You have reviewed the literature, developed a hypothesis, operationalized your variables, and given sound reasons for testing your hypothesis. However, your preliminary work is still not done. You must now decide exactly what specific actions you will take. In other words, although you probably have decided on the general design (such as a simple experiment or a 2 X 2), your plan is not complete until each detail has been thought through and written down.
You should write down exactly what procedures you will follow. For example, what instructions, word for word, will participants be given? Who will administer the treatment? Where? Will participants be run in groups or individually? In answering these questions, you must take into account issues of validity. However, your paramount concern must always be ethics. You do not have the right to harm another.
Ethics should be the foundation of your research plan. Therefore, you should read Appendix A (Ethics) before conducting a study.
In addition to reading Appendix A, you must be extremely careful not to harm your participants. Ideally, your participants should feel just as well when they leave the study as they did when they began the study. Unfortunately, even in the most innocuous studies, protecting your participants from discomfort is much easier said than done.
Realize that any experience may be traumatic to some participants. Trauma can occur from things you would never think of as being traumatic. Because any study has risks and because you won’t know all of the risks, do not run a single participant without your professor’s permission.
To begin to sensitize yourself to the risks involved in your proposed study, list the 10 worst things that could possibly happen to participants. If you are using human participants, be aware that not all participants will react in the same way. Some may experience trauma because the study triggers some painful memory. Some participants may feel bad because they think they did poorly. Other participants may feel bad because they think their behavior ruined your study. Realize that some of your participants may be mentally unbalanced and any attack on their self-esteem might lead to disastrous consequences. Because participants are often fragile, you should list some serious consequences in your worst-case scenario.
Because any study has the potential for harm, the possibility of severe consequences does not mean that your professor will not allow you to do the study. However, you and your professor should think about ways to minimize the risks.
One method of minimizing risks is to screen out “vulnerable participants.” For instance, if there is any reason to believe that your study may increase heart rate or blood pressure, you may want to make sure that only people in good health participate in your study. If your study might harm people with low self-esteem, you may want to use only well-adjusted participants who have high levels of self-esteem. Therefore, you might give a measure of self-esteem to potential participants to eliminate those with low self-esteem.
Not only should you screen participants, but you should also let participants screen themselves. That is, participants should be volunteers who give their informed consent: They should know what the study is about before volunteering for it.
How informed is informed consent? Very informed, when it comes to telling participants about any unpleasant aspect of the study. If participants are going to get shocked or exposed to loud noises or extreme cold, they should be informed of this before they volunteer. Consequently, if your study does involve unpleasantness, you may have difficulty getting participants to volunteer.
Informed consent is considerably less informed when it comes to more innocuous aspects of the study. After all, the study would be ruined if participants knew everything that would happen (and why it happened) before it happened. So, although participants are usually told the truth, they are not always told the whole truth. For example, a memory experiment’s description would mention that participants have to memorize words, but might omit the fact that the researcher is looking at the order in which facts are recalled or that there is a surprise recall of all the lists at the end of the study.
Because participants are not fully informed about your study, there may be some things about it that they dislike. For example, suppose a participant finds the task too difﬁcult or finds it upsetting to try the surprise recall task. What can you do?
One protection against these unexpected problems is to make sure participants understand that they can quit the study at any time. So, before the participants begin your study, tell them that if they find any aspect of the study uncomfortable, they can and should escape this discomfort by quitting the study. Assure them that it is their duty to quit if they experience discomfort and that they will still get full credit.
You have seen that you can minimize ethical problems by letting participants know what they are in for and by letting participants gracefully withdraw from the study. You should also minimize harm by making your study as humane as possible. You can make your study more ethical by reducing the strength of your treatment manipulation, carefully selecting stimulus materials, and by being a conscientious researcher.
Although using extreme levels of your predictor variable may help you get a significant change in the criterion variable, extreme levels may harm your participants. For example, 24 hours of food deprivation is more likely to cause hunger than 12 hours. However, 24 hours of deprivation is more stressful to the participant. If you plan an unpleasant manipulation, remember your participants’ welfare and minimize the unpleasant consequences as much as possible. Consider using levels of the predictor variable that are less severe than you originally intended.
By modifying your stimulus materials, you may be able to prevent them from triggering unpleasant memories. For instance, if you were interested in the effects of caffeine on memory for prose, you wouldn’t want the prose passage to cover some topic like death, divorce, alcoholic parents, or rape. Instead, you would want to use a passage covering a less traumatic topic such as sports. If the sports article referred to someone’s death or hospitalization, you might want to delete that section of the article.
Often, it’s not the study that causes ethical problems, it is the researcher’s arrogance. For example, an arrogant researcher may rush through research sessions providing only superficial explanations and almost no time for questions and feedback. Although we know of a few participants who were hurt as a direct result of a research manipulation, we know of many more who were hurt because the researcher treated them like dirt. To ensure that you are sensitive, courteous, and respectful to all of your human participants, you should do two things.
First, when scheduling your research sessions, make sure you leave a 10-minute gap between the end of one session and the beginning of the next session. Some investigators feel that, like a physician, they should efficiently schedule people one after another. Their “efficiency” results in participants having to wait for the investigator, the investigator having to rush through the formalities of greeting participants, or—even worse—the investigator rushing through debriefing. Thus, the “efficient” investigator, like the efficient physician, is seen as unconcerned. Although this conduct does not become physicians, it’s intolerable for psychological researchers! After a research participant has given an hour of his or her time, you should be more than willing to answer any questions the participant has. Furthermore, if you rush through greeting or debriefing each participant, the participants will see you as uncaring. Consequently, they will be less likely to tell you about any psychological discomfort they felt and less likely to accept any aid you might offer. Thus, the first step is to walk, rather than to run, participants through your study.
Second, give the participants power. That is, allow participants to rate your study on a scale such as the one in Table C–1. Give each participant’s rating sheet to your instructor. Following this simple procedure helps you to be a conscientious and courteous researcher.
Although you should try to anticipate and prevent every possible bad reaction a participant may have to being in your study, you won’t be successful. Inevitably, your procedures will still cause some unpleasantness. After the study is over, you should try to remove this unpleasantness by informing participants about the study, reassuring them that their reactions were normal, and expressing your appreciation for their participation.
You should also listen to participants and be sensitive to any unexpected, unpleasant reactions to your study. By being a good listener, you should be able to undo any damage you have unwittingly done. This process of informing your participants about the study and removing any harm done is called debriefing. Occasionally, ordinary debriefing will not undo the harm caused to the research participant. In those cases, there are several steps you may take to alleviate distress. For participants who are upset with their responses, you should ask them whether they want you to destroy their data. For participants that you cannot calm down, you should take them to talk to a professor, counselor, or friend—even if this means canceling a research session you had scheduled.
In summary, you should be very concerned about ethics. Since ethics involves weighing the costs of the study against the potential benefits, you should do everything you can to minimize the risk of participants becoming uncomfortable. If, despite your efforts, a participant experiences discomfort, you should try to reduce that discomfort during debriefing.
With animal participants, you incur the same responsibilities that you did with human participants—you must protect animal participants from undue stress and discomfort. In many ways, you have even more responsibility to animal participants because they depend on you for their mere existence. You must keep them fed, clean, warm, and comfortable—24 hours a day. To fulfill your responsibility to animal participants, you must follow APA’s guidelines for proper housing, food and water, and handling (see Appendix A).
Furthermore, because your animal participants do not have the power to give their informed consent, nor the power to quit the study, you must carefully question the value of your study. Ask yourself and your professor this question: “Is the potential knowledge gained from the study worth the cost to the animals?” Finally, if you must euthanize (kill) your animal participants at the end of your study, follow APA’s guidelines to ensure that this is done in the most humane way.
We have discussed ways of minimizing harm to participants. However, minimizing harm is not enough to ensure that your study is ethical. For your study to be ethical, the potential benefits must be greater than the potential harm. Thus, an extremely harmless study can be unethical if the study has no potential benefits. So, just as you owe it to your participants to reduce potential harm, you owe it to your participants to maximize the potential benefits of your study. You maximize that potential by making sure your study provides accurate information. To provide accurate information, your study needs to have power and validity.
One of the most serious obstacles to obtaining accurate information is lack of power. Remember, null results don’t prove the null hypothesis. They only make people wonder about the study’s power. There is no point in doing a study that is so powerless that it will lead to inconclusive, null results.
To have power, you should use a strong manipulation, a sensitive dependent measure, well-standardized procedures, a sensitive design, and enough participants.
Perhaps your most important obstacle to finding a significant effect is a lack of participants. As a general rule, you should have at least 16 participants in each group. However, the number of participants you need in each group will be affected by the sensitivity of your design, the heterogeneity of your participants, the number of observations you get from each participant, the size of the difference you expect to find between conditions, and the sensitivity of your dependent measure.
If you have a within-subjects design, a reliable and sensitive dependent variable, and expect a rather large difference between your conditions, you may be able to use fewer than 16 participants per group. If, on the other hand, you are using a simple, between-subjects design, heterogeneous participants, a manipulation that may have little effect, and a relatively insensitive dependent measure, you may want at least 100 participants per condition.
Hunting for Participants
How are you going to get all the volunteer participants you need to conduct a powerful study? The threat of death is not ethical.
Some researchers rely on “captive” samples. For example, many colleges “volunteer” students in introductory psychology courses for the research draft. In fact, most of the research strength of modern psychology has been built using this research draft. If your school has such a draft, count yourself among the blessed. All you have to do is ask your professor how to become a recruiter.
If your school does not have a draft, an effective way of getting participants is to ask professors to request volunteers from their classes. Many professors will gladly do this. Some will even give volunteers extra credit as an incentive for participating in your study.
But what if you don’t want to use college students in your study? For example, suppose you want to study children or retirees? Or, suppose you agree with the skeptics who claim that results from studies done on college students cannot be applied to “normal people.” Then, you would look beyond college classrooms for participants. A note of caution: You may find that getting real-world participants takes as much work and creativity as planning your study.
If you want to study children, you may be able to take advantage of the “captive” audience approach. After all, most children have to go to school. However, obtaining access to those children may turn into a nightmare of red tape. You will have to obtain permission from all or many of the following: the school board, the superintendent, principal, teacher, parent, child, your professor, and university. If you are going to get these permissions in time for your study, you’ll need to plan ahead—and be very lucky.
Finding adult participants can be even more challenging than finding children. For example, one of your textbook’s authors wanted an adult population for her doctoral dissertation. Her first thought was to contact a major company and gain access to its employees. This tactic failed. Next, she tried to run a newspaper ad asking for volunteers. One newspaper refused to print it. Another would only run it in the “Personal” section. Thus, her appeal for participants appeared with ads for astrological advice, massage services, and people wanting dates. Although a few “volunteers” called, most wanted either a date or an obscene conversation. We do not recommend newspaper ads—especially if your goal is to get a representative sample of the adult population.
The authors have had greater success recruiting elderly participants. Nutrition centers, retirement communities, friendship networks, and nursing homes have been fruitful sources of participants. In addition, we recommend the “grandmother connection”: having an older relative or friend introduce you to other prospective participants.
Obviously, finding human participants will take planning, perseverance, and luck. Once you contact prospective participants, you should explain your study to them and have them sign a permission form. The permission form will protect both you and your participants. Basically, it states that you have explained the study to the volunteers and that they agree to participate (see Table C–2). If minors are participating in your study, you need to have separate forms for both the participants and their parents.
Our experiences with recruiting human participants might have increased your enthusiasm for animal research. In many ways, animals are easier to study than humans. You do not have to worry about permission slips, extra credit, or obscene phone calls. Consult with your instructor about obtaining animals for your research. Often, schools have rat colonies or purchase animals for student research.
After ensuring that your study has adequate power, we would like to be able to tell you that you can take it easy and relax. Unfortunately, however, you can’t relax. Power is not your only concern when conducting psychological research. You must also ensure that the construct validity of your results is not destroyed by
If you use more than one investigator, you may be able to detect researcher effects by including the researcher as a factor in your design. In other words, randomly assign participants to both a condition and to a researcher. For example, if you have two treatment conditions (A and B) and two researchers (1 and 2), you would have four conditions: (1) A1, (2) B1 (3) A2 and (4) B2. After having Research 1 run conditions 1 and 2 and Researcher 2 run conditions 3 and 4, you could do an analysis of variance (ANOVA) using researcher as a factor to see whether different researchers got different results.
Using ANOVA to detect researcher effects can be useful. However, there are at least two reasons why using ANOVA may not eliminate researcher effects. First, this statistical approach will only tell you if one researcher is getting different results than other researchers. If all your researchers are biased, you may not get a significant researcher effect. (Besides, if you are the only researcher, you can’t use researcher as a factor in an ANOVA.) Second, and more importantly, detecting researcher effects is not the same as preventing researcher effects.
To prevent researcher effects, you must address the three major causes of researchers failing to conduct studies in an objective and standardized manner. What are these causes? First, researchers may not know how to behave because the procedures for how the researchers should conduct the study have not been spelled out. Second, researchers may not follow those procedures. Third, the researchers may strongly expect participants to behave in certain ways.
Often, the researchers aren’t behaving in an objective and standardized way because of the loose-protocol effect: The instructions aren’t detailed enough. Fortunately, the loose-protocol effect can avoided.
Before you start your study, carefully plan everything out. As a first step, you should write out a set of instructions that chronicles the exact procedure for each participant. These procedures should be so specific that by reading and following your instructions, another person could run your participants the same way you do.
To make your instructions specific, you might want to write a computer program based on these instructions. Since computers don’t assume anything, writing such a program forces you to spell out everything down to the last detail. If you can’t program, just write the script as if a robot were to administer the study. Write out each step, including the actual words that researchers will say to the participants. The use of such a script will help standardize your procedures, thus reducing threats to validity.
Once you have a detailed draft of your protocol, give it a test run. For example, to ensure that you are as specific as you think you are, pretend to be a participant and have several different people run you through the study using only your instructions. See how the different individuals behave. This may give you clues as to how to tighten up your procedures. In addition, you should run several practice participants. Notice whether you change procedures in some subtle way across participants. If so, adjust your instructions to get rid of this variability.
Unfortunately, even if you write out your protocol (procedures) in detail, you or your co-investigators may still fail to follow that protocol. To avoid the researcher failure-to-follow-protocol effect, you need to make sure that: (1) All investigators know the procedures, and (2) that everyone is motivated to follow the procedures.
To make sure that investigators learn the procedures, you should hold training sessions. Supervise investigators while they practice the procedures on each other and on practice participants.
Once researchers know the right way to run the study, the key is to make sure that they are motivated to run the study the same way every time. To increase researchers’ motivation to be consistent, you might have them work in pairs. While one researcher runs the participants, the other will listen in through an intercom or watch through a one-way mirror. You may even wish to record research sessions.
If your researchers still have trouble following procedures, you may need to automate your study. For instance, you might use a computer to present instructions, administer the treatment, or collect the dependent measure. Computers have the reputation for following instructions to the letter, so using a computer may help standardize your procedures. Of course, computers aren’t the only machines that can help you. Some of the machines that could help you give instructions and present stimuli include automated slide projectors, tape recorders, and videotape players. Countless other devices can be used to record your data accurately, from electronic timers and counters to noise-level meters.
The final source of researcher bias is the researcher-expectancy effect: Researchers’ expectations are affecting the results. You can take three steps to prevent the researcher-expectancy effect:
1. Be very specific about how investigators are to conduct themselves. Remember, researcher expectancies probably affect the results by changing the investigator’s behavior rather than by causing the investigator to send a telepathic message to the participants.
2. Don’t let the investigators know the hypothesis.
3. Don’t let investigators know what condition the participant is in—making the investigator “blind.” Although making investigators blind is easiest in drug experiments where participants take either a placebo or the real drug, you can make investigators blind in nondrug experiments.
For example, if you present stimuli in booklets, you can design your booklets so that booklets for different conditions look very similar. In that way, an investigator running a group of participants might not know what condition each participant is in. For some studies, you may be able to use a second investigator who does nothing except collect the dependent measure. This second investigator could easily be kept in the dark as to what condition the participant was in.
Whether you are the only investigator or one of a team of investigators, researcher effects may bias your results. Therefore, you should always try to prevent the loose-protocol effect, the failure-to-follow-protocol effect, and the researcher-expectancy effect.
Unfortunately, in psychological research, you must be aware not only of researcher effects, but also of participant effects: Participants may see through the study and try to play along with what the investigator wants. Fortunately, there are various ways of preventing participants’ expectancies from biasing your results.
For starters, you might make your researcher blind to reduce the chance that the participant will get any ideas from the researcher. Thus, the techniques for reducing research effects that we just discussed may also reduce the effects of participants’ expectancies.
In addition, you may also be able to prevent participants’ expectancies by skillfully choosing your research design. In experimental investigations, for example, you might use a between-subject design rather than a within-subject design because participants who are exposed to only one treatment condition are less likely to guess the hypothesis than participants who are exposed to all treatment conditions.
Another design trick you can use to reduce the impact of participants’ expectancies is to use placebo treatments. Placebo treatments prevent participants from knowing that they are in the “no-treatment” condition. Therefore, if you have comparison condition(s), use placebo treatment(s) rather than no-treatment condition(s). That way, all groups think they are receiving the treatment. Thus, any treatment effect you find will not be due to participants changing their behavior because they expect the treatment to have an effect.
Participants are less likely to know the hypothesis if they don’t know what you are measuring. Obviously, if they don’t even know they are being observed—as in some field studies—they won’t know what you are measuring. Thus, if your hypothesis is an obvious one, you might try to do a field study.
Although field studies lend themselves to unobtrusive recording, unobtrusive recording can even occur in a laboratory study. That is, participants will assume that if you are not in the room with them, you are not observing them. However, thanks to one-way mirrors and intercoms, you can monitor participants’ behavior from the next room.
Even if the participant knows you are watching, the participant doesn’t have to know what you are watching. That is, you can use unobtrusive measures. For example, you might put the participant in front of a computer and ask the participant to type an essay. Although the participant thinks you are measuring the essay’s quality, you could have the computer programmed to monitor speed of typing, time between paragraphs, number of errors made, and times a section was rewritten. In addition, you might also have tape-recorded and videotaped the participant, monitoring his or her facial expressions, number of vocalizations, and loudness of vocalizations.
Rather than trying to obscure or confuse participants as to the purpose of the study, you might try to prevent participants from thinking about the purpose of the study. How? By designing a study that has a high degree of research realism: a study that involves participants in the task. Experimental realism means that participants aren’t constantly saying to themselves: “What does the researcher really want me to do?” or “If I were a typical person, how would I behave in this situation?” Note that experimental realism doesn’t mean the study is like real life; it means that participants are engrossed in the task. In this age of video games, even a fairly artificial task can be very high in experimental realism.
Before now, you might have been surprised to see experimental realism and other strategies for reducing participant effects in a section on ethics. However, you now know that planning an ethical study involves taking into account many factors. Not only must you ensure the safety of your participants, but you must also demonstrate the validity of your methods. To avoid overlooking an important ethical consideration, consult Table C–4, Appendix A, and your professor.
Beyond the Proposal: The Pilot Study
Even after you have carefully designed your study, modified it based on comments from your instructor, and been given your professor’s go-ahead to run it, you may still want to run several participants (friends, family members, other members of the class) just for practice. By running practice participants, you’ll get some of the “bugs” out of your study. Specifically, by running and debriefing practice participants, you will discover
1. whether participants perceived your manipulation the way you intended.
2. whether you can perform the study the same way every time or whether you need to spell out your procedures in more detail.
3. whether you are providing the right amount of time for each of the research tasks and whether you are allowing enough time in between tasks.
4. whether your instructions were clear.
5. whether your cover story was believable.
6. whether you need to revise your stimulus materials.
7. how participants like the study.
8. how long it takes you to run and debrief a participant.
In short, running practice participants helps you to fine-tune your study. Because running practice participants is so useful, many professional investigators run enough practice participants to constitute a small study—what researchers call a pilot study.
Conducting the Actual Study
The dress rehearsal is over. Final changes in your proposal have been made. Now you are ready for the real thing—you are ready to “run” your study! This section will show you how.
As you may imagine, some of your prospective participants may be apprehensive about the study. Participants often aren’t sure whether they are in the right place, or even whether the researcher is a Dr. Frankenstein.
To put your participants at ease, let them know they are in the right place, and be courteous. You should be both friendly and businesslike. The expert investigator greets the participant warmly, pays close attention to the participant, and seems concerned that the participant knows what will happen in the study. The expert investigator is obviously concerned that each participant is treated humanely and that the study is done professionally.
Being professional doesn’t hurt how participants view you. Why? First, most participants like knowing that they are involved in something important. Second, some will view your professionalism as a way of showing that you value their time—which you should.
So, how can you exude a professional manner? Some novice investigators think that they appear professional when they act aloof and unconcerned. Nothing could be less professional. Participants are very turned off by a disinterested attitude. They feel that you don’t care about the study and that you don’t care about them.
To appear professional, you should be neatly dressed, enthusiastic, well-organized, and prompt. “Prompt” may be an understatement. You should be ready and waiting for your participants at least 10 minutes before the study is scheduled to begin. Once your participants arrive, concentrate exclusively on the job at hand. Never ask a participant to wait a few minutes while you socialize with friends.
What do you lose by being a “professional” investigator? Problem participants. If you seem enthusiastic and professional, your participants will also become involved in doing your study—even if the tasks are relatively boring. Thus, if you are professional in your manner and attitude, you will probably not even have to ask the participants to refrain from chatting throughout the study. Similarly, if you are professional, participants will stop asking questions about the study if you say, “I’ll explain the purpose at the end of the study.”
After you have established rapport, you need to give your participants instructions. To get participants to follow instructions to the letter, you might take the following six steps.
1. Be repetitive.
2. Have participants read the instructions.
3. Orally paraphrase those instructions.
4. Run participants individually.
5. Invite participants to ask questions.
6. Have participants demonstrate that they understand the instructions by quizzing them or by giving them a practice trial before beginning the study.
Once the study has begun, try to follow the procedure to the letter. Consistently following the same procedures improves power and reduces the possibility of bias. Therefore, don’t let participants change your behavior by reinforcing or punishing you. For instance, imagine you are investigating long-term memory. You want to expose participants to information and then see what they can write down. However, if you do this, participants may be writing down information that is in short-term memory. Thus, you would not be assessing long-term memory. Therefore, you add a counting backwards task that should virtually eliminate all of the information from short-term memory. Specifically, in your memory study, participants are exposed to information, are supposed to count backwards from a number like 781 by 3s for 20 seconds, and then are asked to recall the information. Ideally, their recall will represent only what they have in long-term memory. Unfortunately, many participants will and the counting task unpleasant, embarrassing, or simply an unwanted nuisance. Consequently, some participants will thank you for telling them they can stop; others will plead nonverbally for you to stop. Clearly, you cannot let any of these strategies stop you from making them count backwards for the full 20 seconds. If you vary your procedures from participant to participant based on each participant’s individual whims, your study will have questionable validity.
Once the study is over, you should debrief your participants. In debriefing, you should first try to find out whether the participants suspected the hypothesis. Simply ask participants what they thought the study was about. Then, explain the purpose of your study.
If you deceived your participants, you need to make sure they are not upset about the deception. You also need to make sure that they understand why deception was necessary. Participants should leave the study appreciating the fact that there was one and only one reason you employed deception: It was the only way to get good information about an important issue.
Making sure participants accept your rationale for deception is crucial for three reasons. First, you don’t want your participants to feel humiliated or angry. Second, if they get mad, they may not only be mad at you, but also at psychologists in general. Perhaps that anger or humiliation will stop them from visiting a psychologist when they need help. Third, the unhappy participant may spread the word about your deception, ruining your chances of deceiving other participants.
After explaining the purpose of the study, you should answer any questions the participants have. Although answering questions may sometimes seem like a waste of time, you owe it to your participants. They gave you their time, now it’s your turn.
After participants’ questions and doubts have been dealt with, give them an opportunity to rate how valuable they felt the study was. These ratings (1) encourage you to be courteous to your participants; (2) let you know whether your study is more traumatic than you originally thought; and (3) make participants feel that you respect them because you value their opinions.
After the rating, you should assure participants that their responses during the study will be kept confidential. Tell them that no one but you will know their responses. Then, ask the participants not to talk about the study because it is still in progress. For example, you might ask them not to talk about the study until next week. Finally, you should thank your participants, escort them back to the waiting area, and say goodbye.
Protecting Data: Confidentiality
You might think that once a participant leaves the study, your responsibilities to that participant end. Wrong! You are still responsible for guaranteeing the participant’s privacy. Knowledge about a given participant is between you (the investigator) and the participant—no one else. Never violate this confidentiality. To ensure confidentiality, you should take the following five precautions:
1. Assign each participant a number. When you refer to a given participant, always use the assigned number—never that participant’s name.
2. Never store a participant’s name and data in a computer—this could be a computer hacker’s delight.
3. If you have participants write their names on booklets, tear off and destroy the cover of the booklet after you have analyzed the data.
4. Store a list of participants and their numbers in one place and the data with the participants’ numbers on it in another place.
5. Watch your mouth. There is rarely a reason to talk casually about a participant’s behavior. Even if you don’tmention any names, other people may guess or think they have guessed the identity of your participant. We realize that it is hard to keep a secret. But to talk freely about someone who participated in your study is to betray a trust. Furthermore, keeping secrets will, for many of you, be an important part of your professional role: Therapists, researchers, consultants, lawyers, and physicians all must keep their clients’ behaviors confidential.
 The more participants, the more power. Indeed, some (Cohen, 1990) would consider 64 participants per group to be a reasonable minimum. We have talked about minimums. Are there maximums? Could you have a design that was too powerful? Some would argue that, in some cases, researchers use so many participants that even the smallest of effects, no matter how practically and theoretically insignificant, would be statistically significant. However, having a design that is too powerful is rarely a problem for novice researchers.
 You may want to consult with your professor as to the type of ANOVA you should use. There is some debate as to whether conventional ANOVA should be performed or a “random effects” model should be used.
A guide from APA about conducting research for high school science projects (has some nice forms and examples for dealing with ethical issues--as well as some practical advice about planning and conducting research)
Back to Research Design Explained Student Main Menu