Data Lab · Year 10 Statistics

How it works

Four steps. Your booklet walks you through all of them, and you get your question signed off before you touch a computer.

Pick a dataset. Scroll down, read what one row means, and download it.
Write your question. It needs two number columns and it asks whether they are related.
Plot it. Scatterplot, explanatory variable across the bottom, then add the line of best fit.
Answer it. Name the association, read the gradient in context, make one prediction, then argue with yourself about whether to trust it.

Write it up here

Download this, open it in Google Docs, and type straight into it. Paste your screenshot into the box at Question 5, then submit the file to the Compass learning task at the end of the lesson. Hand up whatever you got done.

Download the write-up template

The data

Ten to choose from, plus four extras. Every file is small, already cleaned up, and opens straight into a spreadsheet or CODAP. Hit Peek at the data to see what you are getting before you commit.

Not sure what a finished one looks like?

Walk through a whole investigation, start to finish, on the AFL file: the question, the plot, r, the line, and the honest answer at the end. It takes two minutes to read and it will save you a lot longer than that.

See the worked example

Supercars and everyday cars for sale

1627 rows · 7 columns · 100 KB

One row is: one car advertised for sale in Europe
Good numbers to plot: distance_driven_km, price_aud, age_years
Groups you can split by: brand, gearbox
Also in the file: model, year
You could ask: Do cars with more kilometres on them sell for less? Do older cars always sell for less, and which cars flatly refuse to?

Download CSV

Peek at the data

brand	model	year	age_years	distance_driven_km	price_aud	gearbox
Honda	Honda Accord 2.2d Executive	2009	17	265000	12540	Automatic
Mercedes-Benz	Mercedes-Benz GLC	2016	10	122000	55062	Automatic
Ford	Ford Kuga	2014	12	232040	14782	Manual
Ford	Ford Mondeo	2012	14	255000	6840	Manual
Volkswagen	Volkswagen Polo	2013	13	94550	11362	Manual

WNBA players

163 rows · 13 columns · 13 KB

One row is: one WNBA player and their 2019 season
Good numbers to plot: minutes_per_game, player_efficiency_rating, minutes_total, win_shares, usage_pct, true_shooting_pct
Groups you can split by: team, position
Also in the file: player, age, games_played, rebound_pct, assist_pct
You could ask: Do players who spend longer on court score better ratings? Does shooting more often mean shooting well?

Download CSV

Peek at the data

player	team	position	age	games_played	minutes_total	minutes_per_game	player_efficiency_rating	true_shooting_pct	…
Renee Montgomery	Atlanta Dream	Guard	32	34	949	27.9	11.1	52.0	…
Elizabeth Williams	Atlanta Dream	Centre-Forward	26	32	909	28.4	16.7	52.1	…
Brittney Sykes	Atlanta Dream	Guard	25	34	880	25.9	11.3	44.5	…
Tiffany Hayes	Atlanta Dream	Guard	29	29	817	28.2	15.1	49.7	…
Jessica Breland	Atlanta Dream	Forward	31	33	767	23.2	15.2	42.5	…

Antarctic penguins

344 rows · 8 columns · 17 KB

One row is: one penguin measured at a research station
Good numbers to plot: flipper_length_mm, body_mass_g, bill_length_mm, bill_depth_mm
Groups you can split by: species, island, sex
Also in the file: year
You could ask: Do penguins with longer flippers weigh more? Is it the same for all three species?

Download CSV

Peek at the data

species	island	bill_length_mm	bill_depth_mm	flipper_length_mm	body_mass_g	sex	year
Adelie	Torgersen	39.1	18.7	181.0	3750.0	male	2007
Adelie	Torgersen	39.5	17.4	186.0	3800.0	female	2007
Adelie	Torgersen	40.3	18.0	195.0	3250.0	female	2007
Adelie	Torgersen	—	—	—	—	—	2007
Adelie	Torgersen	36.7	19.3	193.0	3450.0	female	2007

AFL players

677 rows · 5 columns · 28 KB

One row is: one AFL player
Good numbers to plot: height_cm, weight_kg
Groups you can split by: position, team
Also in the file: player
You could ask: Are taller players heavier? Do ruckmen sit above or below everyone else?

Download CSV

Peek at the data

player	team	position	height_cm	weight_kg
Scott Thompson	Adelaide	Midfielder	185	85
Richard Douglas	Adelaide	Midfielder	180	80
Rory Sloane	Adelaide	Midfielder	183	85
Patrick Dangerfield	Adelaide	Midfielder	185	92
Bernie Vince	Adelaide	Midfielder	186	87

NBA players

438 rows · 13 columns · 40 KB

One row is: one NBA player and their season
Good numbers to plot: minutes_per_game, points_per_game, height_cm, weight_kg, salary
Groups you can split by: position, team
Also in the file: player, age, years_experience, games_played, rebounds_per_game, assists_per_game
You could ask: Does more time on court mean more points? Is height or court time the better predictor? Does the pay match the points?

Download CSV

Peek at the data

player	team	position	age	years_experience	height_cm	weight_kg	salary	games_played	…
Brad Wanamaker	Boston Celtics	Point guard	30	1	190.5	95.3	1445697.0	99.0	…
Carsen Edwards	Boston Celtics	Point guard	22	0	180.3	90.7	1228026.0	35.0	…
Daniel Theis	Boston Celtics	Centre	27	2	203.2	111.1	5000000.0	187.0	…
Enes Kanter	Boston Celtics	Centre	27	8	208.3	113.4	4767000.0	634.0	…
Gordon Hayward	Boston Celtics	Small forward	30	9	200.7	102.1	32700690.0	634.0	…

Australian Institute of Sport athletes

202 rows · 10 columns · 11 KB

One row is: one elite Australian athlete
Good numbers to plot: height_cm, weight_kg, lean_body_mass_kg, haemoglobin, haematocrit_pct
Groups you can split by: sport, sex
Also in the file: red_cell_count, white_cell_count, ferritin
You could ask: Are taller athletes heavier? Which sport has the tallest athletes?

Download CSV

Peek at the data

sport	sex	height_cm	weight_kg	lean_body_mass_kg	red_cell_count	white_cell_count	haematocrit_pct	haemoglobin	…
Basketball	Female	195.9	78.9	63.32	3.96	7.5	37.5	12.3	…
Basketball	Female	189.7	74.4	58.55	4.41	8.3	38.2	12.7	…
Basketball	Female	177.8	69.1	55.36	4.14	5.0	36.4	11.6	…
Basketball	Female	185.0	74.9	57.18	4.11	5.3	37.3	12.6	…
Basketball	Female	184.6	64.6	53.2	4.45	6.8	41.5	14.0	…

Australian weather, three cities

300 rows · 13 columns · 19 KB

One row is: one day in Hobart, Uluru or Wollongong
Good numbers to plot: min_temp_c, max_temp_c, humidity_3pm_pct
Groups you can split by: location, rain_today
Also in the file: rainfall_mm, wind_gust_kmh, humidity_9am_pct, temp_9am_c, temp_3pm_c, pressure_9am, pressure_3pm, rain_tomorrow
You could ask: Does a cold morning mean a cold afternoon? Does the rule work as well in Uluru as in Hobart?

Download CSV

Peek at the data

location	min_temp_c	max_temp_c	rainfall_mm	wind_gust_kmh	humidity_9am_pct	humidity_3pm_pct	temp_9am_c	temp_3pm_c	…
Hobart	7.9	11.4	1.2	44.0	79	74	9.0	10.5	…
Hobart	6.1	10.8	0.0	31.0	86	86	8.5	10.3	…
Hobart	8.9	19.2	0.8	30.0	52	53	15.9	18.3	…
Hobart	10.6	20.1	0.2	44.0	86	76	12.5	17.5	…
Hobart	7.8	19.0	0.0	35.0	72	42	10.1	17.4	…

Used cars for sale

90 rows · 4 columns · 2 KB

One row is: one second-hand car advertised for sale
Good numbers to plot: age_years, price_thousands_usd, distance_driven_thousand_km
Groups you can split by: model
You could ask: How much value does a car lose each year? Does age or distance driven matter more?

Download CSV

Peek at the data

model	age_years	distance_driven_thousand_km	price_thousands_usd
Mazda6	3	28.6	15.9
Mazda6	2	30.6	16.4
Mazda6	1	33.6	18.9
Mazda6	2	38.6	16.9
Mazda6	2	38.6	20.5

Fast food nutrition

515 rows · 14 columns · 43 KB

One row is: one item on a fast food menu
Good numbers to plot: calories, total_fat_g, sodium_mg, protein_g
Groups you can split by: restaurant, salad
Also in the file: item, calories_from_fat, sat_fat_g, trans_fat_g, cholesterol_mg, carbs_g, fibre_g, sugar_g
You could ask: Can you predict an item's fat from its calories? Which chain comes out worst? Are the salads actually better?

Download CSV

Peek at the data

restaurant	item	calories	calories_from_fat	total_fat_g	sat_fat_g	trans_fat_g	cholesterol_mg	sodium_mg	…
Mcdonalds	Artisan Grilled Chicken Sandwich	380	60	7	2.0	0.0	95	1110	…
Mcdonalds	Single Bacon Smokehouse Burger	840	410	45	17.0	1.5	130	1580	…
Mcdonalds	Double Bacon Smokehouse Burger	1130	600	67	27.0	3.0	220	1920	…
Mcdonalds	Grilled Bacon Smokehouse Chicken Sandwich	750	280	31	10.0	0.5	155	1940	…
Mcdonalds	Crispy Bacon Smokehouse Chicken Sandwich	920	410	45	12.0	0.5	120	1980	…

Pokemon stats

800 rows · 13 columns · 44 KB

One row is: one Pokemon
Good numbers to plot: attack, total_stats, defense, hp
Groups you can split by: type_1, generation, legendary
Also in the file: id, name, type_2, sp_attack, sp_defense, speed
You could ask: Do high attack Pokemon give up defense, or do strong ones just get more of everything? Are legendaries a different breed?

Download CSV

Peek at the data

id	name	type_1	type_2	total_stats	hp	attack	defense	sp_attack	…
1	Bulbasaur	Grass	Poison	318	45	49	49	65	…
2	Ivysaur	Grass	Poison	405	60	62	63	80	…
3	Venusaur	Grass	Poison	525	80	82	83	100	…
3	VenusaurMega Venusaur	Grass	Poison	625	80	100	123	122	…
4	Charmander	Fire	—	309	39	52	43	60	…

Spotify songs

350 rows · 19 columns · 56 KB

One row is: one song
Good numbers to plot: energy, loudness, acousticness, popularity
Groups you can split by: genre, subgenre, playlist_name
Also in the file: title, artist, album, danceability, key, mode, speechiness, instrumentalness, liveness, valence, tempo, duration_ms
You could ask: Are high energy songs mixed louder? Are sad songs slower?

Download CSV

Peek at the data

title	artist	popularity	album	playlist_name	genre	subgenre	danceability	energy	…
On & On	Alok	79	On & On	Dance Room	pop	dance pop	74.7	77.60000000000001	…
All The Lies	Alok	56	All The Lies (Remixes)	Cardio	pop	dance pop	70.7	80.60000000000001	…
Hear Me Now	Alok	75	Hear Me Now	ElectroPop 2020	pop	electropop	77.8	46.3	…
The Wall	Alok	65	The Wall	Electropop	pop	electropop	68.2	91.0	…
Hear Me Now	Alok	52	Hear Me Now	Chillout & Remixes 💜	pop	indie poptimism	78.8	43.0	…

Hollywood movies

970 rows · 16 columns · 100 KB

One row is: one film released between 2007 and 2013
Good numbers to plot: budget_m, world_gross_m, critic_score, audience_score, opening_weekend_m, domestic_gross_m
Groups you can split by: genre, studio, year
Also in the file: movie, story, theatres_opening_week, avg_per_theatre_opening, foreign_gross_m, profitability, opening_profit
You could ask: Does a bigger budget buy a bigger box office? Do critics and ordinary viewers agree, and which films split them hardest?

Download CSV

Peek at the data

movie	studio	critic_score	audience_score	story	genre	theatres_opening_week	opening_weekend_m	avg_per_theatre_opening	…
Spider-Man 3	Sony	61.0	54.0	Metamorphosis	Action	4252.0	151.1	35540.0	…
Shrek the Third	Paramount	42.0	57.0	Quest	Animation	4122.0	121.6	29507.0	…
Transformers	Paramount	57.0	89.0	Monster Force	Action	4011.0	70.5	17577.0	…
Pirates of the Caribbean: At World's End	Disney	45.0	74.0	Rescue	Action	4362.0	114.7	26302.0	…
Harry Potter and the Order of the Phoenix	Warner Bros	78.0	82.0	Quest	Adventure	4285.0	77.1	17998.0	…

Also available

Smaller or stranger. Good if you have finished early or you want something nobody else will have.

Countries: money and lifespan

187 rows · 5 columns · 9 KB

One row is: one country
Good numbers to plot: income_per_person, life_expectancy_years
Groups you can split by: region
Also in the file: country, population
You could ask: Does more money always buy more years of life, or does it level off? Warning: this one is not a straight line, which makes it interesting.

Download CSV

Peek at the data

country	income_per_person	life_expectancy_years	population	region
Afghanistan	1925	57.63	32526562	south_asia
Albania	10620	76.0	2896679	europe_central_asia
Algeria	13434	76.5	39666519	middle_east_north_africa
Andorra	46577	84.1	70473	europe_central_asia
Angola	7615	61.0	25021974	sub_saharan_africa

Australian possums

104 rows · 13 columns · 7 KB

One row is: one possum trapped in Victoria or NSW/Qld
Good numbers to plot: head_length_mm, skull_width_mm, total_length_cm
Groups you can split by: region, sex, trap_site
Also in the file: age_years, tail_length_cm, foot_length_mm, ear_length_mm, eye_width_mm, chest_girth_cm, belly_girth_cm
You could ask: Can you tell a Victorian possum from a northern one by its skull?

Download CSV

Peek at the data

trap_site	region	sex	age_years	head_length_mm	skull_width_mm	total_length_cm	tail_length_cm	foot_length_mm	…
1	Victoria	Male	8.0	94.1	60.4	89.0	36.0	74.5	…
1	Victoria	Female	6.0	92.5	57.6	91.5	36.5	72.5	…
1	Victoria	Female	6.0	94.0	60.0	95.5	39.0	75.4	…
1	Victoria	Female	6.0	93.2	57.1	92.0	38.0	76.1	…
1	Victoria	Female	2.0	91.5	56.3	85.5	36.0	71.0	…

Cars: engine size and fuel use

234 rows · 12 columns · 18 KB

One row is: one car model
Good numbers to plot: engine_litres, highway_l_per_100km, highway_miles_per_gallon
Groups you can split by: class, drive, manufacturer
Also in the file: model, year, cylinders, transmission, fuel_type, city_l_per_100km
You could ask: How much fuel economy do you lose per extra litre of engine? This file has fuel use measured two opposite ways, which is worth a look.

Download CSV

Peek at the data

manufacturer	model	year	engine_litres	cylinders	transmission	drive	class	fuel_type	…
audi	a4	1999	1.8	4	auto(l5)	Front wheel	compact	Premium	…
audi	a4	1999	1.8	4	manual(m5)	Front wheel	compact	Premium	…
audi	a4	2008	2.0	4	manual(m6)	Front wheel	compact	Premium	…
audi	a4	2008	2.0	4	auto(av)	Front wheel	compact	Premium	…
audi	a4	1999	2.8	6	auto(l5)	Front wheel	compact	Premium	…

How mammals sleep

83 rows · 8 columns · 4 KB

One row is: one mammal species
Good numbers to plot: sleep_hours, dream_sleep_hours, body_mass_kg, brain_mass_kg
Groups you can split by: diet
Also in the file: animal, sleep_cycle_hours, awake_hours
You could ask: Do animals that sleep more also dream more? Do carnivores sleep longer?

Download CSV

Peek at the data

animal	diet	sleep_hours	dream_sleep_hours	sleep_cycle_hours	awake_hours	brain_mass_kg	body_mass_kg
Cheetah	Carnivore	12.1	—	—	11.9	—	50.0
Owl monkey	Omnivore	17.0	1.8	—	7.0	0.0155	0.48
Mountain beaver	Herbivore	14.4	2.4	—	9.6	—	1.35
Greater short-tailed shrew	Omnivore	14.9	2.3	0.133333333	9.1	0.00029	0.019
Cow	Herbivore	4.0	0.7	0.666666667	20.0	0.423	600.0

Getting your question right

This is where this task actually goes wrong, so spend a minute on it. Use this shape:

Is there a relationship between ______ and ______ in ______ ?

Check all four before you start

It names two columns that both hold numbers.
It asks whether they are related, not what one single value is.
You can answer it from your file, without going and finding more data.
It is not secretly a compare-two-groups question. That is boxplot work, not a scatterplot. Both are fine, you just have to know which one you are doing.

The question	Verdict
✔ Do penguins with longer flippers weigh more?	Two number columns, and it asks about a relationship.
✔ Does a bigger movie budget mean a bigger box office?	Both are numbers, and the file can answer it.
✔ Is court time or height the better predictor of points?	Ambitious and good. Two scatterplots, then compare the two r values.
✘ What is the best Pokemon?	‘Best’ is not a number, and nothing is being related to anything.
✘ How many songs are in the Spotify file?	That is a fact you look up in one second, not an investigation.
! Do midfielders weigh more than ruckmen?	Careful. This compares two GROUPS, so it is boxplots, not a scatterplot. Fine to do, but know which one you are doing.
! Does more court time CAUSE more points?	Careful. Your data can show two things go together. It can never prove one causes the other. Ask whether they are related, then discuss cause separately.
✘ Does eating fast food make people unhealthy?	There are no people in that file, only menu items. Check your file actually contains the thing you are asking about.

How to plot it

We are using CODAP. It is free, it runs in the browser, there is no login and nothing to install, so it works the same on a Chromebook as on a lab computer.

CODAP

codap.concord.org/app · free, any browser, no login, nothing to install

Open the data. Drag your CSV onto the CODAP window. Or use the menu button, choose Import, then the Local File tab. Your data appears as a table.
You can also use Copy link for CODAP above and import straight from the web.
Make an empty graph. Click Graph in the top toolbar.
Drag your explanatory variable onto the horizontal axis, across the bottom.
Drag your response variable onto the vertical axis. That is your scatterplot.
Add the line. Click the ruler icon in the strip of icons down the right of the graph, then tick Least Squares Line. You get the equation and r in a yellow box, which you can drag if it covers your dots.
Colour by group (optional, looks great). Drag a word column onto the middle of the graph.
Screenshot it, with the line and the box of numbers both readable, because you write the rest up off that picture. Chromebook: Ctrl + Show Windows, or Ctrl + Shift + Show Windows to drag a box around just the graph. Windows: Win + Shift + S. Paste it into your template at Question 5.

Stuck? If a column will not go on an axis, click its name and choose Treat as Numeric. If nothing happens when you drag, you forgot to make the graph first.

Reading r

CODAP prints r straight into the yellow box beside your line, so there is no working to do. Read it off.

Then check it against your own eyes. A plus sign means the line runs uphill, a minus sign downhill. If the sign and the picture disagree, you have read the wrong number.

And remember: the sign only tells you the direction. Strength is how far r sits from zero, so r = −0.9 is strong and r = +0.2 is weak.

Strength bands, ignoring the sign: 0.75 to 1 strong, 0.5 to 0.75 moderate, 0.25 to 0.5 weak, below 0.25 nothing worth talking about. A guide, not a law: 0.74 and 0.76 are not different animals, and a curved plot can carry a healthy r and still deserve no straight line. Look at the picture as well as the number.

If your box shows R squared instead, older versions do that. Square root it and take the sign off the slope: R squared of 0.67 gives r = 0.82, the penguin number from class.

SALT, before you call it finished

Scale, evenly spaced, on both axes.
Axes, explanatory variable across the bottom.
Labels, both axes named, with units.
Title, saying what the graph is of.

Find your own data

Finish the investigation first. Then, if you want to go hunting, this is the big public list:

github.com/awesomedata/awesome-public-datasets

Fair warning, and this is genuinely what working with data feels like: a lot of those links are dead, a lot of the files are far too big to open, and a file that will not load is not a maths problem you can solve. If that link is blocked at school, try it at home.

What makes a dataset usable

It is a .csv file you can actually download.
It has two columns of numbers that might be related.
It is small. A few thousand rows at most, or it will not open.
One row means one thing, and you can say what that thing is.

Get it approved before you commit to it.