H 515 Osu Research Methods Midterm

Front 1

Unit of Analysis

Back 1

Definition: the entities (objects or events) under study
-often the individual
-other units include dyads, families, social networks,
-organizations, and geographical units
Generally speaking if you are talking about rates, your not talking about individuals.

Front 2

Ecological Fallacy

Back 2

-occurs when properties of groups or geographical areas are used to make inferences about individuals.
-"fallacy of division" -assuming that what holds true of a group also is true of individuals within the group
-Ex an article looking at the relationships between nativity and illiteracy found a correlation between foreign birth and illiteracy (but there are other possible explanations)
-Ex: if Sally went to a college whose students had relatively low average SAT scores, you would commit this fallacy if you assumed Sally herself had low SAT scores

Front 3

Individualistic Fallacy

Back 3

may occur when scientists generalize from individual behavior to collective relationships.

Ex: At the state level-the SES factors are better predictors of mortality than other levels

Front 4

Cross-level Fallacy

Back 4

may occur when making an inference from one sub-population to another at the same level of analysis.

Front 5

Universal Fallacy

Back 5

the error of generalizing a universally true relationship from one particular subsample.

Front 6

Explanatory variables

Back 6

variables that are the object of study- part of some specified relationship
2 types of explanatory variables:
-Dependent variables
-Independent variables

Front 7

Dependent variables

Back 7

-type of explanatory variable; the one the researcher is interested in explaining and predicting
-Variation on the dependent variable is thought to depend on or to be influenced by certain other variables.

Front 8

Independent variables

Back 8

type of explanatory variable; the variables that do the influencing and explaining
-Also called predictor variables because their values or categories may be used to predict the values or categories of dependent variables

Front 9

Extraneous variables

Back 9

Not part of the explanatory set; looking at, but not at the thing you are trying to explain
-everything that s not an explanatory variable
-May be antecedent or intervening

Front 10

Antecedent variables

Back 10

type of extraneous variable; occurs prior in time to both the independent and dependent variable

Front 11

Intervening/ mediating variables

Back 11

type of extraneous variable; a variable is intervening if it is an effect of the independent variable and a cause of the dependent variable.

Front 12

Control variables

Back 12

-held constant, or prevented from varying, during the course of observation or analysis
-this may be done to limit the focus of the research or to test hypothesis pertaining to specific subgroups.
*Whenever a variable is held constant in research, that variable cannot account for (or explain) any of the variation in the explanatory variables.

Front 13

Moderating/modifying variables

Back 13

A variable the affects the relationship between 2 variables

Front 14

Quantitative variables

Back 14

values or categories consist of numbers and if differences between its categories can be pressed numerically.

Front 15

Qualitative variables

Back 15

have discrete categories, usually designated by words or labels, and nonnumerical differences between categories.

Front 16

Evidence for inferring causality

Back 16

-Association between the IV and DV
-DV varied after the IV (direction of influence)
-Association probably not spurious (nonspuriousness; elimination of rival hypotheses).

Front 17

Spurious relationship

Back 17

-When a correlation has been produced by an extraneous third factor and neither of the variables involved in the correlation has influenced the other
-Ex: Population size influenced # of strokes and # of births but # of storks does not influence # of babies
-Ex: Size of the fire accounts for both # of firefighters and amount of fire damage but does not mean firefighters caused the damage

*when an association or correlation between variables cannot be explained by an extraneous variable, the relationship is said to be "nonspurious"

Front 18

Hypothesis

Back 18

an expected but unconfirmed relationship between 2 or more variables
-should speculate about the nature and form of a relationship
-should indicate which variable predicts or causes the other and how changes in one variable are related to changes in the other

Front 19

Conditional Statement (if/then)

Back 19

Type of hypothesis; these statements say that if one phenomenon or condition holds, then another will also hold
-Ex: if members of a couple communicate about sexual topics frequently, they will use condoms more.

Front 20

Mathematical Statements

Back 20

Type of hypothesis; represent precise formal statements of hypotheses
-Ex: Y is a function of X
-social scientists seldom state hypotheses in this form

Front 21

Continuous Statements

Back 21

Type of hypothesis; increases in one variable are associated with increases (or decreases) in another variable
Ex: The greater the sexual communication, the greater the condom use

Front 22

Difference Statement

Back 22

Type of hypothesis; statements in this form assert that one variable differs in terms of the categories of another variable
Ex: Couples with high sexual communication use condoms more than couples with low sexual communication.

Front 23

Theory in public health research

Back 23

-Public health is applied and is frequently driven by health concerns
-We can draw on multiple theories for help in understanding a specific problem in a specific setting or context.
-Theories can help guide the development of conceptual models (ex. Selecting variables and specifying hypothesized relationships).

Front 24

Conceptual Models

Back 24

-A diagram of proposed causal linkages among a set of concepts believed to be related to a particular public health problem.
-Used in research development, implementations, and analysis.
-Included in grant application and sometimes research articles.

Front 25

Variables and relationships shown in conceptual models:

Back 25

-Independent/predictor variables
-Dependent variable
-Antecedent variables
-Mediating/intervening variables
-Modifying/moderating variables
-Confounding variables

Front 26

Things to think about (conceptual models)

Back 26

Control variables that are held constant by design are not shown in conceptual models
-It may be appropriate to indicate those control variables in the title (Ex. Conceptual model of factors influencing condom use among young women).
-Be sure you include concepts/constructs/variables-not values of variables-in the model.
-Remember the model should not be an attempt to show all variables affecting your variables of interest.

Front 27

Some important points about measurements

Back 27

-Moving from the abstract to the concrete
-Conceptualization to operationalization
-Manipulated and measured variables
-Operational definitions of measured variables
-Verbal or self-report
-Observation
-Archival records

Front 28

Levels of measurement:

Back 28

-Nominal- anything that has a name. No rank or order.

-Ordinal- Ranking, but the spaces between them have no meaning. (ex. Likert scale)

-Interval-Equal meaning between numbers

-Ratio-There is a true 0. Any kind of count. (Income, birth rate, weight, caloric
intake, # of condoms)

Front 29

Reliability

Back 29

Is it measuring your construct/variable consistently and dependably?
-concerned with questions of stability and consistency

Front 30

Validity

Back 30

-concerned with congruence or "goodness of fit"; accuracy
Is it measuring what it’s supposed to?

Front 31

Sources of Error in Measurement

Back 31

-Systematic measurement error
-Random measurement error

Measures can be reliable but not valid. If measures are unreliable they cannot be valid.

Front 32

Reliability Assessment

Back 32

-Test-retest reliability
-Split-half and internal consistency reliability
-Intercoder reliability

Front 33

Validity Assessment

Back 33

-Subjective validation
-Face validity-subjective feedback (ex. A focus group).
-Content validity
-Criterion-related validation
-Construct validation (We are most concerned with this)

Front 34

4 main types of construct validation:

Back 34

1. Correlations with related variables (positive/negative)
2. Looking at consistency across indicators of different measurements.
3. Looking at relationships between your measure and other variables you would not expect your measure to be correlated with groups you know should be different= discriminate validity
4. Difference between known groups

Front 35

Target Population

Back 35

Sampling; entire set of elements to which findings of the survey are to be extrapolated (ex. Homeless youth in Portland, Oregon households)
-Group that you want to learn about.

Front 36

Sampling unit

Back 36

Sampling; unit which we sample

Front 37

Frame

Back 37

Sampling; (ex. List), the population is divided into parts that are called sampling units. The sampling units must cover the entire population but not overlap each other. The frame is this list of sampling units. It is the mechanism to access and sample the population.

Front 38

Probability sampling

Back 38

every element in the population has a known, nonzero probability of being included in the sample.
-every element has a known chance of being selected in sample
-reliability of estimate can be evaluated

Front 39

Non-probability sampling

Back 39

sampling by nonrandom methods.
-Convenience Sampling (researcher simply selects a requisite number from cases that are conveniently available
-Purposive sampling (investigator relies on his or her expert judgment to select units that are "representative" or "typical" of the population)

Sampling that is based on a sampling pan that does NOT have the feature that every element in the population has a known, nonzero probability of being included in the sample. (Ex. Convenience sampling).
-faster and cheaper than probability sampling

Front 40

Simple Random Sampling (SRS)

Back 40

Sampling design; one selected by a process which gives every possible sample (of that size from that population) the same chance of selection. Most SRS is done without replacement).

Front 41

Stratification and Stratified Random Sampling

Back 41

Sampling design; the sampling frame is partitioned into groups or strata.
-Sampling is performed separately within each stratum

Front 42

Cluster Sampling

Back 42

A sampling design in which clusters are chosen by simple random sampling and, within each sample cluster (primary sampling unit; PSU), all observational units are selected.
-Sample selection is actually conducted with the clusters
-No subsampling within a cluster is done

Front 43

Systematic Sampling

Back 43

Sampling design; A method of probability sampling in which the sample is selected by applying an interval of constant length after a random start. The selection of the first unit determines the whole sample.

Front 44

Multi-Stage Sampling

Back 44

-Extension of cluster sampling where subsampling is implemented at various stages. In multi-stage sampling, the sampling units are of different types at different stages.
-Two Stage Sampling- a sampling design in which:
-Primary sampling units (PSUs) are chosen by some probability sampling scheme at the first stage.
-Secondary Sampling units (SSUs) are selected at the second stage by probability sampling within each PSU selected at the first stage.

Front 45

How to Take a Simple Random Sample

Back 45

1. Assign a number from 1 to N to teach observation unit in the population (ex. Sampling frame)
2. Pick a sample of n of these numbers by use of a random process, provided that the numbers selected are all different and less than N
3. If a number is selected again, discard it and go to the next number
4. The observation units corresponding to these numbers are taken as the sample.

Front 46

Advantages of Simple Random Sampling
and
Disadvantages of Simple Random Sampling

Back 46

Advantages of Simple Random Sampling:
-easy to implement
-easy to analyze data
Disadvantages of Simple Random Sampling:
-expensive
-not feasible in practice: it requires that all elements be identified and labeled prior to the sampling.

Front 47

Why would one use stratified sampling?

Back 47

1. we wish to obtain separate estimates for population parameters for each subgroup within overall population
a. ex. We wish to isolate segments of the population we want to oversample
2. we wish to ensure that our sample is representative of the population.
3. If stratification is done correctly, will give more precise estimate of the population parameters.

Front 48

Construction of strata

Back 48

Construct the strata so that they are homogeneous with respect to the variable under consideration.
-Create the strata by using a variable correlated with the variable of interest.

Front 49

Advantages of stratification over simple random sampling
and
Disadvantages of stratification over simple random sampling

Back 49

Advantages of stratification over simple random sampling:
1. Precision may be increased over simple random sampling
2. It is possible to obtain estimates for each of the strata that have been established.

Disadvantages of stratification over simple random sampling:
1. It may take more time (also money) to select the sample than would be the base for simple random sampling. Complete frames are necessary within each of the strata.
2. Increased complexity in data analysis.

Front 50

Allocation of sampling units across strata
Types of allocation

Back 50

-Equal (balanced): same sample size selected in each stratum.
-Proportionate to population composition: each case has equal opportunity of being selected and one can generalize directly from sample to population.
-Disproportionate: used to oversample subgroups.

Front 51

When is cluster sampling used?

Back 51

-The cost of obtaining information on every observation unit within a cluster is no higher, or only slightly higher, than the cost of obtaining a sample of all observation units.
-A sampling frame of observation units may be impossible to construct.

Front 52

Disadvantages of cluster sampling

Back 52

The sampling errors from this design are generally higher than those obtained from a simple random sample of the same number of observational units.

Front 53

How to take a simple cluster sample

Back 53

-List all clusters in the population
-Take a simple random sample of clusters
-Within each selected cluster, we include and take measurements on all sampling units.

Front 54

Advantages of systematic sampling:

Back 54

-Easy to apply and implement by statisticians and non-statisticians
-particularly advantageous when drawing is done in the field
-The systematic sample is spread evenly over the population (ex. Obtain spatially balanced sample as compared to simple random sampling)

Front 55

Variation of two-stage cluster sampling

Back 55

-Primary sampling units in the population have an equal or unequal number of secondary sampling units.
-The number of secondary sampling units selected from the primary sampling units can be similar or different across the primary units
*Used when a frame for all observational units is not possible to create due to expense or time constraints to develop frame.

Front 56

Sampling Error

Back 56

-Error that results from studying a subset of the population.

Front 57

Nonsampling errors: 3 sources

Back 57

Definition: other types of errors associated with surveys
3 sources
1. Frame or coverage error- attributed to some people not being included on the frame (or appearing multiple times). Therefore, these people do not have an opportunity to be measured.
2. Nonresponse error-missing data defined as either:
a. Unit nonresponse-the entire observation unit is missing (ex. The person cannot be located or refuses to complete the survey).
b. Item nonreponse- some measurements are present for the observation unit but at least one item is missing.
3. Measurement error- attributed to many sources such as the interviewer, instrument, or respondent.

Front 58

Types of Qualitative Data:

Back 58

-Interviews
-Observations (Participant and Non-participant)
-Written documents (Correspondence, reports, meeting minutes, newspapers)
-Case studies
-Historical records

Front 59

Design and Methods depend on

Back 59

-study topic and purpose
-characteristics of the population
-what data are already available
-resources