491/691 Psychology Spring, 2004 Professor Dunning
Major Assignment 1 Assigned: 1/26/04 Due: 2/19/04
TIME-SAMPLED BEHAVIORAL OBSERVATIONS Frequently psychologists are interested in observing and measuring the behavior of people in natural settings. One of the most common general techniques for the measurement of ongoing interactions involves time-sampling and coding of an individual's behavior. The purpose of this project is to acquaint you with these techniques and the problems involved in defining and quantifying social behavior in a systematic and reliable manner. In Phase 1, you will be asked to design a reliable behavioral/observational protocol. In Phase 2, you will be asked to use that protocol to test a specific hypothesis of interest. You will work in pairs (occasionally, in threes). You will have two fundamental choices to make at the outset. Of course, each is subject to change (frequently). First, you must choose a property of a person or interaction (e.g., aggressiveness, nervousness, friendliness) or of a social interaction (e.g., cooperation, intimacy) to measure. This is your conceptual variable. Second, you must choose a setting in which to measure the variable. You can look at anything from unstructured interactions (e.g., people talking at the Straight) to structured activities (e.g., intramural basketball, classroom discussions). I recommend more structured activities. First, people do not leave. Second, they do not become aware that they are being watched. The setting you choose must be a public one, one in which people understand that they are not in a private setting and, thus, can be observed. PRE-PHASE: QUALIFYING YOURSELF FOR RESEARCH Federal agencies require us to make sure that you have received an adequate training in ethical principles before you begin any type of research, including research as a class exercise. Thus, before you begin, you will have to take and pass the "qualifying" test that Cornell assigns to researchers. To do so, you will have to go to the following URL: http://www.osp.cornell.edu/HSCompliance/index.html Once there, go through all the training modules at the site. It should take about an hour to 90 minutes, then take the quiz associated with the modules. To pass, you must get 90%. At the quiz site, there is a place where we can be informed that you passed the quiz. Please tell the site to send notification to Caludia at [e-mail]. Note: You must go through this step. We cannot allow you to do any research exercise in this class if you have not passed this qualification step. PHASE 1: (the hard one) DESIGNING AN OBSERVATIONAL SYSTEM For the first phase of this project, observe people in some public setting, and get some ideas about how to measure your conceptual variable. Your measurement system must consist of a set of behaviors whose presence or absence reflects the conceptual variable. For example, if I wanted to measure "pleasantness," I would look for the presence or absence of: smiles, head nods, touching. HELPFUL HINT: Avoid looking for behaviors that are either rare (e.g., hitting) or overly common (e.g., breathing). The first phase will probably take several measurement sessions. First, you must receive approval from me or Wil about your choice of public setting and which personality dimension you wish to assess. Then, in your first session at this public setting, pick out at least three people to observe. Observe each for 10 minutes, if possible. For each person, spend three minutes looking just
watching, for behaviors that indicate the conceptual variable. Then spend 2 minutes writing down the behaviors you saw. Repeat this 3 minute and 2 minute procedure, and then compare notes with your partner. After you've settled on a conceptual variable, setting, and believe you know what behaviors you are going to watch, please email Joyce at [e-mail]. We must approve of your setting before you dig in with actual systematic observation. Dig in. Keep watching people until you and your partner (1) agree on the set of behaviors you will use for your measures, and (2) usually see the same behaviors from the same person at the same time. You and your partner should independently score your own observations - no discussion should occur about how ambiguous events are scored until after the initial coding has been done. Then, you both can compare scoring, and discuss how and why you disagreed. You should come to agreement about how to treat the kinds of behavioral items on which you disagreed. Now you are ready to create the more formal behavioral coding system. Create a "coding sheet" that lists the various behaviors you are searching for. What you and your partner will do is this: Watch a person for three minutes - or rather, watch a person for 12 consecutive 15 second intervals. For each 15 second interval, independently mark down what behaviors that person exhibited. Follow this procedure for at least four individuals. Then, calculate a figure of the percentage agreement between you and your partner on each measure. This measure, called interjudge reliability, is obtained by considering only the positive instances in which one or both of you said a behavior occurred. Reliability is calculated as the number of intervals in which you both scored the behavior as present divided by the total number of intervals in which either observer or both observers scored the behavior as present. The goal of the first phase is to develop an observation system that will yield better than 80% agreement, as calculated above. Keep plugging, collecting data, and refining your system until you achieve 80% agreement overall the specific behaviors you are coding. PHASE 2: (easy) HYPOTHESIS TESTING In the second phase of the assignment, you and your partner can work independently. In this second stage, your job is twofold: (1) to identify one or more variables that you believe "control" or influence the specific social behavior you have chosen to study, and (2) to use your observational system to test your hypotheses. Exactly how this hypothesis-testing phase should be designed will, of course, depend on the hypothesis or hypotheses you have generated. One set of hypotheses might concern "person variables" (e.g., comparing men to women, older to younger). Other hypotheses may involve situational factors (e.g., time of day, presence of police officer). WRITE-UP: A written report of your work on this project will be due February 19. This report should be brief. Seven pages, maximum. It should include two major items: (1) a discussion of the difficulties you initially encountered in designing an observational system and how they were overcome, leading to a description of the behaviors you used in your final coding system (along with data about reliability), and (2) a discussion of the hypothesis you tested using this system and the data you collected to examine it. You should append a copy of the coding sheets you used to calculate agreement.