, one division and a chi-squared test. That is the whole method.
What it found
The four numbers the SUMIFS gave, and the share each one makes. Four areas, 1996 to 2023.
The test is a chi-squared test of independence on the 2 by 2 table: gear against white shark or other shark, with one degree of freedom. It runs on the counts, never the percentages, because a percentage has thrown away how many sharks it was made from. The result: chi-squared = 11.6, p about 0.0007, far above the critical value of 3.84 at p = 0.05. Every expected count is above 50 (the smallest is about 54), so the test is valid.
So the hypothesis is refuted. It was the drumlines, not the nets, that caught the larger share of white sharks, and the same is true in each of the four areas on its own:
The season behind the gap
The headline is significant. That does not mean it is explained. Adding up the same columns by month explained it.
White sharks pass in winter and spring: 124 of the 147 were caught June to October. The nets catch most of their other sharks in summer. Only 30% of the nets’ sharks were caught June to October, against 43% of the drumlines’.
So I compared the two gears in those five months alone. Nets: 63 white sharks of 1,002 (6.3%). Drumlines: 61 of 832 (7.3%). Chi-squared 0.8, p about 0.38: no significant difference.
Read carelessly, the headline says drumlines are worse for white sharks. The months say something else. Most of the difference comes from when each gear catches its other sharks, not which sharks it catches. The nets’ big summer catch of other sharks dilutes their white-shark share. Season is a confounding variable: it is tied to the gear’s catch and to the white sharks at once. Finding it is the kind of analysis Criterion E (Analysis and conclusion) is there for.
The conclusion, and what it cannot claim
A conclusion claims exactly what was found, and no more. This is mine.
“Across 1996 to 2023, in the four areas that set both gears, white sharks made up a larger share of the drumlines’ shark catch (3.8%) than of the nets’ (2.2%), a difference unlikely to be chance (chi-squared 11.6, p about 0.0007). But in the months white sharks pass, June to October, the two shares were close (7.3% and 6.3%) and not significantly different. The difference comes mostly from when each gear catches its other sharks, not from which sharks it catches.”
Four sentences the file cannot carry, each tempting:
“Drumlines catch more white sharks than nets.”
It is a share, not a rate. The file does not say how many nets and drumlines were in the water, or for how long, so it cannot say which gear catches more white sharks a day.
“Drumlines are worse for white sharks.”
The season test says most of the difference is timing. And if anything the fate data points the other way: 37 of the drumlines' 74 white sharks were alive when the gear was checked, against 28 of the nets' 73.
“Shark control makes beaches safer.”
Nothing here measures bites or swimmers.
“White sharks are declining.”
Catch depends on gear and effort, not only on how many sharks there are.
Eight things wrong with it
None of these were added on purpose. They are all in the programme’s own file, and each one is taught somewhere in the guide. Where one pushes the answer, it says which way.
- 1Place. Drumlines are also set from Cairns to Bundaberg, where there are no white sharks. Take in the whole state and the drumlines' share is dragged down until the answer reverses. step 2 →
- 2The download can arrive empty: the portal answers a script with an empty file. Click the link in a browser and check the rows are there. step 4 →
- 3The gear changed in 2024, and the programme records its own catch, so more gear looks exactly like more sharks. Hence the stop at 2023. step 4 →
- 4A row is not an animal: 4,003 rows in the file record more than one. Counting rows undercounts. step 5 →
- 5A month with no white shark has no row, not a row with 0. Only adding up NumberCaught treats the missing zeros correctly. step 5 →
- 6Season. The nets' big summer catch of other sharks dilutes their white share, which pushes it down and inflates the gap between the gears. step 6 →
- 7Effort is not in the file, so it gives a share and never a rate. Which way that pushes the answer is unknown. step 6 →
- 8Identification. 18 UNKNOWN SHARK and the starred groups could hide white sharks. That would push the true shares slightly up, by an amount nobody can tell for each gear. step 7 →
The evaluation
Students usually list weaknesses as if they were equal. They are not. Mine, ranked by how much they mattered:
- 1Season. It explains most of the difference. The sensitivity test: keep only June to October, and chi-squared falls from 11.6 to 0.8. This is a sensitivity test that failed, and it was worth more than one that passed: it turned a finding about gear into a finding about timing.
- 2Place. The second sensitivity test. Across the whole state the answer reverses: nets 1.8% (73 of 4,162), drumlines 0.9% (132 of 14,175), chi-squared 19.7. That is why the restriction to four areas was fixed before looking, and why it goes in the method with its reason.
- 3No effort data. A share, not a rate. Without the amount of gear in the water, no sentence about which gear catches more white sharks can be written.
- 4Identification. 18 unknown sharks sit in "other sharks". Any of them could have been a white shark.
The two sensitivity tests, in one table: the same comparison made four ways.
- The programme's equipment locations file, on the same dataset page ("Queensland Shark Control Program equipment locations"), gives where each net and drumline is. With it, catch could be divided by the amount of gear, turning a share into a rate. It lists current locations, so it would need matching to the years it covers.
- The biological information file on the same page ("Biological information from caught animals") records lengths. It could test whether the two gears catch white sharks of different sizes.
- Compare the gears month by month rather than pooled, or pair the 15 beaches that set both gears.
Each has a different focus from my question, and each bears on the strategy.
- Does a white shark caught on a drumline survive more often than one caught in a net? The file has 37 of 74 alive against 28 of 73. I did not test it.
- Do nets and drumlines catch white sharks of different sizes?
- What did the 2024 expansion, with new drumlines and daily servicing, do to the white-shark catch?
The sources
All three are real and public. Follow each link to the original; the summary is mine.
The programme's own record of every animal its nets and drumlines catch, by beach, month, gear and species. It has one author, the programme, which both collects the data and publishes it, so there is no second compiler to credit.
The same page holds the equipment locations file and the biological information file named in the evaluation.
Open the dataset pageThe government's own KPMG review advised moving away from nets and drumlines. The 2025 plan expanded them, and scientists said culling does not make beaches safer.
Cite it for who said what. Cite the data file for what was caught.
Read the reportCatch alert drumlines send a satellite alert so a team can release the shark. Queensland now runs them in the Great Barrier Reef Marine Park, after the tribunal's April 2019 ruling against killing sharks there.
Open the page- Queensland Government, Department of Primary Industries. "Queensland's Shark Control Program Data and Information: Number caught by area, calendar year and species group." Queensland Government Open Data Portal, 2026. CC BY 4.0. Accessed 26 September 2026. https://www.data.qld.gov.au/dataset/fisheries-queensland-s-shark-control-program-information-and-data
- Will Murray, "Plan to increase lethal shark control in Queensland 'baffling' to scientist," ABC News, 8 June 2025, accessed 26 September 2026, https://www.abc.net.au/news/2025-06-08/qld-shark-control-program-against-advice-of-kpmg-report/105387824
- Queensland Department of Primary Industries, catch alert drumlines page. https://www.dpi.qld.gov.au/news-media/campaigns/sharksmart/equipment/catch-alert
The page’s own title could not be read, so the third reference describes the page rather than quoting one. Never invent a title to make a reference look complete.
The natural word for a shark-control programme is killed, and it is wrong. The file records animals caught, and 65 of the 147 white sharks were alive when the gear was checked. Say what the number is: caught. step 8 →
- 3.2 Human impact on biodiversity, A protected species caught by gear set for a different purpose, beside 1,456 turtles and 590 marine mammals in the same file.
- 3.3 Conservation and regeneration, A protected shark, a tribunal ruling against killing inside a marine park, and catch alert drumlines that let a team release what is caught.
- 1.2 Systems, The file counts a flow, the animals taken from the coast each month, not a storage, how many sharks there are.
- 1.1 / 1.3 Perspectives and sustainability, Swimmers and a beach economy against wildlife, and a plan that expanded the gear its own government's review advised moving away from.
- Across the whole state nets had the larger share; in the four areas, drumlines did. Which comparison answers the choice the programme actually makes, and why?
- In June to October the gap almost disappears. Is the headline finding wrong, or only badly explained?
- 65 of the 147 white sharks were alive when the gear was checked. Does that change what "worse for white sharks" should mean?
- The file shows what each gear caught, not how much gear was in the water. How would you word a conclusion that stays inside what a share can say?
- The KPMG review advised moving away from nets and drumlines, and the 2025 plan expanded them. Could anything in this file settle that argument?
In your IA: notice how much of this happened before any analysis. The three drafts, the inclusion rule fixed before looking, and the check that four SUMIFS add up to 3,296 and 1,932 are most of Criterion C (Method). The test at the end is the smallest part.
In class: the download is one click in a browser. Give a class the file and the headline, and ask why the nets’ share is lower. Adding Month to the SUMIFS finds the season, and teaches more about confounding than any amount of explaining.
The white shark investigation used on the secondary-data route of this guide is the author’s own analysis of a published dataset: the Queensland Shark Control Program’s record of every animal caught on its nets and drumlines, published by the Queensland Government under CC BY 4.0 and downloaded on 26 September 2026. The choice of the four areas, the analysis and the conclusions are the author’s, not the Queensland Government’s.