INT 301 Great Basin College Types of Data Discussion Questions

Library
< Back
Go to v
??
ili
Q
Chapter 6
Steps of Quantitative and
Qualitative Research Designs
The concept, “gravitation toward social media? is hard
to measure itself. You will need to use tangible measures
such as “time spent on social media each night.? Likewise,
the concept ?emotional effects? can be specified into
multiple questions gauging how happy the child is, how
social the child is, or how energetic and curious the child
is, and so on.
How Do You Select a Sample to Study
from Your Target Population?
Since the guiding principles and procedures for
quantitative and qualitative research are quite different,
the two kinds of research design call for somewhat
different knowledge and skills. In this chapter, we are
going to illustrate more detailed steps of quantitative and
qualitative research designs and some issues to consider
at each step. We will first discuss steps of quantitative
research designs. Qualitative research designs will be
discussed in the second half of the chapter. If you have
already determined that your study calls for a qualitative
research design and are only interested in qualitative
research, skip directly to the section on qualitative
research design.
Quantitative research designs include various methods
including surveys, experiments, and content analysis.
Since the most commonly used quantitative research
method is questionnaire survey, we will focus on the steps
involved in survey research and assume the discussion
will help those interested in other quantitative research
methods as well. Students frequently ask us these
questions when designing their survey research projects:
? What are my independent variable and dependent
variables?
variables are already implied by your research questions.
The independent variable in this example would be
relationships with parents and the dependent variable
would be school performance. Similarly, if your research
question is ?Are teenagers’ grades negatively affected by
gravitation toward social media?? then your independent
variable is “gravitation toward social media? and your
dependent variable is ?grades.? Since you are likely to have
more than one research questions in your study, you may
have multiple independent and dependent variables.
Sometimes, you may have several independent variables
and one dependent variable, and vice versa. For example,
questions such as ?Do regular medical check-ups, exercise,
and sufficient vegetable intakes reduce the likelihood
of cancer?? and ?Do cigarette bans in public buildings
and higher cigarette taxes encourage smokers to quit
smoking?? have multiple independent variables and a
single dependent variable. On the other hand, a research
question on the academic and emotional effects of bedtime
reading during early childhood assumes one independent
variable (“bedtime reading”) and multiple dependent
variables (the various “academic and emotional effects”).
It is a good practice to write down your research questions
and label your independent and dependent variables.
When you identify and label your independent and
dependent variables, you should be quite clear in your
mind that an independent variable is the cause of the
dependent variable and a dependent variable is the effect
of the independent variable. A dependent variable must
be able to vary or be affected when it is influenced by
the independent variable. In another example, if you
use education as an independent variable and salary as
a dependent variable, then, you are anticipating that
the salary of your respondents will change when their
level of education changes. If a variable cannot vary or
cannot be affected, then it cannot be used as a dependent
variable. For example, someone’s race and gender cannot
be changed by the influence of other variables; thus, they
cannot be used as dependent variables.
In the examples above, abstract concepts such as
?relationship with parents,” “gravitation toward social
media,” and “emotional effects,” need to be more specified
and operationalized into measurable indicators so
that you can quantify them. Operationalization is a step
where you identify very specific indicators or measures
for your concepts. For example, ?relationships with
parents? are not something you can directly observe,
but you can use some very specific indicators for a good
or a bad relationship. Quantifiable indicators, such
as ?number of arguments a teenager had with his/her
parents within a month,? ?number of times a teenager
received a punishment from parents within a month,”
and ?number of times a teenager violated rules set by
parents? are all good ways to measure whether a teen
has a good relationship with his/her parents. Or, you can
simply ask the teen respondents to rate the quality of
their relationship with parents on a scale of one to ten.
What group of people or cases is your research about? Do
your research questions concern the general population,
a particular group of people, countries, schools, or other
social organizations? The answer to these questions will
be your study population, or target population; the term
refers to the group of people or cases about whom you
will conduct your study and to whom you will apply the
findings of your study (Babbie 2013). Your population is
also the pool of cases from which you will select a sample,
or a subgroup of the cases you will actually study.
As you can imagine, if you select a sample that resembles
your population closely, you will be able to use your
findings to tell something about your study population.
But if your sample does not resemble your study
population, your ability to use your study findings to
predict the patterns in the study population is limited.
Suppose you have selected a group of students from
your university whose average grade is an A. You
know it is unlikely that this sample will reflect what
the average grade is in your university. The extent to
which your sample “looks like” your study population
is called “representativeness”; study findings from a
representative sample can be generalized to the study
population. For instance, if a group of spectators selected
by random drawing of numbers happens to have the same
demographic characteristics as the spectators in the entire
stadium, this sample will be representative of the crowd
in the stadium. This means that, if there is more rooting
for Team A in this sample, you can generalize that there
will be more support for Team A among the entire stadium
crowd.
to draw a particular type of sample, consult some of the
references listed in this chapter.
In the broadest sense, sampling methods fall within two
groups: probability and non-probably sampling. In the
above list, all the variety of random sampling and cluster
sampling fall in the probability sampling category. Quota
sampling, snowball sampling, purposive sampling and
availability sampling are non-probability sampling.
Probability sampling methods select participants based
only on random chance. Sampling theory considers this
as the best way to obtain a representative sample. To
use probability sampling methods, you need access to the
sampling frame, which refers to the roster of all units
in your study population (e.g., an approximate list of
citizens of a country, a list of residents of a community, a
list of all schools, organizations, student roster, and so on),
so that everyone is in the pool of available subjects and
only random chance can determine whether someone is
selected to be included in the sample.
Non-probability samples are used when researchers
do not have access to the sampling frame, or do not
have a clearly identifiable study population (such as
undocumented immigrants, homeless population).
Research done by students like you often has to resort
to non-probability sampling methods simply due to
insufficient time and resources. Non-probability sampling
methods are likely to introduce sampling biases because
factors other than random chance will affect the selection
process. It is okay to study a non-probability sample,
especially for a small scale exploratory study, or if you are
conducting qualitative research. Just keep in mind that
your findings will have limited generalizability, and the
limitation should be included in the discussion of your
findings.
When you select your study population, make sure that
you can gain access to them. If your study population
is minors (such as children or juveniles) or people with
limited power (such as prisoners), you may face particular
difficulties obtaining informed consents from guardians
or getting permission from the heads of the institutions to
enter the sites to collect data for your research. Therefore,
think carefully about access before you decide to study a
particular population.
After you decide on your study population, decide on what
your unit of analysis is. It may be individuals, universities,
organizations, or countries, depending on what is most
appropriate for your research.
you need a sufficient number of respondents in your
sample. As a general guideline, a minimum of 400
cases will be amenable for statistical data analysis. This
suggestion is to reduce sampling error due to sample
size. If you have a sample size of 400 cases, the standard
sampling error will always be 5% or smaller no matter
what the variation is in the study population (Babbie
2013). In reality, however, it may be unrealistic for a
student researcher to be able to draw a sufficient size
sample; you are more likely to work with much smaller
size sample due to the time and resource constraints.
In deciding your sample size, consult with your project
supervisor, or professor, as they may have specific
guidelines or requirements for sample size. Generally
speaking, three principles are useful in determining
sample size.
First, the larger the sample size, the smaller your standard
sampling error will be. At the same time, your sample
is more likely to resemble the characteristics of your
population and you will be able to generalize your findings
to the target population.
Second, if you are conducting a quantitative study, most
statistical analysis techniques used in social sciences
assume a normal, or the bell-curve distribution of data.
If the sample is too small, say less than 100 cases, there
is a good chance that you will not meet this assumption
of a normal distribution. Our advice is to obtain at least
a sample size of 100 respondents, if you plan to use
statistical analysis techniques. With that number you will
be able to use commonly used descriptive and inferential
statistical techniques such as cross-tabulation analyses,
chi-square tests, t-tests for comparison of means, and so
on. Keep in mind that you may receive invalid answers,
which you will exclude from your analysis; to obtain
a sample size of a 100 valid cases; you may need to go
slightly beyond your targeted sample size when you collect
the data.
? How do I select a sample to study from my target
population?
? What is an acceptable sample size for survey research?
? How do I turn my concepts into variables in survey
questionnaire?
? What are levels of measurement and why do they
matter?
The first half of this chapter responds to these questions
and relevant issues. In designing a survey research, the
following steps are usually necessary:
What Are Your Independent and
Dependent Variables?
How do we select a representative sample? Social
science methods teach us that we can approximate a
representative sample by reducing systemic selection
biases in sample drawing process. In general, a selection
method which only relies on random chance is considered
as having no systemic selection biases (Babbie 2013).
There are a variety of different ways to draw a sample from
the study population: simple random sampling, systematic
random sampling, stratified random sampling, cluster
sampling, quota sampling, snowball sampling, purposive
sampling, and availability sampling. Some of these
sampling techniques select participants using random
drawing while others do not. The specific steps and details
of different sampling strategies are beyond the scope of
this book. If you need to refresh your memory on how
The term “independent variable? is commonly used in
social sciences to refer to the cause, or the variable that
affects the other in a hypothesized relationship. The term
“dependent variable? refers to the effects or outcomes in
a hypothesized relationship. For example, let’s consider
the research question, ?How do relationships with
parents affect teenagers’ school performance?? Suppose
you expect that teenagers who do not have the typical
quarrelsome relationships with their parents will do
better in school than those who have a lot of conflicts
with their parents. The independent and dependent
Third, a small sample size may produce insignificant
statistical results simply as a function of the sample size.
Sometimes, small samples require you to use special
statistical measures other than the commonly used
measures mentioned above. According to probability
theory, when sample size decreases, the standard error
increases. For example, if you do a chi-square test with
a very small sample, you may find that many of the cells
in your cross-tabulation have fewer than five cases and
your chi-square value is not statistically significant. If
you have more than 25% of the cells with fewer than
five counts, your chi-square analysis is not acceptable
(George and Mallery 2000). If this is the case, you cannot
use chi-squire analysis to test whether two variables are
statistically independent of each other. On the other hand,
if you only intend to use simpler descriptive statistics
such as percentages and graphs to answer your research
questions, a sample size smaller than 100 can still work.
A ?robust” sample, or a sufficiently large and
What Is an Acceptable Sample Size for
Surveys?
Another issue to consider is sample size. Regardless of
whether you use probability sampling or non-probability
sampling, the size of the sample is an independent issue
which requires your attention. If you want to conduct
surveys and use computer software to do data analysis,
44% Page 105 of 267 ?
Location 3037 of 7225
Library
< Back
Go to v
??
i
Q
Chapter 9
Quantitative Data Analysis
answer your respondent may have checked off. The
number 1 stands for those who go to the university library
very often, 2 stands for those who go to the university
often, 5 stands for those who never visit, and 6 stands
for those who do not want to tell you how often they visit
or who simply do not want to answer the question. After
you have coded your surveys or structured interview
questionnaires, you may assign a case number to each of
your respondents and start entering the data into the SPSS
software.
?
then you may need to read more systematic textbooks
on statistics and computer data analysis. This chapter
does not attempt to give you a comprehensive training
in computer data analyses; instead, it focuses on guiding
student researchers, as they try to conduct quantitative
data analyses using actual research projects.
Students who are interested in learning systematic
data analysis may want to consult How to Use SPSS,
Eighth Edition: A Step-by-Step Guide to Analysis and
Interpretation by Brian Cronk (2014). Another textbook,
PASW Statistics 18 Guide to Data Analysis by Marija
J. Norusis (2010) may be helpful for students who need
more detailed instructions.
Quantitative data analyses are very useful in student
research. Although you might have been required to take
statistics, research methods, or computer data analysis
classes, doing quantitative data analysis in your own
research may still pose a great challenge. The following
are some frequently asked questions by students who are
expected to undertake computer data analysis in their
empirical research. Let’s take a look to see if any of these
questions are yours.
? I conducted my questionnaire survey; now, how do I do
data entry?
? Why do I have to know the levels of measurement in my
data analysis?
variable is gender; to enter this information to SPSS, you
can type in gender in the first cell in the second column.
For the second row, you may type in the second variable in
the survey, which is age. For the third row, you may type
in the third variable, which is university status or year of
study at a university. Since SPSS only allows you to enter
one word for your variable name, you will need to use
abbreviations or acronyms for your variable if one word is
not appropriate for your variable. As for the third variable,
you may use ustatus as your variable name instead of
university status. Then later on, you may use a more
descriptive variable label to inform you what the variable
is. You can continue until you type in all the names of your
variables in the second column.
The purpose of this survey is to understand your
experiences on campus at the university. There
is no right or wrong answer to the questions. We
would appreciate your giving us the most truthful
and accurate answers possible. This questionnaire
survey is anonymous and your name is not needed.
All data collected from this survey will remain strictly
confidential and only be used for computer data
analysis. Participation in this survey is voluntary,
but any information you can provide will be helpful
and appreciated. You may refuse to participate or
stop answering questions at any time. It should take
approximately 5 minutes to complete the survey.
PLEASE DO NOT WRITE YOUR NAME ON
THIS QUESTIONNAIRE
Starting SPSS Software
How Do You Start Entering Data From
Your Survey or Interview Questionnaire?
After you started the IBM SPSS Statistics (Version 22)
software, your computer screen looks like Figure 9.1. Two
windows are now available: a Data View and Variable
View window. Both are indicated at the lower left-hand
corner of the screen. Select either by clicking on the tab.
By default, the Data View window is activated and the
computer software is ready for data entry.
First of all, please tell us something about
yourself.
1. What is your gender? Are you o X Male 1
Female?
Eye Dructurn
? I learned different procedures of data analyses, but
which ones are most appropriate for my research?
? What do I do, if I just want basic descriptive analyses of
my data?
2. What is your age? 26.
3. What is your university status?
? Which procedure should I perform to determine
whether two variables are related to each other?
1.
Freshman
2.
Sophomore
? Which analysis procedure should I use to see how
several variables are related?
3. X
Junior
Data View
Variable View
4.
Senior
5.
Graduate
? Which analysis procedure should I use to compare
different groups of people?
? How do I use my data to explain the causal
relationships between my independent and dependent
variables?
Figure 9.1 SPSS Window. Source: IBM SPSS Statistics
Software (SPSS), version 22. Reproduced with permission
of International Business Machines Corporation.
6.
Coding
After completing your questionnaire survey or structured
questionnaire interviews, you are ready to use SPSS
(or other software) to do data entry and data analysis.
If you did not assign a number to each of the values in
your survey questions, you need to code your questions
first because the computer software mostly analyzes
numbers but not words. What we mean by coding is to
change respondents’ answers into numbers and enter the
numbers into the SPSS program so that you can analyze
them. For example, if you asked your respondents the
following question about gender without assigning a
number to the two possible answers:
What is your gender? male female
Then, you will need to assign a number to each of the two
options, for example:
What is your gender? O
female
In this way, the number o stands for males and the
number 1 stands for females. This is the task of coding. In
other words, if one respondent checked ?male,” you will
code it as ?o? for a male respondent. If the respondent
checked ?female,? you will use the number 1 for a female
respondent.
Here is an example of coding that has already been built
into the question. You may have asked how frequently
your respondents go to the university library:
How often do you go to the university library?
Other
4. What is your racial background?
1.
Caucasian
? How do I use several independent variables to explain
or predict a dependent variable?
male 1
2.
African
3.
Asian
The Data View window allows you to enter the number
codes for the answers the respondent selected and the
Variable View window allows you to define your variables,
including variable names, value labels, missing values,
data measurement levels, and other data specifications
such as decimals. There is a tutorial, if you need to
refresh your memory about how to use the software. The
helpful tutorial is usually available when you first start
the SPSS program on your computer. If it is not readily
available, you may click on Help and then click on
Tutorial to start the tutorial program.
4. X Hispanic
5.
Native
6.
Other
5. What is your current marital status?
1.
Married
Defining Your Variables
2.
Living together as an unmarried couple
? How do I interpret the data after my data analysis?
? What should I include in my paper when reporting the
findings of my data analysis?
This chapter answers these questions and shows you how
to start your data entry, select appropriate procedures
for your specific data analysis, and report findings in
your final report or thesis. Although there are different
kinds of computer software available for data analysis,
this chapter will show you how to use IBM SPSS Statistics
software version 22 for computer data analysis. The basics
in earlier versions are similar to this version. SPSS stands
for Statistical Package for the Social Sciences and it is
powerful and the most frequently used computer data
analysis software for social science research.
This chapter is written with the perspective that you have
studied statistics and have done some practice in data
analysis but may need to refresh your skills or need more
help with your specific research. If you have never learned
statistics or how to use SPSS to do computer data analysis,
3
Divorced
1. Very often
4.
Separated
2. Often
5. X
Never married
3. Sometimes
We recommend that you begin your data entry by setting
up your variables in the SPSS first. To do this, click
on the Variable View tab and switch to the variable
view window so that you can start to define and enter
information about your variables. In this Variable View
screen, each row represents all the information about one
variable and each column is for one kind of information
about all of the variables. For example, the first column
automatically provides you with a serial number for
each variable. The second column is to enter the variable
names. In the sample survey illustrated below, the first
6.
Widowed
4. Seldom
7.
Other
5. Never
6. What is your father’s education?
6. I do not know
In this case, you already have a number code for each
68% Page 173 of 267 . Location 4789 of 7225
Library
< Back
Go to v
??
i
Q
Chapter 10
Qualitative Data Analysis
meanings implied in the data. Regardless of the type of
qualitative data you work with, your analysis is based on
the principle of interpretation. Suppose your interviewee
stated:
become even more complicated. Do not worry, for almost
everyone runs into this situation. While staying focused is
important, it does not mean that you ignore when you find
something unexpected or something new in the process of
analysis. Remember that being able to immerse yourself
deeply into the data and find emerging stories in them are
the benefits of qualitative research.
?
Constructing a Theoretical Story with Your Data
If you have conducted a qualitative research, you should
have a set of non-numeric data such as texts, images,
observational notes, and voice-recording. Summarizing
and analyzing them can be challenging because there is no
one standard procedure that fits all types of data. When
you are ready to analyze and interpret your qualitative
data you might ask:
? What is the purpose of qualitative data analysis?
? Do I need to transcribe all my interviews?
I had a lot of difficulty juggling work and the family;
finding someone on a short notice when my regular
childcare arrangement falls through was a nightmare.
You will need to figure out the meaning of this statement
(i.e., interpret it). Does it mean that this person wants
to spend more time with the child but cannotbecause
of work? Why is it so difficult to balance work and
childcare? Is it due to particular circumstances, or is it
an experience common to most workers? Is childcare the
only challenge this person feels as a working parent? What
is her ?regular childcare arrangement? and why it would
not work sometimes? By considering these questions,
you are beginning to interpret the meaning of the data.
Obviously, thinking about contexts is an important part of
interpreting your data.
In the end, the goal of qualitative analysis is to tell the
story of your data. Regardless of which analytic route you
decide to take, you will report the prevailing patterns,
claims, and ideas about your topic. It is through the
process of constructing concepts and telling the story
of your data that you will find answers to your initial
research questions. Let’s consider some more questions
you might have when you analyze qualitative data.
Transcribing is straightforward. You will simply play the
recorded interviews and type them onto a word processing
program. Transcribing is a time-consuming work, taking
longer than the interviews themselves. You will probably
have to allocate about three to five hours to transcribe a
one-hour interview, for instance. Your university libraries
may have transcribing machines which you can borrow.
This will help you with the tedious task of listening to the
recorded interviews, stopping, and typing. The machine
allows you to use a convenient pedal to stop the recording
and restart as you transcribe.
Once fully transcribed, the voice data become text which
you can analyze. You may also have taken field notes
about the settings, and any non-verbal cues such as
gestures, smiles, laughs, and facial expressions during the
interviews and any group dynamics you observed during
focus group discussions. These written notes should also
be included in the transcribed data. If you conducted
interviews in languages other than the language in which
you will write your analysis, you may need to translate
the transcribed interviews. For instance, if you conducted
interviews with immigrants in their native languages
but wish to write the report in English, you will need
translated interview transcripts.
follow multiple stages of analysis to find first small units
of meaning and then gradually merge and group them into
a few broader themes. This is the case with the grounded
theory (Glaser and Strauss 1967), a popular technique
in sociology, anthropology, and related disciplines for
analyzing and theorizing text data. We will explain this
technique in greater detail below.
When you are working with images, you may treat them as
symbols or signs for certain meanings; examining closely
each piece of image data, you first assign a code or codes
based on your interpretation of the image and, in a later
stage, merge and group similar codes to construct broader
themes and categories of the meaning embedded in the
images. But for visual data, we frequently find studies
using pre-constructed coding schemes; in this case, codes
are predetermined and the researcher identify and count
the images in the data, which correspond to the coding
scheme. This is a common strategy in content analysis. For
example, if your research investigates racial stereotypes in
magazine advertisements, you may first construct, based
on previous literature, categories of stereotypes on which
you want to focus (e.g., Black athletes, White nuclear
families, Asian women in service roles, and so on). Then,
you can systematically examine the advertisements in your
data to identify the images which contain the different
stereotypes in your coding scheme. Content analyses
report the counts and percentages of each of the thematic
codes and interpret what those statistics tell us about the
research topic at hand.
? Where do I start?
? How do I conduct an inductive analysis?
Do You Need to Transcribe All Your
Interviews?
? What is the process of analysis when I use deductive
coding?
? What tools do I use to organize and summarize the
codes?
Coding or Identifying Themes
? How do I write about my findings from qualitative
data?
Where Do You Start?
Analyzing qualitative data frequently requires reducing
long texts, video footage, and complex images into shorter
and simpler labels that capture the idea. We call these
“codes.” Codes represent units of meaning. For example,
you may use the code “work-family balance? for the above
quotation. In other parts of the interviews, you may find
other codes and themes such as “career disadvantages,”
or ?unable to do everything,” “feeling torn,” and so on.
Coding is a process to identify small and large units of
meaning, which is to be done throughout the analysis
process.
What is the Purpose of Qualitative Data
Analysis?
Qualitative data analysis shares some similarities with
quantitative data analysis (Neuman 2011). Both methods
systematically summarize and compare data to obtain
theoretical ideas from empirical data. There are key
differences, however, in the purpose and procedure
of qualitative data analysis that distinguishes it from
quantitative data analysis. Unlike quantitative data
analysis which follows standardized procedures and
techniques, qualitative analysis uses a variety of creative
techniques that require open and flexible approaches.
While the purpose of quantitative data analysis is to test
already established theories, qualitative data analysis
is most often used to ?conceptualize and build a new
theory? (Neuman 2011: 509). For this reason, qualitative
data analysis is most often inductive or “bottom-up,”
starting from concrete data to extract more generalizable
theoretical ideas embedded in the data.
There is truly a wide range of techniques for qualitative
data analysis. Although we cannot cover all of them here,
we will describe some of the more popular techniques in
this chapter. But no matter which technique you use, there
are principles common to various qualitative data analysis
strategies which you can keep in mind before reviewing
the different