Beruflich Dokumente
Kultur Dokumente
Whats so Bad about Lab Experiments? Laboratory setting is artificial Participants know they are being studied Demand characteristics Experimenter Expectancies Smart vs. Dull rats Reactivity Socially desirable responses!
Field Experiments
Natural settings leads to high external validity Higher ecological validity Higher temporal validity Better generalizability Can use samples which can not come to lab Greater experimental realism Useful to ensure generality of lab studies
Field Experiments Why are they Bad? Hawthorne effect Ethical issues Invasion of privacy Manipulating variables might be dangerous Failure to obtain informed consent Less internal validity Loss of control MANY extraneous variables
Descriptive Methods
Descriptive methods do not involve manipulation of an independent variable. When we use descriptive methods, we can only speculate about causation that may be involved. Comparisons with the Experimental Method We are not able to exercise any control with regard to gathering these data and cannot make any statements regarding cause-andeffect.
Descriptive Methods
What
are the advantages 1. Descriptions are informative 2. Descriptions are the starting point for building operational definitions 3. Description is sometimes the only ways to study behavior
Deciding how to present the results of your research project. Qualitative presentation of results: Report consists of a description of the behavior in question (a narrative record) and the conclusions prompted by this description. Quantitative or numerical presentation of results: Need to know how behavior under investigation is going to be measured and how these measurements will be analyzed.
Archival and Previously Recorded Sources of Data Refers to use of data recorded by other individuals for other purposes public health and census data The General Social Survey is an archival source that can be accessed online http://www.icpsr.umich.edu/gss/home.htm
Potential Problems With Archival and Previously Recorded Sources of Data You will not know exactly who left the data you are investigating. The participants may have been selective in what they chose to write. This problem is also known as selective deposit. Archival and previously recorded sources of data may not survive long enough for you to make use of them.
Good We see behavior which is real. Bad No informed consent Susceptible to experimenter expectancies We usually cannot randomly sample Usually we cannot quantify responses Low internal validity
Case Studies: Involves intense observation and recording of behavior of a single (perhaps two) participant(s) over an extended period of time. There are no guidelines for conducting a case study and the procedures employed, behaviors observed, and reports produced may vary substantially. Frequently used in clinical settings to help formulate ideas and hypotheses for further research.
Descriptive Methods Observational Techniques Naturalistic Observation: Involves seeking answers to research questions by observing behavior in the real world.
The
first goal of naturalistic observation is to describe behavior as it occurs in the natural setting without the artificiality of the laboratory. The second goal of naturalistic observation is to describe the variables that are present and the relations among them.
Why should the researcher be concealed or unobtrusive in a study using naturalistic observation? The reactance or reactivity effect Refers to biasing of the participants responses because they know they are being observed. The reactivity effect is also known as the Hawthorne effect because of the location of the original study.
What are the drawbacks and weaknesses of the participant observer technique? An extended period of time may be necessary before the participant observer is accepted as a member of the group that is under study. Cannot make cause-and-effect statements.
Using More than One Observer There are two main reasons for using more than one observer: One observer may miss or overlook a bit of behavior. There may be some disagreement concerning exactly what was seen and how it should be rated or categorized.
Using More than One Observer When two individuals observe the same behavior, it is possible to see how well their observations agree. The extent to which the observers agree is called interobserver reliability. Low interobserver reliability indicates that the observers disagree about the behavior(s) they observed. High interobserver reliability indicates agreement.
85% agreement is generally considered to be an acceptable minimum level for interobserver reliability.
Clinical perspective Schein (1987) argued convincingly that the clinical perspective or model is not a subcategory of participant observation because:
A client typically chooses the clinician, whereas the participant observer chooses the others to be studied. Unlike participant observers, clinicians cannot be unobtrusive because they have been asked to participate in the situation. Although the participant observer can remain passive, clinicians must intervene in the situation. The participant observers goal is understanding, whereas the clinicians goal is helping. Participant observers validate their findings by replication while clinicians validate their findings by being able to predict the results of a given intervention.
different time periods in order to obtain a more representative sampling of the behavior of interest.
of time periods may be determined randomly or in a more systematic manner. The use of time sampling may apply to the same or different participants.
Selection
Situation Sampling: Involves observing the same behavior in several different situations. This techniques offers the researcher two advantages:
By sampling behavior in several different situations, you are able to determine whether the behavior in question changes as a function of the context in which you observed it. You are likely to observe different participants in the different situations and because different individuals are observed, your ability to generalize any behavioral consistencies across the various situations is increased.
A test should measure the nature and extent of individual differences A good test differentiates (discriminates between groups) people based on their true scores Why Use Tests? To help determine outcomes of experiments To diagnose strengths and weaknesses To assist in placing individuals in appropriate programs To assist in selecting applicants To evaluate a programs effectiveness
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories Surveys typically request our opinion on some topic or issue that is of interest to the researcher. Types of Surveys Descriptive Analytic
Descriptive seeks to determine what percentage of the population has a certain characteristic, holds a certain opinion, or engages in a particular behavior. Examples: The Gallup Poll Nielsen television ratings
Analytic seeks to determine what the relevant variables are and how they might be related. Questions need to be chosen carefully. Pilot testing is generally necessary to determine the type of questions that will be used in the final survey instrument.
Developing a Good Survey or Questionnaire (see table 5-1, p. 100). Step 1 decide what type of instrument to use. How will the information be gathered? Will you use a mail survey, conduct a phone interview, or conduct the survey in person? Will you use trained interviewers?
Yes-No Questions The respondent answers yes or no to the items. Forced Alternative Questions The respondent must select between two alternative responses. Multiple-Choice Questions The respondent must select the most suitable response from among several alternatives. Likert-Type Scales The individual answers a question by selecting a response alternative from a designated scale. A typical scale might be the following: (5) strongly agree, (4) agree, (3) undecided, (2) disagree, or (1) strongly disagree. Open-Ended Questions A question is asked to which the respondent must construct his or her own answer. Step 2 identify the types of questions to use.
Step 1 decide what type of instrument to use. How will the information be gathered? Step 2 identify the types of questions to use. Step 3 write the items. They should be clear, short, and specific.
Use familiar vocabulary and be at the reading level of the individuals you intend to test.
It is important to ask others, especially professionals who have expertise in your area of research interest, to review your items. They may be able to detect biases and unintended wordings that you had not considered.
Step 5 determine the relevant demographic data to be collected. Includes items such as age, sex, annual income, size of community, academic major, and academic classification. Step 6 determine administration procedures and develop instructions. Instructions must be clear, concise, easy to follow, and thoroughly rehearsed (for in-person or phone interviews). How will informed consent be obtained?
researcher does not have to be present while the survey is being completed. Surveys can be sent to a much larger number of participants than a single researcher could ever hope to contact in person.
Disadvantages
The
researcher cannot be sure who actually completes the survey. There is no guarantee the respondent answered the questions in the same order in which they appeared in the survey. Low return rate.
a trained interviewer administers a survey in a respondents home, the response rate climbs dramatically. The trained interviewer is able to clarify ambiguous questions, make sure that all questions are answered in the proper sequence, and generally assist with any problems that the respondents may experience.
Disadvantages
The
potential for considerable expenditure of time and money exists. People not always at home.
Telephone Interviews - Disadvantages Incoming calls may be screened and potential respondents lost. It is easy to say no to a phone interviewer. Cannot use visual aids to clarify questions. Cannot evaluate nonverbal cues such as facial expressions, gestures, and posture. It is more difficult to establish rapport.
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories Tests and Inventories Tests and inventories are designed to assess a specific attribute, ability, or characteristic possessed by the individual being tested. Characteristics of Good Tests and Inventories Validity a test or inventory has validity when it actually measures what it is supposed to measure.
Aptitude
Are used to assess an individuals ability or skill in a particular situation or job. Measure specific aspects of an individuals motivational state, interpersonal capability, or personality.
Personality
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories To Use Multiple Choice Items or Not?
Disadvantages Advantages Limit students options Versatile No opportunity to Easy to score practice writing Simple to take Some people dont do Poor writers not penalized well on them Good items used again Limits content to be Good distracters are assessed diagnostic Items must be well Hard to fake correct answer written
Agree
No Strong Feeling
____ ____
Disagree
____ ____
____ ____
____ ____
____ ____
____ ____
____
____
____
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories Likert Scales Statements indicating an attitude toward a topic are written Items with clearly positive or negative attitudes are selected Statements are listed with a space for respondent to indicate degree of agreement
Government has no business funding child care programs. Child care should be supported by federal, state, and local tax dollars.
SD SD
D D
U U
A A
SA SA
Items can be weighted Weights of unfavorable items are reversed Average score is computed
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories Questionnaires What they are
and pencil tests with structured questions Self-administered Advantages Can be mailed out Survey broad geographic area Cheaper than one-on-one interview Respondents may be more honest Data easy to share with other researchers Disadvantage Low return rate (sample unrepresentative)
Paper
Descriptive Methods Surveys, Questionnaires, Tests, and Inventories Basic Assumptions of a Questionnaire Does not make unreasonable demands upon the respondent Does not have a hidden purpose Requests information that respondents presumably have Contains interesting questions Does not request information that could be obtained by other means
Are
more reliable
strong are demand characteristics? Must responding be structured for the participants? What breadth of information is needed?
Interviews are preferred when the researcher must react to a participants responses. But questionnaires are more reliable, especially if demand characteristics are an issue.
Are questions worded correctly? Are double-barreled questions avoided? "Sales reps are polite and responsive." Are leading questions avoided? "How satisfied are you with the compensation policies of our company? Implies satisfaction even before answering Are Barnum statements avoided? Do you sometimes worry? Are undefined terms avoided?
What is the response scale? Are measurements sensitive to subtle differences? Are multiple-choice or Likert-type questions appropriate? What description should anchor Likert-type questions? How many choices should each scale contain? Is an odd or even number of choices needed?
practice questions needed? Are funnel or filter questions needed? Are demand characteristics a problem? Are ethical obligations being met? Have order effects been avoided? Are alternate forms needed? Are catch trials needed? Are clear instructions provided? Is a pilot study needed?
Qualitative Research
is defined as an inquiry process of understanding a social or human problem, based on building a complex, holistic picture, formed with words, reporting detailed views of informants, and conducted in a natural setting (Creswell, 1994, p. 2). The qualitative researcher believes that a full description of human behavior includes peoples feelings in addition to what they are doing and how they are doing it. The qualitative research style is much less formal and impersonal, and the reader of a qualitative research report can expect to find such additions as definitions that evolved during a study (Creswell, 1994, p. 7).
Qualitative Research
The qualitative researcher is committed to studying particular people in specific settings. Qualitative researchers prefer to use inductive logic. Qualitative research begins with guiding hypotheses reflecting a global issue of interest (Marshall & Rossman, 1989). Qualitative researchers typically analyze their data simultaneously with data collection, data interpretation, and narrative reporting writing (Creswell, 1994, p. 153). Grounded theory is one of the most popular forms of contemporary qualitative research.
Qualitative Research
Grounded
1990)
The
ultimate goal of this approach is to derive theories that are grounded in (based on) reality. Grounded theory is not advocated for all types of research questions. Knowing the literature too well can hamper the creativity necessary to doing grounded theory research. The heart of the grounded theory approach occurs in its use of coding, which is analogous to data analysis in quantitative approaches.
Qualitative Research
Types of coding in grounded theory research (Strauss & Corbin, 1990): Open Coding: The researcher labels and categorizes the phenomena being studied. Axial Coding: Involves finding links between categories and subcategories from open coding. Selective Coding: Entails identifying a core category and relating the subsidiary categories to this core.
Qualitative Research
Grounded Theory (Strauss & Corbin, 1990) From the process of selective coding, the grounded theory researcher moves toward developing a model of process and a transactional system, which essentially tells the story of the outcome of the research.
Process refers to a linking of actions and interactions that result in some outcome . A transactional system is grounded theorys analytical method that allows an examination of the interactions of different events. The transactional system is depicted in a conditional matrix such as that shown in figure 4.2
Qualitative Research
Participatory Action Research (Fine et al, 2003) Goal of a PAR project is usually to evaluate and understand the impact of some social program on the community. Typically involves the participants as equal researchers in the study. PAR approach can use one or more specific methods to gather information. These may be both quantitative and qualitative.
Correlational Research
Looks to the degree of relationship between two variables Relationships are only discovered ex post facto (after the fact, as opposed to a priori), observations are collected, then the correlation is computed. The absolute magnitude of a correlation indicates the degree of relationship.
absence of random assignment allows for potential confoundings The order of occurrence of the variable is unknown
Ethics Inexpensive Lab experiments are sterile Allows us to summarize an entire relationship with one number. Lead to better prediction than experiments
# hours spent volunteering
30 25 20 15 10 5 0
Correlational Research
The sign of the correlation indicates the direction of the relationship. With positive correlations, as one of the variables increases, so does the other, conversely, as one of the variables decreases, so does the other. With negative correlations, as one of the variables increases, the other decreases.
Correlational Research
The Pearson r is by far the most commonly used correlation Can only be used if both variables are measured with interval or ratio scales of measurement and a linear relationship is assumed.
25 20 15 10 5 0 0 50000 Income 100000 150000
Happiness
Happiness
Happiness
15 10 5 0 40000
50000 Income
60000
70000
Happiness
Correlational Research Linear Regression Attempts to predict the value of one variable based on the value of another by creating a line of best fit. Y = bX + a Standard error of the estimate Coefficient of determination (r2)
16 14 12 10
Happiness
8 6 4 2
0 70000
90000
110000 Income
130000
150000
The ability of a test to measure the same thing over and over. Validity: The ability of a test to measure what it is intended to measure. All tests have SOME reliability and validity, some are better than others though.
A reliable measure may or may not be valid but a valid measure must be reliable.
Conventional Views on Reliability A score consists of two things, true score + error If error is high, reliability is low. Temporal Stability-The same form given on separate occasions should yield similar scores. Form Equivalence-If given alternate forms of the same test scores should be similar. Internal Consistency-Within the test itself, scores on individual questions should be similar to each other.
Reliability is a necessary but not sufficient condition for validity. For instance, if the needle of the scale is five pounds away from zero, I always over-report my weight by five pounds. Is the measurement consistent? Yes, but it is consistently wrong! Is the measurement valid? No!
Correlational Research Increasing Reliability Increase sample size Eliminate unclear questions Standardize testing conditions Use both easy and difficult questions Minimize the effects of external events Standardize instructions Maintain consistent scoring procedures
Internal Validity: The extent to which a good (clean) measure (or experiment) has been used, a clean measure is far more likely to reflect the concept intended.
External Validity: The extent to which a measure truly measures what is intended.
Face Validity
The extent to which a measurement procedure appears to measure what it is intended to measure. Not always useful because things that look right may not be and things that dont look right might be.
Construct Validity
Construct validity refers to the degree to which inferences can legitimately be made from the operationalizations in your study to the theoretical constructs on which those operationalizations were based.
In order to demonstrate construct validity, a measure must establish both convergent and discriminant validity.
Convergent Validity
The extent to which scores from one procedure are positively correlated with scores obtained from another procedure that is already accepted as valid.
The extent to which the scores obtained from one procedure are not correlated with the scores obtained from another procedure that measures other variables or constructs.
My new test which is intended to measure selfesteem is discriminantly valid if the scores do NOT correlate with scores on Rotters Locus of Control test.
Predictive Validity
The extent to which a procedure allows for accurate predictions about a subjects future behavior.
Concurrent Validity
The extent to which a procedure correlates with the present behavior of participants.
Correlational Research Where do you get data? Often, data can be gathered from other places. If studying relationship between ACT score and GPA, all I have to do is look it up! Other cases you may need to measure! If possible, use a published measuring device such as an IQ test. If no measure exists, make one!
Sampling Techniques
Probability Sampling Every potential participant in the population has an equal likelihood of being selected for participation in the study. Nonprobability Sampling Every potential participant does not have an equal likelihood of participating.
Simple
Random Sampling ALL members of population have equal chance, this is the ideal.
Stratified Random Sampling Identify important subgroups, then proportionate random selection from them. If TWU has 60% female and 40% male, then, if you need N = 100, randomly sample 60 females and 40 males.
Cluster Sampling Certain groups are randomly selected, then all members of the group participate (i.e. a state might go to a few randomly selected schools to see if a program works).
Sampling Techniques Nonprobability Sampling Quota Sampling We ensure that the sample has the same percentage of each subgroup as the population. If the left/right handed ratio is 10/90, then we would ensure that we choose 1 lefty for every 9 right handed people. The people are chosen by convenience.
Snowball Sampling Researcher identifies participants based on identification by previous participants. Useful if your population is hard to find. Members of rare populations tend to know each other and know how to find others like themselves. If you are studying people with Hodgkins Disease, most participants will know others from support groups or other places.
Small N Designs
Baseline
A-B
Reversal
A-B-A
Multiple-baseline
A-B-A-B
Variations A-B-A-C-A-B-A
Quasi Experiments
Participants
8 7
Driving Abililty
Driving Ability
Quasi-independent variables come from three sources Participant variables Environmental events Passage of time
Gender
Height
Often
we want to create groups based on a pretest score perhaps top and bottom 25% on a selfesteem test
Might
come from common experiences People that live in LA vs. NY People that experienced an earthquake Time-series designs
SF
Seattle Before After
time-series design
time-series design
Presidents
Others
Maturation
Longitudinal
designs
O O O O O O