Observational Methods of Research and Observer Biases (VII)

Observers Biases and Their Effect on Results

Observational Methods of Research and Observer Biases and how theses biases affects results.

    Although observation is an important method of data collection, both unstructured and structured observations are vulnerable to biases. Human perceptual errors and inadequacies are a continuous threat to the quality of information obtained. Observation and interpretation are demanding tasks, requiring attention, sensation, perception, and conception. 

    To accomplish these activities in a completely objective fashion is challenging and perhaps impossible. Several types of observational bias are especially common. One bias is the enhancement of contrast effect, in which observers distort observations in the direction of dividing content into clearcut entities. The converse effect a bias toward central tendency occurs when extreme events are distorted toward a middle ground. 

    A series of biases are called assimilatory, in which observers distort observations in the direction of identity with previous inputs. This bias would have the effect of mis-categorizing information in the direction of regularity and orderliness. Assimilation to the observer's expectations and attitudes also occurs. Rating scales and other evaluative observations are also susceptible to bias. 

    The halo effect is the tendency of observers to be influenced by one characteristic in judging other, unrelated characteristics. For example, if we formed a positive general impression of a person, we would probably be likely to rate that person as intelligent, loyal, and dependable simply because these traits are positively valued. Rating scales may reflect observers' personality. 

    The error of leniency is the tendency for observers to rate everything positively, and the error of severity is the contrasting tendency to rate too harshly. Biases are especially likely to operate when a high degree of observer inference is required. Although the degree of observer bias is not a function of the degree of structure imposed on observations, it is usually more difficult to assess the extent of bias when using unstructured methods. 

    The careful construction and pretesting of checklists and rating scales (with structured observation) and the proper training and preparation of observers are techniques that can play an important role in minimizing or estimating biases. If people are to become good instruments for collecting observational data, then they must be trained to observe in such a way that accuracy is maximized and biases are minimized. 

    Even when the principal investigator is the primary observer, self-training and dry runs are essential. The setting during the trial period should resemble as closely as possible the settings that will be the focus of actual observations. Ideally, training should include practice sessions during which the comparability of the observers' recordings is assessed. 

    That is, two or more observers should watch a trial event or situation, and observational notes or coding should then be compared. Procedures for establishing the interrater reliability of structured observational instruments are described in Chapter 1

Main Reading Points

    1:Observational methods are techniques for acquiring research data through the direct observation and recording of phenomena.

    2:Researchers focus on different units of observation. The molar approach details observations of large segments of behaviors as integral units; the molecular approach treats small, specific actions as separate entities.

    3:Concealed observation, with people unaware that they are being observed or participating in a study, is done to reduce reactivity (ie, behavioral distortions due to the presence of an observer)

    4:Observational intervention refers to the degree to which observers structure the observational setting in line with research demands, as opposed to being passive observers.

    5:Qualitative researchers collect unstructured observational data, often through participant observation. Participant observers obtain information about the dynamics of social groups or cultures within members' own frame of reference.

    6:In the initial phase of participant observation studies, researchers are primarily observers getting a preliminary understanding of the site. As time passes, researchers become more active participants.

    7:Observations tend to become more focused over time, ranging from descriptive observation (broad observations) to focused observation of more carefully selected events or interactions, and then to selective observations designed to facilitate comparisons.

    8:Participant observers usually select events to be observed through a combination of single positioning (observing from a fixed location), multiple positioning (moving around the site to observe in different locations), and mobile positioning (following a person around a site).

    9:Logs of daily events and field notes are the major methods of recording unstructured observational data. Field notes are both descriptive and reflective.

    10:Descriptive notes (sometimes called observational notes) are detailed, objective accounts of what transpired in an observational session. Observers strive for detailed, thick descriptions.

    11:Reflective notes include methodological notes that document observers' thoughts about their strategies; theoretical notes (or analytic notes) that represent ongoing efforts to make sense of the data; and personal notes that document observers' feelings and experiences.

    12:Structured observational methods impose constraints on observers to enhance the accuracy and objectivity of the observations and to obtain an adequate representation of the phenomena of interest.

    13:Checklists are tools for recording the occurrence or frequency of predesignated behaviors, events, or characteristics. Checklists are based on category systems for encoding observed phenomena into discrete, mutually exclusive categories.

    14:Some checklists categorize exhaustively all behaviors of a particular type (eg, body movements) in an ongoing fashion, whereas others use a sign system to record particular behaviors while ignoring others.

    15:With rating scales, another record-keeping device for structured observations, observers are required to rate phenomena along a dimension that is typically bipolar (eg, passive/aggressive or excellent health/poor health); ratings are made either at specific intervals during the observations (eg, every 15 minutes) or after observations are completed.

    16:Some structured observations use sampling to select behaviors or events to be observed. Time sampling involves the specification of the duration and frequency of both observational periods and intersession intervals. Event sampling selects integral behaviors or events of a special type for observation.

    17:Technological advances have greatly augmented researchers' capacity to collect, record, and preserve observational data. Such devices as audiotape recorders and videotape cameras permit behaviors and events to be described or categorized after their occurrence.

    18:Observational methods are subject to various biases. The greater the degree of observer inference and judgment, the more likely that perceptual errors and distortions will occur. The most prevalent observer biases include the enhancement of contrast effect, central tendency bias, the halo effect, assimilatory biases, errors of leniency, and errors of severity.