ACARA v9 CONTENT DESCRIPTION “select and use equipment to generate and record data with precision to obtain useful sample sizes and replicable data, using digital tools as appropriate”
Builds on choosing equipment and reading it with care. At Year 10 the aim is data others can trust: matching a digital tool to the change you study, telling fine resolution apart from a stable reading, taking enough repeats for a useful sample size, and recording the method and every run so an independent group can replicate the result.
Pick the tool that matches the change, then prove it is stable
Suppose you measure how fast a trolley accelerates down a ramp. A pair of light gates wired to a timer triggers the instant the trolley breaks each beam, recording the interval to a thousandth of a second, far ahead of a stopwatch thumbed by a slow human hand. That fine resolution is a good start, but it is not the whole story. Resolution is how finely the tool can split a reading; precision is how closely repeats of the same run agree. A sensor can show many decimals and still wobble in the last few, so a single logged run never settles the answer. The cure is to repeat the run and look at the spread.
Measured acceleration as repeats are added
The same ramp and trolley were timed many times. The bars show the running mean of the measured acceleration after 1 run, then 3, 5, 8 and 12 runs. Watch it settle.
After a single run the mean is whatever that one run gave. As repeats are added the running mean stops jumping and settles near 204 centimetres per second per second. A larger sample size pulls the mean toward the true value and lets one odd run sway it far less, no matter how many decimals the sensor can show.
Sample size, repeats and the figure you report
Each run carries a little random scatter from tiny differences you cannot control: a slightly different release, a speck of dust on the track, a gate nudged a hair. Average a handful of runs and that scatter partly cancels; average many and it cancels further, so the mean settles close to the true value. This is why a larger sample size gives a steadier, more reliable result, even when each single reading already looks precise. A data logger makes the repeats cheap: it captures every run at the same trigger, evenly and without a biased hand, so you can build a useful sample quickly. Recording the real value from every run, rather than rounding them to look alike, lets you average them honestly.
Twelve logged runs of the same ramp test
Each bar is one logged run of the measured acceleration. The runs cluster, but no two land on the same value. Switch between the table and a graph.
The twelve runs spread from 201 to 207 with no single fixed answer, even though the sensor reported each to a fine resolution. That scatter is normal. Because every run was logged honestly, you can average the whole sample to about 204 rather than trusting any one run alone.
Spot the run that does not belong
Even with careful logging, one careless moment can plant an odd value: a gate knocked out of line, a trolley given a push instead of a clean release, or a beam broken by a stray hand. A run that sits well outside the cluster of the others is the cue to check the apparatus and the record on the spot, while the gear is still set up, rather than after it is packed away. The odd value is noted and investigated, never quietly deleted, so the sample you report stays honest.
A series of logged accelerations: find the stray run
Twelve runs of the same ramp test were logged in order. One value sits far outside the cluster, most likely a knocked gate or a trolley that was pushed.
Click the point that does not fit the pattern of the others.
What makes data replicable?
Data is replicable when another group, following your method with their own equipment, can reproduce a result that agrees with yours within reason. That independent agreement is what builds trust in a finding. Replication depends on choices made while gathering the data: a faithful record of the apparatus and its geometry, the logger settings, an honest log of every run, and a sample large enough to be meaningful. Sort each practice below by whether it genuinely makes the ramp investigation replicable, or whether it would quietly make the result impossible to reproduce.
Judge the case for replicable data
The claim: this investigation produces replicable data that another group could reproduce. Decide which practices support it.
Claim: This ramp investigation produces data that an independent group could replicate.
The ramp angle, the gate spacing and the logger trigger settings are written down with the results.
Every logged run is kept, including the ones that came out a little high or low.
Only the three runs nearest the textbook acceleration are reported, and the rest are dropped.
Enough runs are logged to form a useful sample, and the full set is averaged.
The method is left vague so each group can quietly set the ramp up its own way.
Decide whether each statement is evidence for the claim, or not.
Why this matters
A conclusion is only as sound as the data behind it, and sound data is more than a sensor with many decimals. It comes from a tool matched to the change being measured, enough repeats to form a useful sample, an honest record of every run, and a method clear enough that others can reproduce the result. Digital tools make the repeats easy to gather and log without bias, but the judgement about resolution, sample size and replicability stays with the investigator.
Quick self-check
1. A trolley rolls down a ramp and you want the acceleration measured precisely. Which digital tool suits the job best?
2. Your motion sensor reports speed to 0.001 metres per second, yet the same run logged twice differs in the third decimal. What does this tell you?
3. You log the ramp run three times, then ten times. What does the larger sample size mainly give you?
4. A different class repeats your method with their own light gates and trolley and lands on a close acceleration. This shows your data is...
5. To let another group replicate your ramp investigation, what should the record contain?