Soc Methods Final Exam

Front 1

Subjects

Back 1

Subjects: the name for people who are studied and participate in experimental research

Front 2

stimulus/treatment

Back 2

What the independent variable in experimental research is called

Front 3

pretest/posttest

Back 3

Pretest: the measurement of the dependent variable of an experiment prior to the treatment
Posttest: the measurement of the dependent variable of an experiment after the treatment

Front 4

experimental/control groups

Back 4

Experimental group: the group that receives the treatment
Control groups: the group that does not get the treatment

Front 5

double-blind

Back 5

A type of experimental research in which neither the subjects nor the person who directly deals with the subjects for the experimenter know the specifics of the experiment; used to eliminate experimenter expectancy (an internal validity threat)

Front 6

placebo effect

Back 6

Placebo effect: psychological effects of a treatment; although subjects are receiving a false treatment and not the actual treatment, they still show signs of receiving the treatment.

Front 7

confederate

Back 7

Confederate: people who pretend to be other subjects or bystanders but who actually work for the researcher and deliberately mislead subjects

Front 8

Solomon 4-group design

Back 8

An experimental design in which subjects are randomly assigned to two control groups and two experimental groups. Only one experimental group and one control group receive a pretest. All four groups receive a posttest. This design is supposed to address the issue of pretest effects (pretests can sometimes lead to improved posttest scores)

Front 9

cross-sectional and longitudinal designs

Back 9

cross-sectional is a study done at one point in time; longitudinal is a study with a cohort group over time (longitudinal is often better, but more difficult to conduct)

Front 10

trend study

Back 10

?

Front 11

panel study

Back 11

?

Front 12

cohort study

Back 12

?

Front 13

Experimental design

Back 13

Arranging the parts of an experiment and putting them together

Front 14

logic

Back 14

Experimental logic: The experimenter’s basic logic extends commonsense thinking. The three things researchers do in an experiment are (1) begin with a hypothesis, (2) modify something in a situation, and (3) compare outcomes with and without the modification. Compared to the other social research techniques, experimental research is the strongest for testing causal relationships because the three conditions for causality are best met in experimental designs.

Front 15

notation

Back 15

Shorthand system for symbolizing the parts of experimental design.

O= observation of dependent variable, X= treatment/independent variable, R= random assignment, O1=pretest, O2= posttest, If there are multiple Xs then they are numbered with subscripts to distinguish them (X1, X2, etc.)

Front 16

methods of assigning subjects to groups

Back 16

randomization or matching

Front 17

randomization

Back 17

Randomization (aka random assignment): dividing subjects into groups at the beginning of experimental research using a random process, so the experimenter can treat the groups as equivalent. Random means a case has an exactly equal chance of ending up in one or the other group.

Front 18

matching

Back 18

Matching: researchers match cases in groups on certain characteristics, such as age and sex. Matching is an alternative to random assignment, but it is infrequently used. Matching presents a problem: individual cases differ in thousands of ways—how can the researcher decide which characteristics are relevant and can one locate exact matches?

Front 19

Equivalent groups.

Back 19

Mandatory in an experiment; only achieved through random assignment

Front 20

Isolating treatment effects and establishing causality.

Back 20

A researcher wants to control all aspects of the experimental situation to isolate the effects of the treatment and eliminate alternative explanations. Researchers use deception (intentionally misleads subjects) to control the experimental setting. The use of pretests, posttests, manipulation check, and control groups help to establish causality.

Front 21

Strengths and weaknesses of experimental method

Back 21

Strengths and weaknesses of experimental method: The real strength of experimental research is its control and logical rigor in establishing evidence for causality. In general, experiments tend to be easier to replicate, less expensive, and less time consuming than the other techniques. Experimental research also has its limitations: (1) some questions cannot be addressed using experimental methods because control and experimental manipulation are impossible, (2) experiments usually test one or a few hypotheses at a time. This fragments knowledge and makes it necessary to synthesize results across many research reports and (3) external validity is another potential problem because many experiments rely on small nonrandom samples of college students.

Front 22

laboratory and field experiments

Back 22

Laboratory experiments have strong internal validity because situations are carefully controlled to isolate the effects of treatment. However, findings may not apply outside of the experimental setting (low external validity). Field experiments have external validity because they are in real-world settings. However internal validity is sacrificed because there are no control or equivalent groups.

Front 23

Validity issues

Back 23

include Hawthorne effect; history; maturation; mortality; pretest; contamination;
subject demoralization and rivalry; cover story; manipulation check; debriefing; internal and
external validity

Front 24

Hawthorne effect

Back 24

An effect of reactivity named after a famous case in which subjects reacted to the fact that they were in an experiment more than they reacted to the treatment

Front 25

history

Back 25

A threat to internal validity due to something that occurs and affects the dependent variable during an experiment, but which is unplanned and outside the control of the experimenter

Front 26

maturation

Back 26

A threat to internal validity due to natural processes of growth, boredom, and so on, that occur to subjects during the experiment and affect the dependent variable

Front 27

mortality

Back 27

Threats to internal validity due to subjects failing to participate through the entire experiment

Front 28

pretest

Back 28

Threatens internal validity because more than the treatment alone affects the dependent variable. The Solomon four-group design helps a researcher detect testing effects.

Front 29

contamination

Back 29

A threat to internal validity that occurs when the treatment “spills over” from the experimental group, and control groups subjects modify their behavior because they learn of the treatment

Front 30

subject demoralization and rivalry

Back 30

Subject demoralization: Dr. Marshall discussed demoralization during example of “Laughter is the Best Medicine”—control group heard laughter of experimental group watching the funny movie and wondered why they weren’t having as much fun as the other group;demoralization
Rivalry: control group changes performance to spite experimenters or experimental group

Front 31

cover story

Back 31

Cover story: the deception told to subjects to create experimental reality

Front 32

manipulation check

Back 32

Manipulation check: check for the validity of X; make sure the subject is interpreting the treatment in the way it was designed (i.e., in Good Samaritan experiment, check to make sure subject saw the confederate and knew he/she was in need of help. When subjects arrive at destination they are asked if they saw anything unusual on the way over—checking for X)

Front 33

debriefing

Back 33

Debriefing: ethical obligation to explain to subject the true meaning of the experiment after using deception

Front 34

internal and external validity

Back 34

Internal validity: the ability of experimenters to strengthen a causal explanation’s logical rigor by eliminating potential alternative explanations for an association between the treatment and the dependent variable through an experimental design

External validity: the ability to generalize from experimental research to settings or people that differ from the specific conditions of the study

Front 35

Interview v. questionnaire

Back 35

A questionnaire is a self administered survey; In an interview there is some interaction between the researcher and respondent, and interview is more reactive, interview has open questions

Front 36

Element

Back 36

if you sample colleges, then sample students from those colleges, the students are the sampling elements

Front 37

unit

Back 37

units of analysis: for example, households or colleges, not always individuals

Front 38

population

Back 38

Population- the large pool of element or cases; the unit being sampled; the geographical location, and temporal boundaries of populations.

Front 39

sample

Back 39

Sample- group which you are studying that is representative of the general population you’re observing

Front 40

parameter

Back 40

a characteristic of the entire population that is estimated from a sample.

Front 41

statistic

Back 41

a numerical estimate of a population parameter computed from a sample

Front 42

sampling frame

Back 42

a list of cases in a population, or the best approximation of it

Front 43

weighting

Back 43

Weighting- in an index, the researcher values or weights some items more than others.
Unless otherwise stated, assume that an index is unweighted. Like wise, unless you have a good theoretical reason for assigning different weights, use equal weights. Changes the theoretical definition of the construct. Can produce different index scores, but in most cases weighted and unweighted yield similar results.

Front 44

normal curve

Back 44

Normal Curve- a bell shaped curve; shaped by the sampling distribution when there is a huge number of random samples. *The midpoint of the
curve approaches the population parameters as the sample increases.

Front 45

unit of analysis

Back 45

Unit of Analysis- the kind of empirical case or unit that a researcher observes, measures, and analyzes in a study.

Front 46

sampling distribution

Back 46

Sampling Distribution- the set of many random samples. A distribution of different samples that shows the frequency of different samples outcomes from many separate random samples. The pattern in the sampling distribution suggests that over many separate samples, the true population parameter is more common than any other result. *When many different random samples are plotted, than the sampling distribution looks like a bell curve.

Front 47

inferential statistics

Back 47

Inferential Statistics- a branch of applied mathematics or statistics based on a random sample. It lets a researcher make precise statements about the level of confidence he or she has in the results of a sample being equal to the population parameter.

Front 48

hidden population

Back 48

Hidden Population- people who engage in clandestine, deviant, or concealed activities and who are difficult to locate and study. (e.g. illegal drug users, prostitutes, homosexuals, people with HIV/AIDS, homeless people and others)

Front 49

Logic of sampling

Back 49

First you establish… What you would like to talk about (the population to study; get a sampling frame)
- Then… The Sampling Process
- Finally…What you actually observe in the data (sample)

Front 50

Issues of representativeness

Back 50

include size, heterogeneity v. homogeneity. Probability v. nonprobability designs. Logic & interpretation of standard error, confidence interval, and confidence level

Front 51

size

Back 51

- sample size depends on the kind of data analysis the researcher plans.
- how accurate the sample has to be for the researcher’s purposes, and on population characteristics.
- a large scale size alone doesn’t generate a representative sample.
- samples should have a good sampling frame
- good samples for qualitative purpose can be very small

Front 52

heterogeneity v. homogeneity.

Back 52

Heterogeneity- everything being equal, larger samples are needed if one wants high accuracy if the population has
a great deal of variability, or if one wants to examine many variables in the data analysis simultaneously; The more heterogenous the sample, the larger the sampling error
§ Homogeneity- smaller samples are sufficient when less accuracy is acceptable, when the population is homogenous, or when only a few variables are examined at a time.

Front 53

Probability v. nonprobability designs

Back 53

Probability: better with large samples and generally more accurate, higher external validity, allows for estimating error, must have an exhaustive sampling frame & use random selection
NonProbability: when you don't have a sampling frame, no time or money for probability, for pretesting, for small samples, for theoretical reasons

Front 54

Logic & interpretation of standard error

Back 54

Logic & Interpretation of Standard Error- referred to as sampling error. The deviation between sample results and a
population parameter due to random processes.To specify exactly the probable margin of error in the estimate.

Front 55

confidence interval

Back 55

Confidence Interval- a range of values, usually a little higher and lower than a specific value found in a sample, within which a researcher has a specified and high degree of confidence that the population parameter lies. *We only select one sample and we don’t know the population parameter, we try to estimate it. By random sample selection, we can use probability theory to draw a band around the sample statistic.

Front 56

confidence level

Back 56

Confidence Level- to estimate the odds. We can do this because probability theory specifies that 68% of a sampling distribution will produce estimates falling
within one standard error of the parameter. (This is an extension of what we know about the normal curve, in which 68% of values fall within 1 standard deviation of the mean, 95% of values fall within 2 standard deviations of the mean, 99.9% of values fall within 3 standard deviations of the mean.)

Front 57

simple random

Back 57

prob sample; uses random numbers table

Front 58

systematic random

Back 58

prob sample; fixed interval; k is the interval size: K = sampling frame size/ desired sample size

Front 59

stratified random

Back 59

prob sample; dividing into homogenous strata; good for sampling rare elements

Front 60

area/multistage cluster

Back 60

prob sample; use sampling units (colleges), calculate standard error, sample elements (students), calculate standard error, combine SEs to estimate confidence interval

Front 61

convenience

Back 61

non prob; survey people who walk by; suseptible to bias

Front 62

quota

Back 62

non prob; asks a filter question to match sample characteristics to the aggregate in proportion; limited by how much matching you can do

Front 63

purposive

Back 63

non prob; rejects EPSEM although they may have a sampling frame; seeks out ideal types, key positions, etc

Front 64

snowball

Back 64

non prob; sample of last resort; accumulates slowly, sample size may be 10-20; referrals give homogenous samples; good for hidden populations

Front 65

Definitions of scale and index

Back 65

Scale- A measure in which a researcher captures the intensity, direction, level, or potency of a variable construct. It arranges responses or observations on a continuum. Can use a single indictor or multiple indicators, most are at ordinal level of measurement.
Index- A measure in which a researcher adds or combines several distinct indicators of a construct into a single score. This composite score is often a simple sum of the multiple indicators. It is used for content and convergent validity. Indexes are often measured at the interval or ratio-level.

Front 66

why use composite measures

Back 66

Content Validity- Multidimensional variables. (coverage over consensus)
Reliability- Measured in consistency, use of multiple measures enhances reliability.
Measurement refinement

Front 67

intensity structure

Back 67

?

Front 68

scoring

Back 68

amalgate scores into summary measure

Front 69

rule of parsimony

Back 69

In science, less is best. One score with multiple indicators.(Redundant data violates the rule of parsimony)

Front 70

Index/scale similarities and differences

Back 70

Scales and Indexes must be mutually exclusive, exhaustive, and unidimensional.

1.Mutually Exclusive- An individual or a case fits into one and only one attribute of a variable.

2.Exhaustive- All cases fit into one of the attributes of a variable.

3.Unidimensional- All the items in a scale or index fit together, or measure a single construct. Combine several specific pieces of information into a single score or measure, have all the pieces work together and measure the same thing.

Both indexes and scales produce ordinal or interval level measures of a variable
Both can be combined into one measure
Scales and indexes give a researcher more information about variables and make it possible to asses the quality of measurement. Both increase reliability and validity, and they simply the information that is collected.

Front 71

index and scale construction

Back 71

Purpose- Creates an ordinal, interval, or ratio measure of a variable expressed as a numerical score. (For Example, can help measure the most desirable place to live based on unemployment rates, commuting time, crime rate, recreation opportunities, weather…etc)
Weighting (Unless otherwise stated, assume that an index is unweighted.)
Unweighted Index- Gives each item equal weight. It involves adding up the items without modification, as if each were multiplied by 1 ( or by -1 for items that are negative)
Weighted Index- Researcher values or weights some items more than others. Size of weights can come from theoretical assumptions, the theoretical definition, or a statistical technique such as factor analysis. Weighting changes the theoretical definition of the construct.
Purpose- Creates ordinal, interval, or ratio measure of a variable expressed as a numerical score. Common when a researcher wants to measure how an individual feels or thinks about something. Some call this HARDNESS or POTENCY OF FEELINGS. Two purposes for using scales: Scales help in the conceptualization and operationalization processes- show the fit between a set of indicators and a single construct. Scales produce quantitative measures and can be used with other variables to test hypotheses.

Front 72

item analysis

Back 72

does the item have to do with what you're measuring? does it belong in the scale or index?

Front 73

internal and external validation

Back 73

Internal Validity- Means that there are no errors internal to the design of the research project. Used primarily in experimental research to talk about possible errors or alternative explanations of results that arise despite attempts to institute controls.
**High Internal Validity- Means that there are few such errors
**Low Internal Validity- Means that such errors are likely.
External Validity- Ability to generalize findings from a specific setting and small group to a broad range of setting and people. It addresses the question, if something happens in a lab or among a particular group of subjects can the findings be generalized to the ‘real’ world or to the general public.
** High External Validity- Means that the results can be generalized to many situations and many groups of people.
**Low External Validity- Means that the results apply only to a very specific setting.

Front 74

Likert

Back 74

widely used and very common in survey research. Usually ask people to indicate whether or not they agree or disagree with a statement. Require a miniumum of two categories such as “agree” and “disagree”. (Examples of type of Likert Scales Table 5.4 on p.131). Response choices must be evenly balanced. Ordinal level of measurement where responses only indicate a ranking. Response Set (Response bias or response style)- Tendecny of some people to answer a large number of items in the same way (usually agreeing) out of laziness or a psychological predisposition.
Two Limitations to using: Different combinations of several scale items can result in the same overall score or result and the response set is a potential danger.

Front 75

Index v. Scale

Back 75

Indexes are easier to construct; You can combine several scale questions in a composite index if they are all measuring a single construct

Front 76

Bogardus

Back 76

Measures the social distance separating ethnic or other groups from each other. It is used with one group to determine how much distance it feels toward a target or ‘out-group’.
People respond to a series of ordered statement; those that are most threatening or most socially distant are at one end, and those that might be least threatening or socially intimate are at the other end. The logic of the scale assumes that a person who refuses contact or is uncomfortable with the socially distant items will refuse the socially closer items.
Coding is different: 0= yes 1= no. Scale has an intensity structure empirically, questions go from easiest to hardest.
Two limitations- Researcher needs to tailor the categories to a specific out-group and social setting. Two, it is not easy for a researcher to compare how a respondent feels toward several different groups unless the respondent completes a similar social distance scale for all out-groups at the same time.

Front 77

Guttman scales

Back 77

Differs from Bogardus and Likert Scales because researchers use it to evaluate data after they are collected. Guttman scaling begins with measuring a set of indicators items. These can be questionnaire items, votes, or observed characteristics. Indicators are generally measured in a yes/no or present/absent fashion.
The logical relationship among items in Guttman Scaling is hierarchical.
Scalogram Analysis- Lets a researcher test whether a hierarchical relationship exists among the items.
Scalable- Capable of forming a Guttman Scale, if a hierarchical pattern exists.
The strength or degree to which items can be scaled is measured with statistics that measure wether the responses can be reproduced based on a hierarchical pattern.
Most range from zero to 100 percent. A score of zero indicators a random pattern, or no hierarchical pattern. A score of 100 indicates that all responses to the answer fit the hierarchical or scaled pattern.

Front 78

coefficient of reproducibility

Back 78

w/ Guttman Scales; (1 - # errors/ # of guesses) = (1 - nonscale types/ N cases x N items)

Front 79

Univariate analysis

Back 79

describes one variable in isolation

Front 80

frequency distribution

Back 80

the easiest way to describe numerical data of one variable. It is a table that show the distribution of cases into the category of one variable. Ex: the % of cases in each category

Front 81

table construction

Back 81

label & name the table, say when; list indicators and frequency of each; give N and missing

Front 82

relative v. absolute frequency

Back 82

relative frequency refers to the percentages and absolute refers to the total number of respondents.

Front 83

central tendency

Back 83

measures of the center of the frequency distribution that helps in summarizing the information about one variable into a single #. Ex: mean, median mode—which are often called averages.

-will all be equal if frequency distribution has a normal (bell shaped) curve—if it is skewed the the 3 will not be equal.

Front 84

variation

Back 84

spread dispersion and variability around the center—no variation means everyone is the same
-variation can be measured in 3 ways: range, percentile, and standard deviation
-range: largest and smallest score
-percentile: tell score at a specific place within the distribution
-standard deviation: most comprehensive and widely used.
-require interval or ratio level of measurement
-tells us who is more similar—the lower the S.D. the more similar the subjects in the group are.

Front 85

bell curve

Back 85

bell curve is formed from frequency distribution. In a bell curve (or normal curve) the three measures of central tendency are equal