Lesson 2 of 10 · 18 min
Interpreting the bystander research
Identify the variables, design, controls and ethical issues in Latané and Darley's studies, judge their validity, and analyse bystander data with means, standard deviations and a t-test.
What you will be able to do
- Identify and operationalise the IV and DV and write hypotheses for a bystander study.
- Identify the design, controls and levels of measurement in Latané and Darley's studies.
- Evaluate their ethics and their internal and external validity.
- Analyse bystander data with descriptive statistics and an unpaired t-test, and draw a conclusion.
"Interpret the findings" means more than repeating the numbers. You need to say what the studies measured, how well they measured it, and how far the results can be trusted and generalised. This lesson works through the seizure study as a model, then analyses a class-style dataset.
| Feature | In the seizure study |
|---|---|
| Aim | To find out whether the number of bystanders a person believes are present affects whether, and how quickly, they help in an emergency. |
| IV | Number of other people the participant believed could hear the seizure: none, one or four. |
| DV, operationalised | Whether the participant left the booth to report the seizure before it ended, and the time in seconds from the start of the seizure until they did. |
| Design | Experiment with independent groups: each participant was in one condition only. |
| Controls | The same recorded voices and the same seizure for everyone (standardised procedure); participants were alone in booths and could not see each other; the "others" were recordings, so their behaviour was identical. |
| Levels of measurement | Helped or not: nominal. Time to help: ratio (a true zero). |
| Sample | 72 New York University students in an introductory psychology course: a convenience sample. |
Which is the best alternative hypothesis for the seizure study?
Definition
Deception and debriefing
Deception in research is misleading participants about the purpose or procedure of a study. It is allowed only when the study could not be done otherwise, the deception causes no lasting harm, and participants are debriefed: told the truth, given the chance to ask questions and to withdraw their data, and checked to make sure they leave no worse off.
Ethics. All three studies used deception: the seizure was a recording, the smoke was harmless and the crash was a tape. Participants could not give fully informed consent, and some were clearly distressed while they listened. In their favour, the researchers debriefed each participant straight away, explained the reasons for the deception and dealt with any distress; in the seizure study every participant afterwards said the deceptions were necessary and justified (Darley & Latané, 1968). A modern ethics committee would weigh the distress against the value of the knowledge and would require careful debriefing and the right to withdraw data.
Validity. Internal validity was strong: the procedure was standardised and the only thing that differed between conditions was the number of bystanders, so differences in helping can be put down to the IV. External validity is more limited. The samples were American university students, the emergencies were staged in a laboratory or office, and in the seizure study the "bystanders" were voices, not people the participant could see. A meta-analysis of four decades of bystander research (Fischer et al., 2011) found that the bystander effect is reliable, but smaller in clearly dangerous emergencies, where others can be a source of physical support.
Mark each step of the smoke-filled room procedure as acceptable or name the principle it breaches.
Which ethical principle did the smoke-filled room study most clearly have to manage, given its design?
Analysing bystander data. A modern class cannot stage an emergency, but it can test the same idea ethically with a vignette: a short written scenario. In this illustrative study, 20 Year 12 students were randomly allocated to read a vignette about a cyclist who falls off their bike on a footpath. In one version the reader is alone on the street; in the other, ten other people are walking past. Each student rated how likely they would be to stop and help, from 0 (definitely not) to 100 (definitely).
| Condition | Ratings | Mean | SD |
|---|---|---|---|
| Alone (n = 10) | 78, 85, 70, 92, 66, 81, 74, 88, 79, 83 | 79.60 | 8.02 |
| Ten others present (n = 10) | 60, 72, 55, 68, 74, 50, 63, 58, 70, 65 | 63.50 | 7.78 |
Illustrative vignette data. Independent groups, so the unpaired t-test applies. Try switching the error bars between SD, SE and 95% CI.
Worked example
Test a bystander vignette study with an unpaired t-test
Twenty Year 12 students were randomly allocated to read one of two versions of a vignette about a cyclist who falls on a footpath, and rated how likely they would be to help (0–100). Illustrative data:
| Condition | n | Mean | SD |
|---|---|---|---|
| Alone | 10 | 79.60 | 8.02 |
| Ten others present | 10 | 63.50 | 7.78 |
(a) State the null hypothesis. (b) An unpaired t-test gives , . Draw a conclusion. (c) Identify one limitation of the study.
Step 1: (a) Null hypothesis
There will be no difference in the mean likelihood-of-helping rating between students who read the alone vignette and students who read the vignette with ten others present.
Step 2: Check the test suits the design
Each student read one vignette (independent groups), so the scores are unpaired: an unpaired t-test with .
Step 3: (b) Compare p with 0.05
, so the difference is statistically significant and the null hypothesis is rejected.
Step 4: (b) State the conclusion in context
Students who imagined being alone rated themselves significantly more likely to help (mean 79.60) than those who imagined ten others present (mean 63.50), a difference of 16.10 points. This supports the bystander effect on helping intentions.
Step 5: (c) Limitation
A vignette measures what students say they would do, not what they actually do, so ecological validity is low. The sample (20 students from one school) is also small and not representative.
Try it yourself first, then check the working one step at a time.
Answer
(a) No difference in mean ratings between the two vignettes. (b) : reject the null; the alone group rated helping significantly more likely (79.60 vs 63.50). (c) It measures intentions, not behaviour (low ecological validity); small convenience sample.
Method
Match the test to the design (independent groups → unpaired).
Compare with 0.05 and decide about the null hypothesis.
State the conclusion with the means, in the study's terms.
Give a limitation specific to the study.
Watch out for
Saying the result "proves" the bystander effect. A significant result supports a hypothesis; it does not prove it, and a type I error is possible.
Using a paired test for independent groups.
Common mistake
A vignette measures intentions, not behaviour
Students in a vignette study say what they would do. Latané and Darley's participants showed what they did, and most people predict that they would help far more than they actually do. A significant result in a vignette study supports the bystander effect on intentions; it is weaker evidence about real behaviour (lower ecological validity).
In the vignette study, why is the unpaired t-test the right test rather than the paired t-test?
Exam tip
Interpreting data in the exam
State the result with the numbers (means, percentages, , ), say whether it is significant (), relate it to the hypothesis ("so the null hypothesis is rejected"), and then comment on what limits the conclusion: sample, design, ecological validity or ethics.
Summary
Seizure study: IV the believed number of bystanders; DV whether and how fast the participant reported the seizure; independent groups; standardised recordings.
All three studies used deception, so debriefing was essential. Distress and the lack of informed consent are their main ethical issues.
Internal validity was high; external validity is limited by student samples and staged emergencies. The effect is reliable but smaller in dangerous emergencies (Fischer et al., 2011).
Independent groups give unpaired data: compare means with an unpaired t-test and conclude using .
Vignette studies are ethical but measure intentions, not behaviour.
Finished the lesson?
Try the 3 quick checks above and the lesson completes itself, or mark it complete now.