Home Introduction Cognitive Psychology Cognitive Perspective Social Perception Social Memory Social Categorization Social Judgment Language Automaticity Self Social Neuropsychology Personality Social Intelligence Development Sociology of Knowledge Social Construction Conclusion Lecture Illustrations Exam Information


University of California, Berkeley

Department of Psychology


Psychology 164

Spring 2008


Midterm Examination


Final Exam Feedback

In what follows I provide the scoring guide given to the GSI.  These are simply intended to be samples of adequate (3-point) answers.  Other good answers were, of course, possible.  

Although I worried that the 15-item would prove to be a little too long, in fact the vast majority of students (72%) were able to answer all 15 questions.  In fact, a couple of students answered more than 15 items.  Following the procedure indicated on the exam, for those students we counted only the first five items in each section.

Slide3.JPG (51038 bytes)By these standards, class performance on the exam was very good, with a mean score of 42.72, and a standard deviation of 6.80.  That made the average grade a solid B (85%).  Ten (10) students got perfect scores of 50.

But wait!  I also conducted an item analysis to identify bad questions.  With a standard multiple-choice exam, in which everybody must answer every item, the statistical standards for a bad item are pretty clear: a low item pass percent, coupled with a low item-to-total correlation, is a pretty good indication that the item in question doesn't belong on the test.  But with a short-answer (or essay) exam, with continuous (rather than dichotomous) scoring of items, not to mention the element of choice, the standards aren't quite so clear, and so a little improvisation is called for. 

Here's how I did it.

The mean number of students answering any given question was 74.67, with a standard deviation of 28.64.

The mean score on the 21 items was 2.52 (SD = 0.35).

Slide2.JPG (76423 bytes)A plausible statistical criterion for identifying bad items would be to follow the statistical "Rule of 2" -- that any value that lies more than two SDs away from the mean is an outlier, and thus suspect.  That would mean that:



An item answered by fewer than 17% of the students (75-58) is an outlier.  There weren't any such items, but #s 12, 13, and 21 came close.

An item with a mean score less than 1.82 is an outlier.  There were two such items, #s 12 and 19.

An item that is a double outlier would be particularly suspect.  No item met that criterion, but  Item #12 came closest, so I dropped it.

How do you drop an item?  

The easiest thing to do is to give everyone credit for the item, regardless of their response.  That works when everyone has to answer all test items, but a test with choice creates a problem -- students who chose to answer the suspect item were effectively prevented from answering another item that was not suspect, on which they might have done better.  

So a more equitable solution is to count Item #12 as scored, but then to add 3 points to everyone's score.  That effectively removes the suspect item from the test, but also gives some measure of credit to those students who attempted it anyway.

The downside of this solution is that a few students, who achieved relatively high scores on the exam but did not choose to answer the suspect item, could, in principle, get as many as 53 points on a 50-point test.

So, the final step is to truncate scores above 50.  

So, to calculate your final score:

If you answered more than five questions in any section, count only the first five items.

Give yourself 5 points, assuming that you actually provided your name on every page of the exam.

Regardless of whether you answered #12, give yourself an additional 3 points.

If your total score is greater than 50, truncate it to 50 points, the maximum score allowable for this exam.

Slide5.JPG (50964 bytes)As a result of the rescoring procedure, the mean exam score rose to 45.97 (SD = 6.52).  The average grade is now an A- (92%).  A total of 40 students, more than 1/3 of the class, ended up with perfect scores.


In what follows I provide information on the item analysis, as well as some additional commentary, as appropriate.


Answer five (5) questions from each of Sections 1-3, for a total of 15 questions. Each of these questions is worth three (3) points. Do not answer more than 5 questions per section. If you answer more than five (5) questions in any section, we will count only the first five.

Be sure to print your name and UCB Student ID on every page of the exam. Following this instruction is worth five (5) points. Students who fail to print their names on every page of the exam will lose all five points.

Your responses should be very concise. As a rule, less than 5 sentences will do. Write your answers in the space provided. If absolutely necessary, you may continue on the other side of the page. Write legibly, in complete sentences, in the space provided, and please use ink. Exams written in pencil will not be eligible for regrading.



Section 1. Answer five (5) questions; do not answer more than five (3 points per question).

1. What are the premises of symbolic interactionism?

(1) Action depends on the meaning of the object toward which the action is directed. (2) The meanings of objects are derived from social interaction. (3) Meanings are not inherent in the objects themselves, but are constructed through cognitive processes.  74% of the class attempted the item; mean score = 2.34; item-to-total r = .40.

Comment:  The item-to-total correlation is the correlation of score on this item with score on the test as a whole.  As a rule, items with a low -- that is, not statistically significant -- item-to-total r don't belong on the test.  With a class this size, r  > .19 is statistically significant, p < .05).  


2. Describe three (3) ways in which persons affect their environments.

There are four modes by which this construction can occur: Any three will do. (1) Evocation, where the appearance and behavior of the person unintentionally evokes a response from the environment. (2) Selection, where a person chooses to place him/herself in one environment rather than another. (3) Manipulation, where a person changes the objective environment through overt behavioral activities. (4) Transformation, where the person acts cognitively to alter the subjective, mental representation of the environment.  92%; M = 2.78; r = .43.


3. What is the self-fulfilling prophecy? Distinguish between perceptual and behavioral confirmation effects.

In the self-fulfilling prophecy, a person acts in such a way as to make an originally false conception of a situation come true. In perceptual confirmation, the target's behavior is ambiguous, but is interpreted by the actor as confirming his/her expectancies. In behavioral confirmation, the target's behavior would be viewed by an objective, unbiased observer as confirming the actor's expectancies.  91%; M = 2.5; r = .39.


4. How does Lewin define the total psychological field?

The total psychological field in which behavior takes place consists, first, of the person in the situation; and, second, of cognition and motivation. Both elements in each pair are required for the successful prediction of behavior; neither element, alone, is sufficient for this purpose.  57%; M = 2.69; r = .53.


5. Briefly describe three (3) of the following models of social cognition: Consistency-seeker, Naive scientist, cognitive miser, motivated tactician, or activated actor.

The consistency-seeker is motivated to reduce any dissonance between attitudes, and between attitudes and behavior. The naive scientist engages in a rational analysis of events in the social world. The cognitive miser is motivated to reduce information-processing demand by relying on cognitive strategies that simplify complex problems. The motivated tactician chooses among available cognitive strategies based on goals. The activated actor relies on automatic, unconscious processes rather than conscious, deliberate ones.  44%; M = 2.68; r = .32.


6. Distinguish between priming and chronic accessibility.

Priming is an automatic process in which processing one stimulus facilitates (or inhibits) the processing of another stimulus, by momentarily activating prime-related concepts. For example, "subliminal" presentation of words like angry and fight can lead subjects to interpret a target's behavior as hostile and aggressive. Chronic accessibility is like priming, except that the "prime" is not a stimulus in the environment, but rather a chronically activated construct inside the perceiver's head. Thus, a person who is concerned about hostility may be set to interpret other people's behavior as hostile.  60%; M = 2.46; r = .66.


7. What is the perspective of cognitive sociology?

Cognitive sociology is concerned with individuals as members of thought communities, and with the role of cognitive socialization in shaping the individual's thought processes. Cognitive sociology begins with the assumption that different historical epochs, different cultures, and different subcultures are characterized by distinct differences in both the content and the mode of thought.  70%; M = 2.26; r = .52.



Section 2. Answer five (5) questions; do not answer more than five (3 points per question).

8. What is a central trait and what makes it central?

Asch defined a central trait has a psychosocial characteristic that, when changed, alters the entire impression of a person. Central traits carry more information about the person than peripheral traits do, because they are highly correlated with major dimensions of person perception, such as social and intellectual evaluation.  99%; M = 2.85; r = .46.


9. Why are individuals described with two highly negative traits rated as less likable than those described with two highly negative traits and two moderately negative traits?

According to the weighted averaging model of cognitive algebra, impressions are formed by averaging the values of individual traits, along an estimate of perceiver's initial bias. Assuming a default value of zero for bias, the relatively low values accorded the moderately positive traits are effectively lower the average value of the entire impression -- that is, they make the person with two highly negative traits and two moderately negative traits seem less dislikable!.  94%; M = 2.83; r = .23.


10. How do studies of emotion and babyfacedness bear on the ecological perspective on social perception?

We can "read" people's basic emotional states, such as fear and happiness, on their faces; and organisms with enlarged eyes, chubby cheeks, and large craniums tend to elicit caretaking and inhibit aggression. According to the ecological view, our perceptual systems evolved to automatically extract information about basic emotions and competence or vulnerability. All the information needed to form these impressions is in the stimulus -- no "higher" inferences or judgments are needed.  96%; M = 2.63; r = .61.


11. Distinguish between salience and vividness. What are their effects on social perception?

Both salience and vividness attract attention, so that salient and vivid stimuli play a relatively large role in person perception. Salience is a product of the relation between an object and its surrounding context. Vividness is inherent in the object itself.  85%; M = 2.57; r = .34.  


12. How do causal attributions influence attitude change in response to persuasive communications?

 People respond to persuasive communications, in part, based on their analyses of the causes of the communication. Communicators who might be biased by their own personal opinions (a personal attribute), or influenced by situational factors (a feature of the environment), are less credible than those whose communications appear to be independent of both dispositional and situational constraints -- and thus less persuasive, as well.  19%; M = 1.76; r = .77.  

Comment:  "Causal" was misspelled "caudal" in the original exam, which might have led most students to avoid it, and those who tried it to do relatively poorly.  Still, the item was highly correlated with total exam score.  Anyway, I think the point of the question was clear: there's a lot of discussion in Fiske & Taylor about causal attributions; I don't know what "caudal" attributions would be, except perhaps attributions to the caudate nucleus, a structure located in the basal ganglia of the brain, which plays a role in learning and memory.  But then again, the caudate lights up when people look at their romantic partners....  In the final analysis, it doesn't matter what I think: the statistical item analysis indicates that it is suspect -- although, frankly, the item-to-total correlation is so high that if I had to pick a single item to assess performance in the course, it would be this one.  Go figure..


13. Briefly describe the "elaboration likelihood" model of persuasion.

The elaboration likelihood model assumes that there are two routes to attitude change: a central route involving careful deliberation, or elaborative thinking, about the persuasive communication; and a peripheral route, in which attitude change occurs without much thought -- for example, by virtue of automatic priming from mere exposure. The answer should include either of the following: (a) Attitude change which occurs via the central route is likely to last longer than change that occurs via the peripheral route. (b) The likelihood of elaboration is a function of certain qualities of the communicator (e.g., credibility), message (e.g., argument quality), audience involvement, and individual differences in such factors as "need for cognition".  22%; M = 2.29; r = .37.


14. Define "optical pluralism" and "optical community". How are "social optics" related to attention.

Optical pluralism refers to the fact that there are different "mental lenses" through which we can "see" an event, resulting in a number of different possible perceptions, or interpretations, of an event. "Optical communities" are groups of people who have come, through a process of "optical socialization", to perceive events through the same set of "mental lenses".  69%; M = 2.81; r = .65.



Section 3. Answer five (5) questions; do not answer more than five (3 points per question).

15. What is a cognitive schema? What are the effects of mental schemata (otherwise known as schemas) on person memory?

A schema is a relatively abstract knowledge structure which guides perception and memory. As a result of schematic processing, person memory favors schema-relevant events over schema-irrelevant events, and schema-incongruent events over schema-congruent ones. At the same time, however, schemata can distort memory, leading people to falsely remember events that are congruent with a schema, but which did not in fact occur.  94%; M = 2.88; r = .37.


16. How is person memory organized in a generic associative-network model of memory?

In an associative-network model of memory, each fact about a person -- his or her identity, traits and other general characteristics, and specific behavioral episodes are represented as nodes connected by associative links that represent the relations among them. Evidence from both clustering and reaction-time (priming) studies is consistent with a model in which items of trait (semantic) and behavioral (episodic) information are represented independently of each other.  94%; M = 2.52; r = .58.


17. Compare localist and distributed theories of the neural representation of person memory.

Localist theories hold that items of knowledge are represented by the activity of single neurons, or perhaps small clusters of neurons, centered on a specific location in the brain. Distributed theories of memory hold that individual items of knowledge are represented by patterns of neural activity distributed widely across the cortex. Localist theories predict the existence of "grandmother cells", which invariably fire in the presence of a particular stimulus. Distributed theories are supported by evidence that the specific location of brain damage is not critical to determining the loss of learned responses.  95%; M = 2.97; r = .11.

Comment:  Here, the item-to-total correlation is relatively low -- actually not statistically significant --  but that's because there's very little variance in performance.  The correlation coefficient measures how much variance two variables have in common, and if one variable doesn't have any variance to begin with, the correlation between it and another variable must necessarily be low.


18. What advantage do procedural models of person memory have over declarative models? What disadvantage?

The processing of declarative memories tends to be slow and effortful, consuming limited cognitive resources in working memory. In procedural models, certain aspects of person perception and memory have been automatized by repeated practice, so that information is processed rapidly and efficiently. However, procedural memory is less flexible than declarative memory, more closely tied to the particular domain (verbal or nonverbal, visual or auditory) in which the information is acquired, or the specific operations used to process the information.  62%; M = 2.31; r = .24.


19. What does it mean to say that person memory is "embodied"?

The general idea behind embodiment is that the purpose of perception, memory, and other cognitive faculties is action. Bodily action does not just result from cognition -- it also contributes to it. Thus, people who nod their heads while processing a persuasive communication are more likely to agree with it. And when people imitate the facial expressions of emotion, they are likely to actually feel the corresponding emotion themselves. Thus, motor perceptions, and motor representations, are important elements in social cognition.  40%; M = 1.61; r = .70.

Comment:  This was a hard decision, because the average score almost made this item an outlier.  But it didn't quite cross threshold, and if you start fiddling with the criteria, you can appear arbitrary and capricious.  Besides, the item-to-total correlation was pretty high.


20. Distinguish between the prototype and exemplar views of social categorization.

Both the prototype and exemplar views of categorization are alternatives to the classical view of categories as collections of objects which share certain defining features in common. According to the prototype view, the summary representation is a list of "characteristic" features shared by many category members, but not necessarily all of them. The exemplar view eliminates the notion of "category as summary of features", and asserts instead that a category is simply represented by a list of the objects (exemplars) that are in the category.  71%; M = 2.57; r = .46.


21. What does it mean to say that social memory has a "normative dimension"?

From the perspective of cognitive sociology, memory is not just a matter of an individual remembering and forgetting past events. In sociological terms, each individual is part of a "remembrance community", consisting of people with shared memories. Such a community enforces "rules of remembrance" that indicate which events members should remember, and which they should forget.  22%; M = 2.63; r = .57.



A scoring guide will be posted to the course website by noon, March 5.

Exams will be graded and returned as soon as possible.

Requests for regrading must be made to your GSI

no later than 1 week after exams are returned.


This page last revised 05/12/10 05:52:53 AM.