The history of social psychology is the history of attempts to explain how other people affect what we notice, believe, say and do. Its strongest contributions turn an impression about social life into a comparison that can be checked: what changed in the situation, what changed in behaviour, and which alternative explanations remain? The important lesson is not that people always conform or that groups always make poor decisions. It is that social conditions belong inside an explanation of behaviour.
This guide follows selected primary sources from competition research in 1898 through conformity, cognitive dissonance, bystander research, group categorisation and collaborative replication. It is a history of questions and methods, with practical exercises, rather than an exhaustive chronology of every school or researcher. All numerical examples are invented teaching cases unless expressly identified as published findings. They are not reconstructions of the original experiments.
Use the Gold Standard master for the wider collection, Education: learning transfers into life for the purpose of learning, and the How X Works hub for connected mechanisms. The existing Social Psychology root explains the subject itself. This historical guide asks how particular questions became research programmes, and how to read their evidence responsibly.
Choose your social psychology history route
Chapters 1–3: Start with social questions and comparisons
Chapters 4–6: Examine attitudes, responsibility and groups
Chapters 7–9: Test replication, beliefs and group output
Chapters 10–12: Evaluate causes and transfer the method
Chapter index
Start with social questions and comparisons · Chapters 1–3
1. Begin with a social question that could have more than one answer
Examine attitudes, responsibility and groups · Chapters 4–6
4. Festinger and Carlsmith in 1959: actions and attitudes
5. Darley and Latané in 1968: responsibility in a social situation
Test replication, beliefs and group output · Chapters 7–9
7. Replication: a finding becomes a question again
8. Public silence and private belief: measure the missing quantity
9. Group output: effort, coordination and information are different problems
Evaluate causes and transfer the method · Chapters 10–12
10. Causal claims: what exactly did the comparison change?
11. Ethical reading: historical importance is not permission to reenact
12. Changed-context transfer: a committee chooses a community project
Use questions and primary sources · Chapters 13–14
CHAPTER 1 OF 14 · Start with social questions and comparisons
1. Begin with a social question that could have more than one answer
Imagine a meeting in which nobody challenges a proposal. One observer concludes that everyone agrees. Another concludes that everyone is afraid. A third thinks nobody has read the proposal. The visible behaviour is compatible with all three explanations. Social psychology becomes useful when it helps design an observation that distinguishes them, rather than supplying a memorable label for whichever explanation feels plausible.
Separate three things in your notes: what happened, what someone inferred, and what additional evidence would change the inference. “Nobody spoke during the invitation for questions” is an observation. “Everybody understood” is an inference. Independently collected explanations of the proposal would provide different evidence. The distinction sounds simple, yet it prevents a common failure: treating an interpretation of behaviour as if it were the behaviour itself.
People respond to meanings as well as physical arrangements
Two rooms can contain the same number of people and still create different social situations. In one, participants believe that their answers will remain private. In another, they expect a supervisor to read every answer aloud. The furniture is identical; the anticipated audience is different. An experiment needs to specify what participants were told, what they appeared to understand and what consequence they expected.
This creates a methodological challenge. An instruction intended to change one feature may change several. Asking people to explain their answers publicly can alter accountability, anxiety, effort and the desire to appear consistent. A difference in performance would show that the arrangements mattered; it would not immediately identify which psychological pathway produced the difference.
Read historical research with that distinction in mind. A controlled comparison can support a causal claim about a procedure while leaving its mechanism partly unresolved. Follow-up studies can vary the procedure or measure intermediate processes. The historical story is therefore not simply a succession of discoveries with final names. It includes the construction of better questions, competing interpretations and improved ways to tell those interpretations apart.
For a broad methodological route, How Psychology Works connects observations, constructs, measurements and explanations. Here those distinctions will be applied to situations in which the behaviour of one person becomes part of the environment of another.
CHAPTER 2 OF 14 · Start with social questions and comparisons
2. Triplett in 1898: competition becomes a research problem
Norman Triplett’s The Dynamogenic Factors in Pacemaking and Competition considered cycling records and laboratory work involving a reel-turning task. His reported child data included different responses to competition, rather than a uniform improvement in every participant. The paper also described arrangements intended to address practice across trials. Reading the original reveals a more interesting problem than the slogan that other people make us faster: performance depends on the task, the comparison and the individual response.
Original case: faster does not necessarily mean better
Suppose a person sorts forty cards alone in sixty seconds with no mistakes. With a competitor present, the same person sorts forty cards in fifty seconds but makes four mistakes. The time has fallen by ten seconds, or 16.7% of the original time. Yet correct output has fallen from forty to thirty-six cards. Reporting only speed would hide part of the result.
Correct cards per second are 40 ÷ 60 = 0.667 alone and 36 ÷ 50 = 0.72 in competition. Under this particular efficiency measure, performance has improved slightly. If every mistake requires ten seconds of later correction, however, the competitive condition requires fifty seconds plus forty seconds of repair. Its effective completion time becomes ninety seconds. The conclusion depends on the outcome that matters.
This does not mean that a researcher may choose any outcome after seeing the data. Define the relevant measure before comparing conditions, and retain enough detail to understand trade-offs. A task involving harmless practice cards has different consequences from one in which an error is costly. The same measured change can have different practical value.
Practice and order need their own comparison
If everyone works alone first and competes second, improvement could partly reflect familiarity. Reversing the order for some participants helps investigate this possibility, although fatigue and carry-over still require attention. The historical importance of such design questions is that the social explanation must earn its place against ordinary alternatives.
Keep the original research question attached to its procedure. Cycling records, a laboratory motor task and a modern collaborative workplace are different settings. A historical observation can inspire a new hypothesis about the workplace, but it does not supply the workplace result. That distinction preserves the value of the earlier work without asking it to establish more than it measured.
CHAPTER 3 OF 14 · Start with social questions and comparisons
3. Asch in 1955: agreement, independence and public answers
Solomon Asch’s Opinions and Social Pressure described line-judgement experiments in which a participant encountered a group giving arranged incorrect answers. The account reported both conformity and independence, and variations involving the majority’s size and unanimity. Its historical contribution is the controlled contrast between a judgement task and the social conditions surrounding the answer. It does not establish that every person abandons an evident truth whenever a majority disagrees.
Original calculation: trials and people are different denominators
Imagine twenty participants each answering ten critical questions. There are two hundred responses. Forty responses agree with an incorrect majority, giving a response-level conformity rate of 20%. That figure alone does not tell us how many participants ever conformed. Four people might have conformed on all ten questions, or all twenty might have conformed on two questions each.
In the first arrangement, 20% of participants conformed at least once. In the second, 100% did. Both produce exactly the same response-level rate. A report must state whether its denominator is people, trials or groups. Confusing those quantities can turn a limited result into a dramatic but false claim about human nature.
Repeated answers from the same person are also related observations. Treating two hundred answers as if they came from two hundred independent people can exaggerate the amount of information in the study. The design and analysis should account for the level at which observations are clustered.
A public statement is not a complete reading of private belief
A person may give an answer because they believe it, because they doubt themselves, because they want to avoid conflict, or because they misunderstand the task. A public answer alone cannot fully distinguish those explanations. Private judgements, explanations and carefully designed variations can add evidence, while each introduces its own measurement limits.
For an everyday application, collect initial estimates independently before group discussion. Then record what changed and why. This is a transparent decision procedure, not a recreation of a deceptive historical experiment. It helps a team distinguish agreement that incorporates new evidence from agreement produced by repeating the first confident speaker. The goal is better reasoning, not celebrating disagreement for its own sake.
CHAPTER 4 OF 14 · Examine attitudes, responsibility and groups
4. Festinger and Carlsmith in 1959: actions and attitudes
Leon Festinger and James Carlsmith’s Cognitive Consequences of Forced Compliance examined attitudes after participants performed a boring activity and were paid to describe it positively to another person. The experiment varied payment and compared subsequent evaluations, testing predictions from cognitive dissonance theory. The reported pattern challenged a simple assumption that greater payment must produce greater private attitude change. The paper concerns a specified laboratory arrangement, not a rule that small rewards always improve motivation or honesty.
Original case: the same changed opinion has several possible causes
A volunteer initially dislikes a project, publicly endorses it, and later reports liking it more. Perhaps the volunteer is bringing an attitude into line with a chosen action. Perhaps participation provided new information. Perhaps the volunteer thinks the interviewer wants a positive answer. Perhaps the initial rating was unusually negative and would have changed anyway. The sequence alone cannot select the mechanism.
To investigate, define the original attitude, the action and the later outcome. Distinguish public endorsement from private evaluation. Record whether new experiences occurred between measurements. A comparison group can help, but it must experience a sufficiently similar setting for the contrast to answer the intended question. Merely having two groups does not guarantee an informative experiment.
Suppose one group changes from an average rating of three to five on a seven-point scale, while a comparison group changes from three to four. The first group’s observed increase is two points; the difference between increases is one point. Interpreting that one-point contrast requires assumptions about assignment, measurement and other experiences. It also requires care about the scale: a two-point change is not automatically twice the psychological change of a one-point increase.
Explanations should make risky predictions
A useful explanation predicts something that could fail to occur. If every possible response is described as evidence of dissonance, the explanation has stopped discriminating between outcomes. Specify the conditions under which the proposed tension should arise, what alternative response is possible, and which result would weaken the account.
The everyday lesson is a habit of self-examination. When your opinion changes after you have defended a decision, ask what new evidence supports the change. Changing one’s mind can be sensible; maintaining it can also be sensible. The quality of the reasoning depends on the evidence, not on whether the final position resembles the first one.
CHAPTER 5 OF 14 · Examine attitudes, responsibility and groups
5. Darley and Latané in 1968: responsibility in a social situation
John Darley and Bibb Latané’s 1968 Bystander Intervention in Emergencies: Diffusion of Responsibility investigated responses to a staged emergency when participants believed different numbers of other listeners were present. The reported individual response varied with the perceived social arrangement. The study made responsibility and information about other observers experimental questions. It should not be converted into the claim that crowds never help, or used to infer a particular witness’s motives from an outcome alone.
Original calculation: an individual’s probability is not the group’s probability
Consider a deliberately simplified model. If each of four people independently has a 30% chance of acting, the probability that nobody acts is 0.7 to the fourth power, or 0.2401. The chance that at least one acts is therefore 1 − 0.2401 = 0.7599, about 76%. An individual probability of 30% does not imply a group probability of 30%.
The independence assumption is crucial. People might watch one another, share uncertainty or wait for a signal. Their actions could be strongly dependent. The calculation is not an estimate of real emergency behaviour and must not guide emergency decisions. It demonstrates a statistical distinction between an individual outcome and a collective outcome.
Now consider an ordinary project inbox. Five colleagues receive a request to update a document, but no owner is named. A later failure to update it does not reveal whether everybody lacked motivation. Each might have assumed another person had accepted the task. The organisational repair is to name an owner, confirm acceptance and make completion visible. This application is an explicit design proposal, not a numerical prediction from the historical study.
Mechanism and moral judgement must be separated
Explaining how a situation can affect action does not remove responsibility for action. Equally, moral condemnation does not identify the mechanism that produced a failure. A useful review asks both what obligations existed and what information, resources and coordination were available.
When studying a historical paper, distinguish the experiment from stories that motivated its authors. An introductory anecdote is not the experiment’s evidence, and a controlled experiment does not verify every detail of an anecdote. This distinction matters whenever a vivid story becomes more memorable than the research procedure. Read the actual comparison before repeating the familiar lesson.
CHAPTER 6 OF 14 · Examine attitudes, responsibility and groups
6. Tajfel and colleagues in 1971: categories and allocation
Henri Tajfel, Michael Billig, Robert Bundy and Claude Flament’s Social Categorization and Intergroup Behaviour examined reward allocations under group classifications designed to reduce explanations based on prior hostility or direct individual interest. The reported results included favouring one’s own group and choices sensitive to differences between groups. These findings made categorisation itself a research variable. They do not show that every category necessarily creates hostility, or that a laboratory allocation task explains every historical conflict.
Original allocation exercise
Imagine choosing between two allocations. Option A gives your group twelve tokens and another group ten. Option B gives your group ten and the other group four. Option A provides more for your own group and more overall. Option B provides a larger advantage relative to the other group: six tokens rather than two.
Choosing B therefore sacrifices two tokens for one’s own group while increasing the between-group difference. This arithmetic distinguishes absolute benefit from relative advantage. It does not establish why a person chose B. They might misunderstand the options, believe the tokens have different meanings, or follow a perceived rule. Ask what the design makes each interpretation plausible or implausible.
Suppose ten people choose A and ten choose B. A mean allocation can conceal these contrasting strategies. Report the choices and their structure, not only a total average. If the task is repeated, investigate whether people learn, change their interpretation or respond differently to the labels attached to the recipients.
From laboratory categories to institutions
Real groups have histories, relationships, resources and unequal power. An artificial classification deliberately removes some of that complexity to investigate a narrower question. Applying its result requires restoring the relevant complexity, not pretending that it never matters.
For a fair committee decision, define the allocation objective before presenting group labels: equal shares, demonstrated need, expected benefit or another justified principle. Then inspect whether the decision follows that principle consistently. The appropriate objective is a substantive question that a token task cannot settle.
This chapter’s exercise is useful because the trade-off is visible. A claim about fairness becomes easier to examine when the available options, recipients and consequences are explicit. Social explanations should reveal those relationships rather than using group identity as an all-purpose explanation of any disagreement.
CHAPTER 7 OF 14 · Test replication, beliefs and group output
7. Replication: a finding becomes a question again
The Open Science Collaboration’s 2015 Estimating the Reproducibility of Psychological Science reported a coordinated attempt to replicate one hundred studies drawn from three psychology journals. Its place in this history is the organisation of replication as a shared research project. A selected set of studies is not a random census of all social psychology, and a replication result must be interpreted alongside design, uncertainty and the precise claim being tested.
Original example: two estimates can differ without contradiction
Suppose an original study estimates a ten-point difference between conditions, with substantial uncertainty. A later study estimates a four-point difference. The point estimates differ, but that alone does not establish that the studies contradict each other. We need their uncertainty, procedures and populations. Conversely, two results with the same sign are not automatically strong confirmation if both are extremely imprecise.
A significance threshold creates an additional temptation. An estimate just beyond a threshold and one just short of it can be numerically very similar. Calling one a discovery and the other a complete failure hides that similarity. Compare effect sizes and intervals, and inspect what would matter substantively in the setting.
Replication can test the same procedure in a new sample, or investigate whether a proposed relationship survives a meaningful change in context. These questions overlap but are not identical. If materials or instructions change, explain why. If they remain constant but carry different meanings for participants, physical similarity may not guarantee psychological similarity.
Transparency makes disagreement productive
A reader should be able to identify the planned outcome, the exclusions, the sample, the materials and the analysis. Exploring unexpected patterns is valuable when it is labelled as exploration. Presenting an analysis chosen after seeing the data as if it had been the only planned test creates a misleading impression of certainty.
For students, a useful exercise is to write a short prediction before examining a result. State what would count against it. Then compare the observed outcome with that prediction and retain both in the final report. This practice does not require recreating a famous experiment or collecting sensitive data. It teaches the difference between learning from evidence and constructing a persuasive story after the fact.
CHAPTER 8 OF 14 · Test replication, beliefs and group output
8. Public silence and private belief: measure the missing quantity
Historical conformity research invites a practical question: how can a group discover what its members actually think? Silence is a weak measure because it can reflect agreement, uncertainty, distraction, lack of opportunity or concern about consequences. Counting silent people does not tell us which explanation applies.
The specialist article Pluralistic Ignorance: Private Doubts, Public Silence and False Group Norms examines this distinction directly. Use it alongside the historical material to separate visible conduct from beliefs about what other people believe.
Original survey case
A group of thirty people discusses a proposed schedule. Twenty-four remain silent. It would be incorrect to report 80% approval from that silence. Later, twenty people answer a private survey, and fourteen support the proposal. Support among respondents is 14 ÷ 20 = 70%. Support among all thirty is not known because ten did not respond.
If every nonrespondent opposed the proposal, total support would be fourteen out of thirty, or 46.7%. If every nonrespondent supported it, support would be twenty-four out of thirty, or 80%. Those bounds are wide. They show what the missing responses permit, without pretending that either extreme is likely.
An anonymous questionnaire may reduce some pressures but introduce other limitations. People may misunderstand a question, answer strategically or decline to respond. Explain what is actually anonymous, keep questions specific and report the response rate. Do not promise privacy that the collection method cannot provide.
Repair the decision process, not merely the atmosphere
Before discussion, ask members to record their initial judgement and the evidence they used. During discussion, invite reasons and missing information. Afterwards, distinguish changes caused by new evidence from changes made for practical compromise. A committee may reasonably choose a compromise even when private preferences remain different.
The objective is not to eliminate social influence. Sharing expertise is social influence too. The objective is to make the path from evidence to decision visible enough that confidence, status and repeated statements do not quietly substitute for relevant information.
CHAPTER 9 OF 14 · Test replication, beliefs and group output
9. Group output: effort, coordination and information are different problems
If a team produces less than expected, it is tempting to conclude that its members did not try. Yet weak output may arise from duplicated work, incompatible formats, unavailable resources or a bottleneck in combining contributions. Historical interest in social effects on performance should make us more careful about mechanisms, not quicker to accuse people.
The site’s Social Loafing guide distinguishes individual effort from coordination and task-design problems. The distinction determines which repair could plausibly work.
Original calculation: gross activity and useful output
Four people each produce ten candidate ideas. Their combined lists contain forty entries. If fifteen entries duplicate ideas already present, the group has twenty-five unique ideas. If five of those fail the stated requirements, twenty useful ideas remain. Reporting forty ideas describes gross activity; reporting twenty describes useful unique output under the chosen criteria.
Now a second process produces only thirty entries, but five are duplicates and two fail the requirements. It yields twenty-three useful ideas. Lower activity can accompany higher useful output. Without recording duplicates and the acceptance rule, a team may reward the less effective process.
This example does not prove that any particular brainstorming method works. It defines outcomes that a comparison could measure. To evaluate a method, use comparable tasks, similar resources and a clear procedure for judging ideas. If judges know which method produced each idea, their expectations may affect evaluation; consider a suitably independent review where feasible.
Information can remain private even when everyone participates
A meeting can involve equal speaking time while still missing the decisive evidence. One person may hold an unusual fact that never seems relevant enough to mention. Repeating shared facts creates familiarity without increasing coverage. Ask what new information entered the decision, not just who spoke for how long.
The specialist Hidden Profiles guide develops this problem with a worked case. Together these routes show why group performance needs several measures: effort, coordination, information coverage and final quality. A single participation score cannot identify all four.
CHAPTER 10 OF 14 · Evaluate causes and transfer the method
10. Causal claims: what exactly did the comparison change?
A social intervention often changes several things at once. Suppose a school replaces a public show of hands with a private digital response. More students now disagree with an answer. Privacy may matter, but so might the extra thinking time, the response device or the fact that students can revise before submitting. The observed difference belongs first to the whole procedure.
To investigate a mechanism, vary one relevant feature while holding others reasonably stable, or use a design that can estimate several components. No design is magically free of assumptions. State which comparison is being made and what it can support. Random assignment helps balance pre-existing differences on average; it does not fix a poorly defined outcome or guarantee that the intended instruction was understood.
Original difference-between-changes example
A team introduces independent written estimates before meetings. Its error rate falls from 30% to 18%, a decline of twelve percentage points. Another comparable team retains its procedure and falls from 25% to 20%, a decline of five points. The difference between changes is minus seven percentage points.
Interpreting that contrast as an effect requires more than arithmetic. The teams might face different tasks or experience different staffing changes. If assignment was not random, the assumption that their trends would otherwise have been comparable needs justification. A good report gives the observed comparison and its limitations together.
Do not call a fall from 30% to 18% a twelve-percent reduction. It is a twelve-percentage-point reduction and a 40% relative reduction, because 12 ÷ 30 = 0.4. Both descriptions can be correct, but the denominator must be visible. Relative percentages can sound dramatic when the underlying baseline is small.
Measurement can change behaviour
If participants know that disagreement is being counted, they may disagree more to satisfy the perceived purpose of the exercise. Measure the quality and relevance of reasons, not only the number of objections. The goal of independent judgement is not a maximum disagreement rate. A well-informed group may appropriately converge on the same answer. The measure should reward the intended capability rather than a superficial display of it.
CHAPTER 11 OF 14 · Evaluate causes and transfer the method
11. Ethical reading: historical importance is not permission to reenact
Several studies in this history used deception or situations participants did not fully understand at the outset. Reading them is different from deciding that a classroom, workplace or online community should recreate them. Historical influence does not itself justify a procedure today. A responsible educational activity can teach the inference without placing people under hidden pressure or provoking distress.
Use invented data, transparent role descriptions and voluntary discussion. Explain that a scenario is fictional. Avoid collecting personal disclosures to make a lesson feel realistic. If actual research is proposed, the applicable institutional review and participant protections need to be addressed through the appropriate process; this article is not a substitute for that process.
Original classroom alternative
Give learners three fictional reports of the same meeting. One describes public answers, another describes private initial estimates, and a third describes the evidence shared during discussion. Ask them to identify which conclusions each report supports. No participant needs to be tricked, publicly embarrassed or pressured into an answer.
Then change one fact: the initial estimates were collected after participants saw the leader’s answer. Ask which interpretation becomes weaker. This small change reveals the importance of timing and informational independence. It also teaches why a record labelled private is not automatically independent.
Avoid using concepts as accusations
Calling a colleague conformist, dissonant or biased can shut down the very investigation the concept was meant to support. Describe the observable process instead: “We heard the same evidence repeated, but we have not checked the alternative,” or “The person allocating the work did not know which tasks were already assigned.” Such descriptions are easier to verify and repair.
A social explanation can also be misused to dismiss a disagreement: assuming that another person’s belief merely reflects their group while treating one’s own belief as pure reason. Apply the same evidential questions to your own position. Which information did you encounter, whose judgement did you trust, what alternatives did you inspect, and what would change your mind? Historical knowledge is most useful when it improves those questions rather than furnishing a superior label for other people.
CHAPTER 12 OF 14 · Evaluate causes and transfer the method
12. Changed-context transfer: a committee chooses a community project
Here is an original exercise. A committee of twelve chooses between Project A and Project B. The chair recommends A first. During discussion, eight members praise A and four remain silent. The secretary records unanimous support. After the meeting, a private questionnaire receives eight responses: five prefer A and three prefer B. The committee concludes that public agreement was reliable and that the questionnaire confirms it.
Before reading the answer, identify the unsupported claims. Decide what further information would help and propose a better process for the next decision. Use the historical ideas carefully: do not assume conformity or fear merely because the chair spoke first.
Answer: several quantities have been merged
Eight public endorsements and four silences do not establish twelve endorsements. Five of eight questionnaire respondents prefer A, giving 62.5% support among respondents. Four members did not respond, so total private support is unknown. It could range from five of twelve, or 41.7%, to nine of twelve, or 75%, under the two extreme assumptions about nonrespondents.
The questionnaire was collected after discussion. It therefore cannot recover independent initial preferences. Some people may have learned useful facts during the meeting. Others may have changed their public or private answer for different reasons. The sequence does not isolate a mechanism.
We also lack the evidence used to evaluate the projects. Preference counts are not a substitute for comparing costs, benefits and constraints. A popular option can be poorly supported; a less popular option can also be poorly supported. The process needs both a record of judgements and an examination of their grounds.
A practical repair
Before the next meeting, circulate the same factual brief and ask members to record a provisional preference, uncertainty and missing information. Begin the meeting by collecting distinct evidence rather than repeating endorsements. Compare both options against the agreed criteria. Record unresolved uncertainties and explain whether the final choice reflects evidence, compromise or a constraint that rules out an alternative.
Then transfer the procedure to a different decision, such as selecting a location rather than a project. If the team can still distinguish information coverage, initial preference and final commitment, it has learned a method. If it merely reproduces the questionnaire without understanding what the timing measures, it has copied a format. The test is whether the reasoning survives the changed problem.
CHAPTER 13 OF 14 · Use questions and primary sources
13. Questions readers ask about this history
Did social psychology begin with one famous experiment?
A single starting point is usually a teaching convenience. The selected 1898 paper provides an early empirical entry into this guide, not a claim that nobody previously investigated social life. Intellectual history includes different questions, institutions and traditions. Ask what a proposed origin identifies: a term, a department, a method or a particular experiment.
Do laboratory findings describe everyday life?
They describe the tested conditions first. Applying them elsewhere requires an argument about which features remain relevant and evidence that the relationship travels. A laboratory can isolate a useful contrast precisely because it leaves out parts of ordinary life. That strength also creates a boundary on immediate generalisation.
Does a failed replication make an earlier study worthless?
It requires the claim to be reconsidered in light of the new evidence. Possible issues include uncertainty, procedures, populations, measurement and the original interpretation. None should be assumed merely to protect a preferred conclusion. Compare the actual estimates and methods before deciding what has been learned.
Is agreement a warning sign?
Agreement is compatible with strong evidence, shared information, compromise and social pressure. The quality of the process matters. Independent initial judgements, explicit reasons and a record of new information help distinguish these possibilities. Disagreement is equally ambiguous: it can reveal overlooked evidence or simply different goals.
How can I study the history without memorising slogans?
For each paper, write the question, manipulation or comparison, outcome, main interpretation and one important limit. Then invent an alternative explanation and identify an observation that could distinguish it. Keep the historical source separate from your invented example so a memorable classroom scenario does not become a falsely remembered published finding.
What is a useful practical learning route?
Begin with one everyday decision where the stakes are modest. Record the information available before discussion, the alternatives considered and the final reasons. Review the record afterwards without treating a lucky outcome as proof of a good process. Compare with another decision and look for a specific repair, such as collecting missing evidence or assigning responsibility more clearly.
For the relationship between social settings and change across a life, continue to History of Developmental Psychology. The two histories intersect, but they organise their questions differently: one foregrounds social influence, while the other foregrounds continuity and change through time.
CHAPTER 14 OF 14 · Use questions and primary sources
14. Primary sources and how to keep their claims in scope
Triplett, N. (1898), The Dynamogenic Factors in Pacemaking and Competition, American Journal of Psychology, 9, 507–533. Read the procedures and different response patterns rather than replacing the paper with a universal claim about competition.
Asch, S. E. (1955), Opinions and Social Pressure, Scientific American, 193, 31–35. Track the distinction between public answers, individual differences and changes in the social arrangement. The worked denominators in this guide are invented examples, not the study’s reported rates.
Festinger, L., and Carlsmith, J. M. (1959), Cognitive Consequences of Forced Compliance, Journal of Abnormal and Social Psychology, 58, 203–210. Read the theory’s predictions alongside the actual procedure and outcome. Avoid translating a conditional laboratory result into an unconditional rule about rewards.
Darley, J. M., and Latané, B. (1968), Bystander Intervention in Emergencies: Diffusion of Responsibility, Journal of Personality and Social Psychology, 8, 377–383. Separate the staged comparison from the surrounding historical narrative, and distinguish individual responding from collective outcomes.
Tajfel, H., Billig, M. G., Bundy, R. P., and Flament, C. (1971), Social Categorization and Intergroup Behaviour, European Journal of Social Psychology, 1. The publisher’s abstract states the research aim and reported allocation patterns. The token choices in this article illustrate a distinction; they are not copied experimental materials.
Open Science Collaboration (2015), Estimating the Reproducibility of Psychological Science, Science, 349, aac4716. Read it as a defined collaborative investigation, and avoid treating its selected studies as a census of an entire discipline.
Across these sources, preserve the connection between claim and conditions. What was observed? What comparison makes it informative? Which mechanism is proposed? What else could explain the result? A useful history of social psychology leaves the reader better able to answer those questions when the next confident claim about human behaviour appears.