Research and Statistics Flashcards
What’re the steps of the research process?
-Problem Formulation
-Methodology
-Actual Collection of Data
-Analysis of Results
-Dissemination of Results
The Research Process: Problem Formulaiton
process used by researchers to develop a precise statement that can be operationalized
Problem Formulation: Questions of Hypotheses for Study
development of conceptual frameworks and operational concepts
Study Design: Design Concerns
insure that the data collected is the same for all participants depends on the training of interviewers and management and control of data
Study Design: Research Resources
subjects, the availability of existing data, and the quality of the professional researchers
funding is an important issue and often determines the compromises that researchers make in designing studies
involvement of human subjects: ethical questions
time available for research
Study Design: Ethical Concerns
-Research cannot lead to harming clients
-Denial of an intervention may constitute harm
-Informed consent
-Confidentiality
The Research Process: Methodology
includes select measurement techniques to be used, the setting where the research is to be conducted, and the population or group to be studied
The Research Process: Actual Collection of Data
data security to insure confidentiality
The Research Process: Analysis of Results
were the research questions answered? Were the hypothesis accepted or rejected? Research does not simply test a research hypothesis, but implicates the whole body of logically connected knowledge
analysis of results may include changes in related aspects of social work practice and policy
The Research Process: Dissemination of Restuls
the cumulative nature of science requires reporting results, conclusions, and interpretations in writing with a view toward publication in professional, preferably refereed journals directed to peers
Research Terminology: Concepts
abstractions created by generalizing from specifics
Research Terminology: Operationalizing a Concept
reducing a conceptual variable to a set of directions and actions so that a study can proceed in a systematic and replicable manner
Research Terminology: Hypothesis
statement about a relationship between two or more variables that can be tested with an outcome that one may confirm, fail to confirm, or refute
What does a hypothesis do?
-Asserts there is a relationship between variables
-Specifics the nature of that relationship in a way that allows testing
-Should include empirical referents
-Avoids value judgments
-Is related to a body of theory
Research Terminology: Null Hypothesis
asserts there is no significant relationship between two variables
Research Terminology: Variable
a characteristic possessed by everyone in the population in varying amounts or kinds
opposite of a constant which refers to a characteristic which does not vary with different people
Variable: Independent Variable
the variable believed to cause some variation in another variable
Variable: Dependent Variable
the variable whose variation must be explained
Variable: Intervening and Extraneous Variable
a variable which comes between the independent and dependent variables
modifies or confounds the variation dependent variable which was thought to result from the effects of the independent variable
Research Terminology: Theory
a set of related hypotheses connected so as to explain some phenomena or that predicts phenomena
Research Design: Exploratory Studies
used to explore an area of knowledge where little is known and to gain familiarity with real life settings, problems, or phenomena
What’re exploratory studies used for?
-clarify concepts and to develop hypotheses for subsequent research
-while used to explore new areas, they are built on careful assessment of knowledge that already exists
-may include systematic review of related literature, a survey of expert, analysis of case material, and participant observation
Research Design: Descriptive Studies
used where there is more knowledge than in exploratory studies
What do description studies do?
-concerned with ascertaining facts and carefully designed studies of phenomena
-attempt to develop an accurate qualitative and/or quantitative summary or assessment of the situation or phenomena
-they involve studies of a small representative sample so that interferences can be made about the broader population from which the sample is drawn
-describe characteristics of the population or the relationship among given variables
Research Design: Experimental Studies
the most rigorous of all studies
Research Design: Experimental Studies
-involves testing a prediction by manipulating and independent variable and measuring the effect on a dependent variable
Relationship between the independent and dependent variables?
-done under controlled conditions
-research methodology is designed to eliminate the effect of other extraneous variables, thus eliminating alternative explanations for any relationship that is observed
How do you determine causality?
-Concomitant Variation
-Time Factor
-Alternating Opening Variables
-Threats to Internal and External Validity
Experimental Studies: Concomitant Variation
manipulation of the independent variable is associated with changes in the dependent variable
Experimental Studies: Time Factor
the independent variable occurs before change in the dependent variable
Experimental Studies: Alternate Operating Variables
refer to observed changes caused by extraneous or unknown variables rather than the hypothesized independent variable since these may
Experimental: Threats to Internal and External Validity
alternate operating variables which may influence the results of a study in unknown ways
Threats to Internal and External Validity: History
Maturation: effects which are systematic with the passage of time
other changes resulting from the passage of time that may be more powerful than the intervention
Threats to Internal and External Validity: Measurement Decay
fatigue of judges or observers
bias in the selection process
Threats to Internal and External Validity: Mortality
the loss of some subjects from a sample
Threats to Internal and External Validity: Hawthorne or Test-Taking Effect
the act of being studied may in itself produce some change in the subjects
Threats to Internal and External Validity: Placebo
treatment given to a control group to convince them they are being exposed to the experimental variable, but exposure is only to a neutral stimulus
Types of Experimental Studies and Design: Field Experiment
conducted in a concrete, natural environment
tests a hypothesis but does not have tight control over subjects’ exposure to the experimental variable
Types of Experimental Studies and Design: Laboratory Experiments
the tests are conducted under tightly controlled laboratory conditions
Types of Experimental Studies and Design: Classical Before-After Experimental and Control Group Design with Randomization
randomly assign all subjects to experimental and control groups so the researcher is able to assume the two groups are comparable at the start of the study
Types of Experimental Studies and Design: Ex-Post Facto Analysis
use of statistical analysis of the data to control for the effect of a given variable
Measurement Problems: Issues with Measurement
defining true differences and distinguishing these from variations that result from measurement errors
some measurement errors are constant
other errors in measurement are random
Measurement Problems: Reliability
refers to consistency in the measurement of a variable
Tests of Reliability: Test-Retest
provides evidence of stable scores
administered once, then administered again to the same subjects under the same conditions
the higher the correlation between the initial test and the retest scores, the higher the reliability
Test of Reliability: Split-Half
each item in a test is randomly assigned a place in the test
the test is split into two halves, each half considered comparable to the other
Test of Reliability: Alternate Forms
comparable measures of the same variable are administered to the same subjects at about the same time
the degree of correlation between the two scores is a measure of reliability
Test of Reliability: Inter-Judge Agreement
two or more judges are trained to observe and score the same phenomenon in the same way
the judges’ independent measures of the same observed phenomenon are correlated
high correlation indicates high reliability
What’re the methods increasing reliability?
-standardizing administration of the measurement instrument
-adding additional items (test questions) to cancel out random error
-statistically identifying and eliminating items which do not agree with other
Tests of Validity: Concurrent Validity
the measure used in a study is compared with another instrument presumed to measure the same variable
high correlation of results indicates concurrent validity of the instrument
Tests of Validity: Predictive Validity
a measure in a study is compared with some predicted future outcome
Tests of Validity: Content Validity
analysis of an instrument by a group of persons considered expert in the field studied
if the experts judge the test to be a good measurement of what is being studied, the test has content validity
Tests of Validity: Construct Validity
the degree to which a measure related to other variables expected within a system of theoretical relationships
Measures Problem: Scales of Measurement
in addition to concern with the measurement problems of reliability and validity, research is also concerned with choosing an appropriate measurement scale
Four Categories of Measurement: Nonminal
two or more name categories
categories show differences but not the degree of difference
categories have no additive value
measurement involves simply placing subjects in categories and counting the number in each, there are few available statistical procedures
these measures are not highly sensitive in determining differences
Four Categories of Measurement: Ordinal
the scores show each subject’s position with respect to a particular characteristic
a number of statistics are used with ordinal measurements; these are called non-parametric
Four Categories of Measurement: Interval
higher-level scale shows ordinal positions with equal intervals between scores
not only ranks data but also uses categories of equal size
Four Categories of Measurement: Ratio
an interval scale with an absolute zero
a scale where a score of 80 is twice as high as a score of 40 uses the same statistics as interval scales
Sampling: Sample
part of a large population
the quality of the sample determines the inferences that can be made about the large population from which its drawn
Sampling: Population
has specific variables in common
Sampling: Census
a complete count of all relevant members within the sample and population
Sampling: Statistic
a numerical expression summarizing a sample characteristic of the population from which it is drawn
Types of Samples: Probability Sample
sample that allows the researcher to specify that each element of the population has a known probability of inclusion
it assures representatives, allows statistical inferences about the population from which it is drawn, and is more precise
Probability Sample: Simple Random Sample
subjects from a known population are drawn randomly for the sample
each subject has equal probability of inclusion in the sample
Probability Sample: Stratified Random Sample
subjects are first grouped into strata of interest to the researcher and then drawn randomly from each group
Probability Sample: Cluster Random Sample
multiple-stage sampling in which successive random samples are drawn from natural groups (clusters) when acquiring a list of all elements of a universe is not possible or too expensive
Sampling: Non-Probability Sampling
sample in which it is not possible to specify the probability that each element of the population has an equal chance of being included in the sample
Non-Probability Sample: Accidental Sample
take the first case encountered
Non-Probability Sampling: Quota Sample
include some cases from each segment of a population
Non-Probability Sampling: Purposive Sample
purposely draws a sample from a part of the population assumed to have particular knowledge of what is being studied
Statistics: Descriptive Statistics
computations that describe some characteristics of a group or sample
they enable users of research to summarize information about the group and to make useful comparisons among two or more groups
Statistics: Inferential Statistics
allow researchers to generalize from a sample to some large population of which the sample is a part
inferential statistics are based on probability theory
Descriptive Statistics: Measures of Central Tendency
measures the degree to which findings cluster together
Measures of Central Tendency: Purpose
an average is used to represent all scores in a group, thus giving a concise description of the performance of the group as a whole
allows within group and between group comparisons
Measures of Central Tendency: Mean
sum of the scores divided by the number of scores; an average
Measures of Central Tendency: Median
the midpoint of a series of hierarchically ordered scores
half the scores fall above and half fall below the midpoint
Measures of Central Tendency: Mode
the score or measure occurring most frequently
When is mean used?
with a symmetrical distribution, when the most stable measure is wanted and when other statistics are to be computed
When is median used?
when the exact midpoint of a distribution of scores is wanted, when there are extreme scores
When is mode used?
when a quick and approximate measure is all that is wanted, when the typical value is wanted
Measure of Variability: Purpose
to determine dispersion or spread of scores around the central tendency
Measures of Variability: Measures of Dispersion
spread of scores around the average
Measures of Dispersion: Ratio
the lowest to the highest score
Measures of Dispersion: Standard Deviation
the most stable measure of variability, it uses squared deviations from the mean degree to which scores are spread or dispersed around the mean
Measures of Association: Correlation Coefficient
a numerical index of the degree to which two variables measured in interval units are associated with each other
the symbol “r” represents the correlation coefficient
any change in one variable of the pair causes a direct an proportional change in other variable of the pair
Correlation Coefficient: Magnitude of Correlation
the closer an “r” value gets to 1, the stronger the association r=.80 is stronger than r=.30
Correlation Coefficient: Direction of the Association
this is indicated by the sign (+) or )-) in front of the “r” coefficient
What does (-) mean?
indicates a negative or inverse relationships
What does (+) mean?
indicates a positive or direct association
Statistical Tests of Significance: Null Hypothesis
the hypothesis that there is “no difference” that is actually subjected to staistical testing
Statistical Test of Significance: Type I Error
the chance of rejecting the null hypothesis when it should be accepted
Statistical Test of Significance: Type II Error
the chance of accepting a null hypothesis when it should be rejected
Statistical Test of Significance: Statistical Significance
an observation or result of an outcome measurement is statistically significant if it falls within the region of “unlikely events” or “unlikely outcomes”
Inferential Statistics Definition
used to generalize from a sample to some large population from which the sample is drawn
