Transcript Quantitative Research

QUANTITATIVE RESEARCH
Hypothesizing, counting, and reporting
Quantitative Research



Numbers-based – Quantitative research refers to
the manipulation of numbers to make claims,
provide evidence, describe phenomena, determine
relationships, or determine causation.
Deductive – usually tests a hypothesis based on
previous research. Numbers are important to
determine when a hypothesis has been confirmed or
not. You are looking FOR something.
Generalizable – through statistical or mathematical
modeling, can make predictions about future events.
Quantitative research


Quantitative research often starts with an
expectation about what you are going to find and
then tests that expectation.
Follows a scientific method:
1.
2.
3.
4.
5.
6.
7.
8.
Define the question
Gather information and resources
Form hypothesis
Design experiment
Perform experiment and collect data
Analyze data
Interpret data and draw conclusions that serve as a starting point for
new hypotheses
Publish results
Research Plan

As we follow this scientific method, recognize that it
really is just a research plan, but in a more focused
manner. You would still benefit much from working
out the following BEFORE you conduct your study
 Research
 Method
 Plan
 Timeline
Question
Define the Question

Defining the question, often called your research
question, determines the scope of what you are
able to research. A good research question should
be (FINER):
 Feasible
– is it a realistic question to ask?
 Interesting – will we learn something from it?
 Novel – have very few people done it?
 Ethical – does it respect the participants?
 Relevant – will we be able to do something with the
findings?
Hulley S, Cummings S. (Eds ) Designing Clinical Research. Willimas & Wilkins: Baltimore, 1988
Defining the Question

To create a Quantitative Research Question
 Define
your participants
 Define your issue
 Define the variables of that issue
 Ask a question of the participants, issue, and variable


Do DU students have part-time jobs that they enjoy?
Are college major and character class in World of
Warcraft players correlated?
Gather Information and Resources

Text-Based Research is useful in helping you define your
expectation (hypothesis). You want to find what has come
before in the topic or related topics. You will rarely find your
exact study (if you do, then your research question isn’t novel).
You are looking for elements, pieces of your topic that have
come before.

Previous studies have determined that experienced female MMORPG
players have a primary motivation to play for social reasons (Yee,
2007). Other studies have shown no gender differences when looking at
more than a single primary motivation (Tychsen, Hitchens, & Brolund,
2008). In these studies, the intention was to look at motivation in
experienced RPG/MMORPG players—in my current study, I intend to
look at motivation to play MMORPG World of Warcraft by nonexperienced gamers.
Gather Information and Resources

Your experiences as well as those experiences of your
friends can also be useful in helping you gather
information and resources.

I have spent many hours in arenas with a rating of 2266 in
3v3. I have noticed in the past that my team tends to win
more before noon server time and after 10 PM server time.
Assuming Blizzard’s matchmaking for 3v3 arena matches
remains consistent, it seems that the level of player is better
during the afternoon and evening than it is at other times.
Form Hypothesis


What is your best guess as to the outcome of your Research
Question
The hypothesis is based on your Research Question, but it is not
phrased as a question – it is phrased as your best guess as to the
outcome of that question



DU students do not enjoy their part-time work
College major and character class in World of Warcraft are not related.
You might create sub-hypotheses to account for other variables that
you might consider relevant (e.g. gender, class-standing). You tag
these on to the end of your initial hypothesis.


Seniors tend to enjoy their part-time work more than first-year students enjoy their
work.
Female players tend to pick primarily caster classes whereas male students tend to
pick non-caster classes.
Design Experiment




Determine what best would address the research question
you are asking. In quantitative studies, a survey works the
best because you can control responses.
Determine triangulation questions or observations for two
reasons:
 You don’t want it to be obvious what you are asking
 Other variables may be affecting the outcome.
Refer to Chapter 8 in Situating Research (or the handout
Conducting Surveys) when coming up with your Survey
Questions
It’s a good idea to playtest your survey with one or more
people so that they can give you feedback about what
Design Experiment

Likert-type scale (1-5) will allow you to “quantify” human
beliefs, attitudes, and experiences. Likert-type scales are used
often in social science, descriptive studies.


Usually ask positive questions, and then follow with whether the person
agrees or disagrees.
Usually scaled so that higher numbers are positive/agreement, lower
numbers negative/disagreement
“A good writing class should consist of lectures on grammar”
1-strongly disagree, 2-disagree, 3-agree, 4-strongly agree
“How satisfied are you with University of Denver’s dorms”
1-very unsatisfied, 2-unsatisfied, 3-satisfied, 4-very satisfied
Design Experiment

Planning your survey, your hypothesis, your plan is
vital BEFORE you conduct your survey because you
only get one shot at the survey.
The Final Four




Perform Experiment – remember to be professional,
take notes (you never know what might effect your
results), and ethical.
Analyze Data – Keep track of your data, put it in a
spreadsheet, and work the numbers. We will be
talking a bit about statistics here, but really, all you
will be expected to do is descriptive analysis
Interpret Data and Draw Conclusions
Publish Results (see IMRAD PowerPoint)
Quantitative Research
Ideas and Controversy
Quantitative Research




Experimental – testing whether a “thing” (independent variable) applied to
a subject/group has an effect (dependant variable)

Two equal groups, one control, one experimental

Pre-test, post-test
Quasi-experimental – testing whether a “thing” applied to a subject/group
has an effect but without being able to
 Actively apply the “thing”
 Control for other variables
Descriptive – testing whether an effect is apparent or not
Inferential – testing whether two or more sets of already determined data
are related or not; testing whether two or more sets of data can provide
new data.
Quantitative Controversy

Some scholars believe that human experience, attitudes and beliefs cannot
be quantified.


Some scholars believe that quantitative research is trusted more than it
should be


Bridges still collapse, spaceships still get lost, new cars still break down,
pharmaceutical companies still produce harmful drugs, computers still crash
Some scholars believe that quantitative research is reductive


A person’s “feelings” can change from day to day with little conscious thought of
the fact
Statistically speaking, your SAT has already predicted what your final college
GPA is going to be. Will you only ever be 1400 or 3.6 smart?
Some scholars believe that quantitative research misses important nuances

In a ChangeWave survey, the iPhone had the highest customer satisfaction with
79% of the sample “very satisfied.”
Very Short Guide to Stats for SGR
Basics of aggregate and statistical data
Inferential v. Descriptive

Descriptive statistics “describe” the data of a sample
or population. They are usually aggregate data
Average (Mean) GPA
 Standard Deviation of SAT score


Inferential statistics “infer” (i.e. conclude) relationships
between a sample AND a population, or “infer” past,
present or future results of a sample/population
based on its data.

Regression/correlation analysis of GPA and SAT (relationship
between SAT and GPA, and SAT can be used to predict
GPA)
N = number of participations

In inferential statistics, you would refer to the number
of participants in your survey as N. If it is a sample
or part of a whole, it is n (lowercase), and if it is a
total population, it is N (uppercase).
Population: N = 4,432
 Sample: n = 100


In descriptive studies and descriptive statistics, it is
common to refer to participants as N, subgroups of
those participants as n

Of the total students surveyed (N = 100), only 10% (n =
10) were male.
Measures of central tendency

Central Tendency measures common “middles”






Mean is the arithmetic average of items or values
Mode is the most occurring item or value
Median is the item or value of which 50% are greater and 50% are less.
Sometimes GPA or time can be used as a measure, but another measure is one
of attitudes and beliefs using a Likert-type scale.
Standard Deviation is a measure of the spread of items or values in a series.
Understanding the variation can help you see how close a particular item or
value is to other numbers.
Distribution (Histogram) is a visual representation of the number of a particular
result in an array of numbers.
In this series (number of hours I played WoW over break):
8, 0, 0, 3, 2, 10, 0

Mean = 3.29, Mode = 0, Median = 2, SD = 4.11
In this series (number of hours I worked this week):
8, 8, 8, 8, 6, 6, 5

Mean = 7, Mode = 8, Median = 8, SD = 1.29
Using Excel to do your stats

Mean { =average(range) }



Standard Deviation { =stdev(range) }
You can also count the number of instances of a value
including instances of text: { =countif(range,”value”) }


You can compute mode { =mode(range) } or median
{=median(range) }, but they might not be as useful in this project.
The following example would count every instance of “male” in
the range:
=countif(A2:A7,”male”)
You can create frequency distribution histograms by using
Tools -> Data Analysis, then Historgram. Histograms count the
number of instances of a result in a given array.
You can also find these commands by using Insert -> Function. There are also
far more complex inferential statistics available in Excel

You can do a complete Descriptive Stats Summary by selecting Tools > Data Analysis (If
you don’t see a Data Analysis, then (Excel 2003) Tools > Add-ins > Analysis ToolPak;
(Excel 2007) Excel Options > Add-ins > Manage Add-ins > Analysis ToolPak
Writing Stats in APA



Standard Deviation = SD
Mean = M
Descriptive statistics are often written in parentheses after an
item that the statistic refers to, and symbols and numbers should
be separated by a space

In a survey of DU students, participants (N = 100) responded that money
was more important (M = 4.2, SD = .9) than experience (M = 3.5, SD =
.76) in selecting a summer job.

In a survey of computer game addicts, females (n = 15) were more likely
to be depressed during withdrawal (M = 5.2, SD = .45) than males were
(n = 78, M = 3.2, SD = .98)

Chapter 8 in Situating Research has more about this.
Charts and Graphs
It’s important in doing graphs that you compute an aggregate (sum,
average, SD, something) before graphing information. You cannot just
Select All of data and make a graph out of it.




Pie graphs – good for showing distributions of a total population
(you will have to compute aggregates first)
Line graphs – good for showing time-based, linear progression
Column/Bar graphs – good for showing distribution of individual
responses (you will have to create aggregates first)
Y-Axis (vertical) for variables, X-Axis (horizontal) for participants.