All videos › Biology › Grade 12
Analysing Biological Data - Mean, Spread and the t-Test Two groups of made-up seedlings differ by four centimetres. Is that a real difference? Use the mean, the standard deviation and the idea behind a t-test.
Sign in to save your place
Biology · Grade 12 · Doing biology: methods and ethics · 3.5 minutes · UK English
Transcript 0:02 Five made-up seedlings were given fertiliser, and five were not.0:07 The fertilised ones grew taller on average.0:10 Is that a real effect, or just chance?0:13 You already know how to find a mean,0:15 and how a fair experiment uses a control group.0:18 In this video, you will describe data with a mean and0:22 a standard deviation, and see the idea behind a t-test and0:26 its p value.0:28 The mean tells you the centre of the data.0:31 The standard deviation tells you the spread,0:33 roughly how far the values sit from their mean.0:37 To find it for a sample, square each distance from the0:40 mean and add them up. Divide by one less than the0:43 number of values. Then take the square root.0:46 A small standard deviation means the values cluster tightly.0:50 A large one means they are spread out.0:53 Here is the made-up data. The fertilised heights are thirteen,0:58 fourteen, fifteen, sixteen and seventeen centimetres. The mean is fifteen.1:03 The distances from the mean are two,1:06 one, zero, one and two. Their squares add to ten.1:09 Divide by four to get two point five.1:12 The square root is about one point six centimetres.1:16 The unfertilised heights are nine to thirteen centimetres.1:20 Their mean is eleven, and their squares also add to ten,1:24 so the spread is the same.1:27 The gap between the means is four centimetres.1:30 A t-test asks how big that gap is compared with the1:34 noise. Divide two point five by five for each group,1:38 add them and take the square root.1:41 That standard error is one. So t is four divided by1:45 one, which is four. The gap is four times the noise.1:49 Next, the test turns t into a p value.1:52 A p value assumes there is no real effect.1:56 It asks how often chance alone would give a gap at2:00 least this big. Here, a t of four gives a small2:03 p value. That is strong evidence against the idea of no2:08 real effect.2:09 Your turn. A made-up study of thousands of plants finds a2:13 one millimetre height gap, with p equal to zero point zero2:18 zero one. Which statement is correct? Is it,2:21 a gap this big would be rare if there were no2:24 effect? The fertiliser effect must be both large and important?2:28 Or, there is a zero point one percent chance that there2:32 is no real effect? Pause the video and decide.2:36 Then watch for the answer.2:46 The answer is the first choice. The p value starts by2:50 assuming there is no real effect, so it cannot be the2:54 chance that there is no effect. The common mistake is to2:58 read a small p value as a large or important effect.3:02 A gap of one millimetre may not matter to a gardener3:06 at all. With a big enough sample,3:08 even a tiny difference can give a very small p value.3:13 Here is one tip to keep. Report the size of the3:16 difference, not just the p value. In our example,3:19 say that the fertilised seedlings were four centimetres taller on average,3:24 give the spread and the number in each group,3:27 and then give the result of the test.3:30 Let us recap. The mean gives the centre and the standard3:34 deviation gives the spread. A t-test compares the gap between means3:39 with the noise. A p value says how surprising the gap3:43 would be if there were no real effect.3:46 It does not measure importance. Thank you for learning with Worked3:51 Example School.Spotted a mistake? Tell us at contact@workedexampleschool.com .