Lesson 4 of 9 · 20 min

Interpreting heart rate data from the experiment

Turn heart rate readings into changes, means, standard deviations and standard errors, graph them with error bars, run the paired t-test and write a conclusion with its limitations.

What you will be able to do

  • Calculate the change in heart rate for each participant and the mean change in each condition.
  • Calculate and interpret the sample standard deviation and the standard error.
  • Read a column graph with error bars.
  • Choose the paired or unpaired t-test from the design and use p to draw a conclusion.
  • Identify the limitations of the data and the possibility of type I and type II errors.

A Year 11 class ran the experiment from the last lesson as repeated measures: ten students each watched a 3-minute scary film clip and a 3-minute toothpaste advertisement, in counterbalanced order, with a rest between. Heart rate was counted for 60 seconds at rest before each clip and in the last minute of each clip. These are their results (made up for practice, but typical of a class).

Heart rate (bpm) before and during each clip
ParticipantEmotive: restEmotive: duringInformative: restInformative: during
P172867173
P265796665
P380887881
P470917274
P568756968
P675847477
P762706364
P877957678
P969787070
P1074807376

Step 1: the change for each participant. The DV is the change: heart rate during minus heart rate at rest. P1's emotive change is 86−72=1486 - 72 = 14 bpm; P2's informative change is 65−66=−165 - 66 = -1 bpm. Keep the minus sign: P2's heart rate fell slightly.

What is P4's change in heart rate while watching the emotive clip?

in bpm
Changes in heart rate (bpm)
ParticipantEmotive changeInformative changeDifference (emotive − informative)
P114212
P214−115
P3835
P421219
P57−18
P6936
P7817
P818216
P9909
P10633

Step 2: the mean change in each condition. Add the changes and divide by n=10n = 10:

xˉemotive=11410=11.40 bpm,xˉinformative=1410=1.40 bpm\bar{x}_{\text{emotive}} = \frac{114}{10} = 11.40\text{ bpm}, \qquad \bar{x}_{\text{informative}} = \frac{14}{10} = 1.40\text{ bpm}

Step 3: the spread. The sample standard deviation measures how far the changes typically sit from their mean:

s=∑(x−xˉ)2n−1s = \sqrt{\frac{\sum (x - \bar{x})^2}{n - 1}}

For the emotive changes ∑(x−xˉ)2=232.40\sum(x - \bar{x})^2 = 232.40, so s=232.40/9=5.08s = \sqrt{232.40 / 9} = 5.08 bpm. For the informative changes s=1.58s = 1.58 bpm. People reacted to the scary clip by very different amounts (6 to 21 bpm), but almost everyone barely reacted to the advertisement.

Tip

On the calculator

Enter the changes as a list and use one-variable statistics. Take the sample standard deviation: sxs_x on a Casio, SxSx on a TI. The population value, σx\sigma_x, divides by nn and is a little smaller (4.82 here instead of 5.08).

Step 4: the uncertainty in each mean. The standard error says how much the mean would vary from one sample of ten to another:

SE=sn=5.0810=1.61 bpm (emotive),1.5810=0.50 bpm (informative)SE = \frac{s}{\sqrt{n}} = \frac{5.08}{\sqrt{10}} = 1.61\text{ bpm (emotive)}, \qquad \frac{1.58}{\sqrt{10}} = 0.50\text{ bpm (informative)}

An approximate 95% confidence interval is xˉ±1.96 SE\bar{x} \pm 1.96\,SE: for the emotive clip 11.40±3.1511.40 \pm 3.15, from 8.25 to 14.55 bpm.

Mean change in heart rate for each clip. Error bars show ±1 standard error (n = 10).

Key idea

Reading error bars

A graph must say what its error bars show: standard deviation (the spread of the scores), standard error (the uncertainty in the mean) or a 95% confidence interval. When SE bars are far apart, as here, the difference is likely to be real; when they overlap, it may not be. The t-test is what decides.

The informative changes have s=1.58s = 1.58 bpm and n=10n = 10. Calculate the standard error, to 2 decimal places.

in bpm

Round to 2 decimal places.

Step 5: the right t-test. The same ten people did both conditions, so each person's two scores are linked: the data are paired. The paired t-test works on each person's difference (the last column of the table): mean dˉ=10.00\bar{d} = 10.00 bpm, standard deviation sd=5.27s_d = 5.27 bpm.

t=dˉsd/n=10.005.27/10=10.001.67=6.00,df=n−1=9t = \frac{\bar{d}}{s_d / \sqrt{n}} = \frac{10.00}{5.27 / \sqrt{10}} = \frac{10.00}{1.67} = 6.00, \qquad df = n - 1 = 9

Software gives p=0.0002p = 0.0002 (two-tailed). Had the class used independent groups, with different people watching each clip, the unpaired t-test would be right instead, with df=n1+n2−2df = n_1 + n_2 - 2.

Interactive

Test the class data

The paired lines show each student's two changes. Switch the error bars between SD, SE and the 95% CI.

Key idea

The conclusion

p=0.0002<0.05p = 0.0002 < 0.05, so reject the null hypothesis. The mean change in heart rate was significantly greater for the emotive clip (11.40 bpm) than for the informative clip (1.40 bpm), t(9)=6.00t(9) = 6.00, p<0.05p < 0.05. Watching the scary clip produced more sympathetic arousal than watching the advertisement.

A second class finds a mean difference of 2.1 bpm, t(11)=1.48t(11) = 1.48, p=0.17p = 0.17. Which conclusion is correct?

Choose one answer

Worked example

Mean, standard deviation and standard error of heart rate changes

  • Simple familiar
  • Technology active

Six students' changes in heart rate (bpm) while watching a scary clip were 12, 8, 15, 9, 6 and 10. Calculate the mean, the sample standard deviation and the standard error, each to 2 decimal places.

4 steps

Try it yourself first, then check the working one step at a time.

Method

  1. Mean: sum ÷ n.

  2. SD: square the deviations, sum, divide by n − 1, square root.

  3. SE: SD ÷ √n.

Watch out for

  • Dividing by nn (giving 2.89, the population SD).

  • Rounding the SD to 3.2 before finding SE.

Worked example

Run the paired t-test on six students

  • Complex familiar
  • Technology active

The same six students also watched a toothpaste advertisement. Their changes in heart rate (bpm) were:

StudentS1S2S3S4S5S6
Emotive128159610
Informative213−122

Choose and carry out the appropriate t-test, given that p=0.0007p = 0.0007 (two-tailed), and state the conclusion.

5 steps

Try it yourself first, then check the working one step at a time.

Method

  1. Same participants in both conditions → paired test.

  2. Work on each participant's difference: mean, SD, then t=dˉ/(sd/n)t = \bar{d} / (s_d/\sqrt{n}), df=n−1df = n - 1.

  3. Compare p with 0.05 and state the conclusion in terms of the variables.

Watch out for

  • Using the unpaired test, which ignores the pairing.

  • Using df=n1+n2−2=10df = n_1 + n_2 - 2 = 10 for a paired test.

Common mistake

Errors and limitations

  • Type I error: rejecting H₀ when it is really true (a false alarm). With α=0.05\alpha = 0.05 this happens in about 1 in 20 tests of a true null.
  • Type II error: retaining H₀ when there really is an effect (a miss). Small samples make this more likely.
  • Limitations to mention: a small convenience sample of one class (it may not represent the population); counting a pulse by hand is imprecise; heart rate rises with many emotions (and with surprise or movement), so it shows arousal, not fear itself; the t-test assumes the differences are roughly normally distributed, hard to check with ten people.

Summary

  • Change = heart rate during − resting heart rate, sign kept. Analyse the changes.

  • Mean, sample SD (n−1n - 1) and SE = s/ns / \sqrt{n}, each to 2 decimal places with units.

  • Column graphs show the means with error bars, and say whether the bars are SD, SE or a 95% CI.

  • Repeated measures and matched participants: paired t-test on the differences, df=n−1df = n - 1. Independent groups: unpaired t-test, df=n1+n2−2df = n_1 + n_2 - 2.

  • p < 0.05: reject H₀ (significant). p ≥ 0.05: retain H₀. Neither proves anything; type I and II errors remain possible.

  • State limitations: sample size and sampling, measurement precision, what heart rate does and does not measure.

Finished the lesson?

Try the 3 quick checks above and the lesson completes itself, or mark it complete now.

Lesson 4 of 90% read

Save your progress with a free account