Post on 06-May-2018
R. S. Balkin, 2008 3
Theory
Explanation of an observed phenomena
In the social sciences, it is derived fromtheoretical constructs
Constructs are operationally defined
R. S. Balkin, 2008 5
Limitations toGeneralizability/Transfer Bias Phenomenological Inconsistent realities
Objective reality vs constructed reality
R. S. Balkin, 2008 6
Postmodernism
There is no superior form of inquiry There is no universal truth Rather, there is an agreement among
many voices
R. S. Balkin, 2008 7
Response to postmodernism
Scientific researchers have developedconcepts and procedures to conduct highquality investigations
Findings can be replicated Findings can be refuted Bias and errors can be minimized A good researcher is guided by ethical
principles
R. S. Balkin, 2008 10
Description
We wish to know the form, structure, activity,change over time, and/or relationship toother phenomena
R. S. Balkin, 2008 13
How do we use research to answerquestions? Define what we want to measure:
Purpose of research Type of Research Variables
Develop questions or hypotheses Identify how data will be collected
Population Sampling
R. S. Balkin, 2008 14
Define what we want to measure:Purpose Fundamental research
Deductive method less concerned with applicationmore common in physical sciencesThe benefits of the research may not be
applied immediately
R. S. Balkin, 2008 15
Define what we want to measure:Purpose Applied research
Purpose is to improve a product or processDescribes most social science researchAttempts to develop generalizations about a
population
R. S. Balkin, 2008 16
Define what we want to measure:Purpose Action research
The researcher is actively involved in the process Focus is on immediate application, not generalization Purpose is to improve practice Results may not be generalizable because they often
do not occur in an experimental setting
R. S. Balkin, 2008 17
Define what we want to measure:Purpose Assessment
Fact-finding—examines what existsNo hypothesis testingNot considered research
R. S. Balkin, 2008 18
Define what we want to measure:Purpose Evaluation
Application of findingsExamination of utilization and effectivenessNot considered researchMay be summative—occurs at the conclusion
of a specific period (i.e. assigning a grade)May be formative—evaluation is ongoing and
continuous
R. S. Balkin, 2008 19
Define what we want to measure:Two broad areas Qualitative research
Inductive Identifies the details of a phenomena Develops theory
Quantitative Research Deductive Generalizes to a population Tests theory
R. S. Balkin, 2008 20
Define what we want to measure:Type Descriptive research
May be quantitative: The gathering and analysis of empirical data in
order to accurately describe a population Focus is on descriptive statistics Involves hypothesis testing, analysis, and
generalization May be qualitative:
Focuses on the nature of a phenomena Develops a theory of a phenomena
R. S. Balkin, 2008 21
Define what we want to measure:Type Experimental research
Attempts to identify group differences or relationshipsbetween or among variables
Always involves manipulation of variables andrandom assignment
Quasi-experimental research Combines descriptive and experimental research There is no random assignment
Either may be qualitative or quantitative
R. S. Balkin, 2008 22
Descriptive vs. ExperimentalResearch Both involve hypothesis testing Both employ random sampling (more on
that later) Experimental research uses random
assignment (more on that later too)
R. S. Balkin, 2008 23
Descriptive vs. ExperimentalResearch Descriptive research does not employ random
assignment—that is, some phenomena cannotbe randomly assign, such as sex.
The researcher may be interested in therelationship between two variables that arealready present—this type of descriptiveresearch is known as a correlational research For example, a researcher wishes to study the
relationship between depression and disorderedeating.
R. S. Balkin, 2008 24
Define what we want to measure:Type Historical research
Describes what was by investigating,recording, analyzing, and interpreting events
Purpose is to identify previous trends of thepast to understand what needs to happen inthe present and future
May be quantitative or qualitative
R. S. Balkin, 2008 25
Define what we want to measure:Variables Variables are defined through observation:
i.e. demographic variables (e.g. SES,ethnicity, sex)
OR Variables are defined operationally
Constructs (i.e. mood, creativity, wellness)Based on theory
R. S. Balkin, 2008 26
Develop questions or hypotheses:Questions When developing a research project, it is
sometimes helpful to think in terms of whatquestion(s) may be answered through thisresearch.
R. S. Balkin, 2008 27
Develop questions or hypotheses:Questions Research questions may be descriptive
They describe what is occurringWhat percent of low achieving students come
from single parent families?
R. S. Balkin, 2008 28
Develop questions or hypotheses:Questions Research questions may be relational
The relationship between two variables isexamined
What is the relationship between exercise anddecreases in depression?
R. S. Balkin, 2008 29
Develop questions or hypotheses:Questions Research questions may be causal
Is one phenomena caused by anotherphenomena?
Does drinking alcohol cause an individual toescalate in drug use?
Causal studies are rarely conducted in socialscience research because determiningcausation is very difficult.
R. S. Balkin, 2008 30
Develop questions or hypotheses:Questions
Conditions for causation:Relationship between the two phenomena
Time order—the proper sequence of events
Influence of confounding variables has beeneliminated—no other influences can be attributedto the relationship
R. S. Balkin, 2008 31
Develop questions or hypotheses:Hypotheses An hypothesis is a specific statement of prediction.
There are two types:
1. Scientific Statement about what we think should happen in
our study Statement about the variables that deter the study
2. Statistical contains a null and an alternative hypothesis You accept one Lends credibility, provides evidence DOES NOT PROVE
R. S. Balkin, 2008 32
Statistical Hypotheses
Actually, a statistical hypothesis includes twohypotheses.
Let's say that you predict that there will be differencesbetween males and females in a study of emotionalmaturity in high school students.
The way we would formally set up the hypothesis test isto formulate two hypothesis statements: One that describes no differences between the groups One that describes differences between the groups.
R. S. Balkin, 2008 33
Statistical Hypotheses
Usually, we call the hypothesis thatdescribes no differences as the nullhypothesis.
We call the hypothesis that you supportsdifferences the alternative hypothesis
We use a notation like H1 to represent thealternative hypothesis and H0 torepresent the null case.
R. S. Balkin, 2008 34
Statistical Hypotheses Null Hypothesis
Alternative (research) hypothesis
Ho: µ1 = µ2H1: µ1 ≠ µ2
“OK, I know it's a convoluted, awkward and formalistic way to askresearch questions. But it encompasses a long tradition in statisticscalled the hypothetical-deductive model, and sometimes we justhave to do things because they're traditions. And anyway, if all ofthis hypothesis testing was easy enough so anybody couldunderstand it, how do you think statisticians would stay employed?”(Trochim, 2006)