Observational research is type of correlational (i.e., nonexperimental) research in which a researcher observes ongoing behavior. There are a variety of types of observational research, each of which has both strengths and weaknesses. These types are organized below by the extent to which an experimenter intrudes upon or controls the environment.
Naturalistic observation, also known as nonparticipant observation, has no intervention by a researcher. It is simply studying behaviors that occur naturally in natural contexts, unlike the artificial environment of a controlled laboratory setting. Importantly, in naturalistic observation, there is no attempt to manipulate variables.
Strength: We can measure what behavior is really like. After all, the researcher is observing real-life. This type of research, then, has high ecological validity (the extent to which a situation generalizes to real-life circumstances).
Weakness: We don't know the cause of behaviors, nor do we know if any observation is representative of what normally occurs.
Criteria for Naturalistic Observation
There are three specific criteria for an observational research study to be considered 'naturalistic.' If any one of these three are violated, the research is no longer naturalistic observation.
1) The setting must be natural. A researcher cannot adjust, control, change, or influence the setting or environment.
2) The event must be natural. If you're interested in memory for arguments and you wanted to use naturalistic observation, you'd basically have to wait until an argument to occur to collect data -- bad idea. Staging a fake argument, however real it may seem, is not a natural event and thereby violates this criterion.
3) The behavior must be natural. This requires that a researcher be unnoticed. For example, if you're measuring walking speed, you have to make sure you are sneaky about it; if anyone notices you with a stopwatch and a notepad, their behavior will likely change as a results, thereby violating this criterion.
Violations of the Criteria
Why would it matter that one of the above criteria is violated? Because of reactivity; people you're observing will act differently if they know the situation isn't natural, that the event isn't natural, or that they're being measured. Is this a problem in laboratory research? Don't participants know they're being observed and measured? Wouldn't this affect their behavior? Absolutely, positively, yes! But for whatever reason, reactivity is (for the most part) ignored in laboratory research.
Here, unlike naturalistic observation, the researcher intervenes in the environment. Basically, this refers to inserting yourself as a member of a group in order to observe behavior you wouldn't otherwise have access to. Although it seems like naturalistic observation and participant observation are simply categories, you should understand that there is a really a continuum of intrusion into the environment. It depends on the extent to which the researcher is involved in the research study. For example, if someone sets up an event (e.g., putting trash in a path to see who picks it up), this is not naturalistic observation (see criteria above). This is also not participant observation research because the researcher is not a part of the group being observed. Think of this as a continuum with naturalistic observation on one end and participant observation on the other.
Here are two famous examples of participant observation:
Example 1: David Rosenhan (On Being Sane in Insane Places, 1973)
Rosenhan was interested in how diagnoses of mental illness were made. He and seven associates went to different mental institutions and simply said they were hearing voices. They were all admitted to the hospitals, despite the fact that they all acted normally. The range of stays in the hospitals was from a low of 9 days to a high of 52 days (yikes!). It seems the people who knew the researchers weren't real patients were the real patients themselves! When the eight were discharged, it wasn't on the basis of misdiagnosis but "schizophrenia in remission." Rosenhan would have never been able to have the insight into how labels, diagnoses, and treatments were given without acting as a participant in the observation.
Example 2: Leon Festinger (1956) - Doomsday Cult
Festinger was interested in cult members' attitudes and beliefs. Of course, you cannot use naturalistic observation in this study, so the reasonable alternative was participant observation. How did Festinger do this? He joined the cult (and obviously, didn't tell them he was a psychologist). This particular cult thought the world was going to end on a particular day. Festinger was interested in how the beliefs and attitudes of the cult's members would react when (or...if?) the world didn't end. When that date came and went, most people would probably think belief in the cult would wane. Amazingly, after the world didn't end, the strength of the cult members' beliefs actually increased. Why? Because they thought the world didn't end because of their prayers. Festinger would never have been able to research this without becoming both a participant and an observer in his research.
Advantages of Participant Observation Research
The advantages probably seem obvious to you. Certainly, participant observation research allows one to gain information one wouldn't have otherwise had access to. Secondly, behaviors remain relatively natural, thereby giving the measurements high external validity.
Disadvantages of Participant Observation Research
There are a variety of disadvantages. First, the people being observed have no opportunity to provide informed consent to be a participant in the research. This is a serious ethical consideration that should not be taken lightly. Second, the researcher loses objectivity. How can a researcher be a participant in the observation and remain completely neutral? It's impossible. You'll form opinions and change your behavior accordingly. In the famous Stanford Prison Experiment, Dr. Zimbardo acted as the prison warden, despite the fact that he was the principal investigator. This was unfortunate because he lost his sense of objectivity about the study. He, like the "prisoners" and "guards" truly acted the role and could not see that he should have stopped the study earlier. The third potential disadvantage of participant observation research is that your participation in the study in likely to influence what you're measuring. That is, you'll probably influence the data because you're interacting with your observations. This can sometimes be remedied by using observers who are blind to the purpose of the study.
Other Types of Observational Research
This is a type of observational research that involves a thorough descriptive analysis of a single individual, group, or event. We may learn more about this later. For now, you should understand that this is a type of observational research.
Archival research involves an analysis of already-existing data. An hypothesis is generated and then tested by analyzing data that have already been collected. This is a useful approach when one has access to large amounts of information collected over long periods of time. Such databases are available, for example, in longitudinal research that collects information from the same individuals over many years.