APPENDICES

A
Appendix Answers to the Chapter Study
Questions 605

B
Appendix Determine Size Using Sample Size
Tables 609

C
Appendix Commonly Used Statistics in
Educational Research 613

D
Appendix Nonnormal Distribution 614

E
Appendix Strategies for Defending
a Research Proposal 615

603
This page intentionally left blank
Answers to the Chapter
Study Questions
A
A P P E N D I X

Chapter 1 Chapter 2
1. Three advantages you might identify are that research 1. The parent involvement study (Deslandes & Bertrand,
contributes to knowledge, that it helps with practice, 2005) narrative hook is fair. It is a reference to another
and that it provides information to policy makers. Three article in which the author makes the case for the
disadvantages might be that results are vague, are con- importance of parent involvement. It points to the
tradictory, and may be based on inaccurate data. many studies that have documented parent involve-
2. If I were Maria, I would first clearly identify the issue ment as a problem worth studying. It is, in short, a
that needed to be studied by visiting with a member literature-based hook, but it would be of more interest
or two of the committee. Second, I would go to the to academics and other researchers than to educa-
university library and begin to locate studies on the tional personnel in the schools or to parents. It lacks
issue. Third, I would write down the central question emotion and numbers to back up the stance. The
that I would like answered. mothers’ trust in school principals (Shelden et al.,
3. If the parent involvement study was qualitative instead 2010) is better. It addresses a topic that many readers
of quantitative, the authors would have sought the can relate to. However, it does not use numeric infor-
views of the parents on the question of their involve- mation to indicate how pervasive the issue of parent
ment with their adolescents in school and at home. participation might be. It also does not cause a par-
They may have asked the parents open-ended ques- ticular emotional or attitudinal response in the reader.
tions and explored their different perspectives. 2. The problem in the parent involvement study discusses
Instead, the authors chose to test whether four factors the importance of learning how the four constructs
would help explain parental involvement. If the moth- influence parent involvement. Explaining the relation-
ers’ trust in school principals study was quantitative ship among variables is a key element in quantitative
rather than qualitative, the authors would have set up research. In the mothers’ trust study, however, the
an experiment to test whether one group of mothers authors seek to develop an understanding of mothers’
(e.g., those who were more vocal in their concerns attitudes toward principals. Explanation is a central
about school) differed from another group of mothers characteristic of quantitative research, whereas under-
(e.g., those who voiced few concerns about the standing is a characteristic of qualitative research.
school) in terms of their responses to the principals. 3. The parent involvement study contributes to knowl-
The authors might also have surveyed mothers to edge by helping to fill a void in the existing literature.
determine their trends in views toward principals. But, The mothers’ trust in school principals study contrib-
as a qualitative study, the authors were interested in utes to knowledge by studying a problem that also
exploring the mothers’ attitudes toward principals and fills a void (i.e., we do not know the elements of
identifying themes in their responses. mothers’ trust and need a study to learn about it).

605
606 APPENDIX A Answers to the Chapter Study Questions

4. In the parent involvement study, a justification based 2a. Central phenomenon = conceptions of reading.
on personal experience would require that the 2b. Central phenomenon = aspects of sorority life.
researchers talk about their personal experiences with 2c. Central phenomenon = academic change (sources and
adolescents, with parenting, and with the schools of processes).
their children (if they have them). In the mothers’ trust 3a. Directional hypothesis
in school principals study, the authors might have 3b. Null hypothesis
talked about whether they were mothers, and whether
they had experienced issues of trust with the schools
or with school officials. Chapter 5
5. If we compare the introductions to the two studies, the
1. Cronbach’s alpha is a measure of reliability and, more
authors of both studies cite extensive literature to doc-
specifically, internal consistency. A coefficient of .93 is
ument the importance of the problem of study. In the
a high coefficient; .6 is an acceptable level for deter-
qualitative mothers’ trust study, this might appear to be
mining whether the scale has internal consistency. With
an excessive use of the literature up front in a study
a .72 reliability coefficient, the reliability is satisfactory
that would tend to close down the discussion rather
for the scores.
than open it up. However, the literature review at the
2. Permissions should have been obtained from (a) the
beginning of this qualitative study does a nice job of
administrator in the school district responsible for per-
narrowing the topic down from the broad issue of par-
mitting researchers to enter schools and conduct
ent involvement in schools to the issue of mothers and
research, (b) possibly the building principal, (c) the
their trust of principals.
teachers in the classrooms of the parents’ students,
6. Answers will vary.
and (d) the campus institutional review board at the
authors’ institution.
Chapter 3 3. Steps in locating this instrument: (a) Run an ERIC
1. The key terms would be teens, high schools, and preg- search and examine the references netted in this
nancies. I would need to see if these terms are those search. (b) Next look in the Buros Institute indices,
used in the ERIC system or use the Thesaurus to see TIP, or MYY, to see if there are any instruments that
what terms would be appropriate to use when running measure this variable. (c) Look for published research
my ERIC search. journal articles that contain this variable. See if the
2. I would first examine recent articles, say, those articles authors cite an instrument that they may have used to
published in the last 2 to 3 years. Because they are all measure the variable. (d) Ask your advisor or a spe-
journal articles, they represent a good place to start. cialist in the field who may have studied this variable.
I may not be familiar with the quality or stature of the Find out if this person knows of an instrument that
journals, but they all seem like national, review-based you might use.
journals. This leaves me with scanning the titles. Some 4. Maria needs to make these decisions: Whom will she
titles seem to convey content broader than others. study? She needs to decide whether she will study stu-
I would begin with the broader articles first (e.g., dents in high schools, teachers, administrators, or
“Violence-related attitudes and behaviors of high school other personnel. She also needs to determine how she
students” rather than “Response of African American will select these individuals. A random sampling pro-
adolescents . . .”). cedure would be ideal. Also, she needs to identify
3. The reference might look might look like this: how many individuals she needs for her sample using
the rough guide in the section on “Sample Size.” Next,
Kingery, P. M., Coggeshall, M. B., & Alford, A. A.
based on the questions she plans to answer, she needs
(1998). Violence at school: Recent evidence from
to determine what kinds of data she will collect. For
four national surveys. Psychology in the Schools,
example, she may collect attitudinal data or public
35(3), 247–258.
school document data. With this decision, she will
next look for instruments to use in collecting this
Chapter 4 information and perhaps conduct a search of the lit-
1a. Independent variable = parental involvement; depen- erature for existing instruments. These instruments
dent variable = student reading performance. need to be of high quality, reliable, and valid. The
1b. Independent variable = students’ knowledge of procedure for data collection needs to fit the individu-
accountability; dependent variable = changes in als who will provide the data, and the scales need to
attitudes toward learning. conform to standard types of interval, ordinal, or nom-
1c. Independent variables = mutuality, comprehensive- inal scales. Finally, with an instrument identified, she
ness, gender sensitivity, congruence; dependent will collect data using standardized procedures for all
variable = functional mentoring relationships. participants and ethical practices.
 APPENDIX A Answers to the Chapter Study Questions 607

Chapter 6 tional information through having the mothers journal

1. Is the hypothesis a comparison of groups or the rela- about their experiences, or collecting notes written to
tionship of variables? It is a comparison of three groups. the school by mothers about their children. Additional
insight might have been gained from visiting (through
How many independent variables do you have? interviews) with other members of the families, such
Dependent variables? One independent variable as fathers or siblings of the children.
with three groups; one dependent variable, math
scores.
Are there any covariates being controlled? No. Chapter 8
Is the independent variable categorical or continuous? 1. Some limitations in this study would be that only
Categorical. mothers were interviewed, and that the researchers did
Is the dependent variable categorical or continuous? not have information about how different school con-
Continuous. texts or settings might have influenced the mothers’
Statistical test of choice: Analysis of variance (ANOVA). views toward principals. Also, children with disabilities
vary considerably in their needs, and these were not
2. You fix the responses that are out of range (e.g.,
factored into the study. Future research might involve
6 and 8).
entire families, might develop specific types of model-
3. In this statement:
ing mothers needed by the principals, and the specific
What does F stand for? ANOVA, the statistical test.
steps in coaching for teachers and administrators that
What is (1, 110)? The degrees of freedom.
would encourage the mothers’ participation.
What is 6.81? The calculated value using the ANOVA
2. Responses will vary to this question. Look for multiple
formula.
code words in the left margin and two or three themes
What does p = .001 mean? The p value or the signifi-
in the right margin of the transcript.
cance level of the calculated value; the groups differ
3. Responses will vary to this question. Students could
statistically at the alpha level of .001.
provide feedback on their experiences in using the
software tutorial.
4. Answers will vary for this description. Look for
Chapter 7 answers related to context, broad-to-narrow descrip-
1. The purposeful sampling snowballing strategy involved tion, use of action verbs, vivid modifiers, and quotes.
three approaches: to distribute district-level material 5. Answers will vary for this description. Look for the
through administrators, to hand out material through name of the theme, evidence for multiple perspectives
school personnel, and to have mothers refer other on the theme, subthemes, and quotes.
mothers. The approach of distributing materials through
administrators resulted in few responses because it did
not protect the identities of the mothers. Consequently, Chapter 9
the researchers asked school principals and various 1. The standards might be those (a) used by the advisor
school personnel to assist in recruiting participants. in assessing past students in the program, (b) held
2. Ethical issues might have developed in how the inter- personally by the advisor, and (c) used by the college
views were collected and whether the researchers or university.
might have created a power imbalance by their role as 2. Correct the discriminatory language:
researchers versus mothers in this study. Also, issues a. Poor (level of specificity problem): Teachers under
might arise in how the results would be used from this 25 adopt an open learning environment for
study. Did the results have a negative impact on some students.
principals? Did the authors use the results to point out Improved: Teachers in their first 2 years out of
problems in the schools (as opposed to opening a dia- teacher education adopt an open learning environ-
logue about constructive change) or to coerce the ment for students.
principals? The authors should certainly have men- b. Poor (insensitive to labels): The at-risk students
tioned the issue of the use of the results of this study scored below the average in math.
in a sensitive project such as this one. Improved: The students who have been identified
3. When a study topic is sensitive (e.g., mothers’ trust in as at risk scored below the average in math.
school administrators), the best approach is to collect c. Poor (fails to acknowledge participation of individ-
one-on-one interview data in which the interviewee uals with specific cultural identification): The
can feel free to express their views. Observations in homeless used the school shelter at night.
the school would not necessarily yield perspectives of Improved: The adults who do not have a perma-
mothers. The researchers might have collected addi- nent residence used the school shelter at night.
608 APPENDIX A Answers to the Chapter Study Questions

3. It was more of a quantitative approach. It had a struc- vide the results to the question in nonstatistical terms.
ture much like a quantitative study, especially in the I would strongly emphasize the results and the impli-
introduction with the extensive use of literature. cations of the study.
4. The bulleted points that I would include would 5–9. Answers will vary.
address the key results for each of the four research
questions. I might pose each question and then pro-
Determine Size Using
Sample Size Tables
B
A P P E N D I X

In the process of collecting data, researchers need to deter- we will explore the calculations involved in the sample
mine the number of participants to use in their studies. size formulas and present tables that simplify the process
Several options are available. They can make an educated of making the calculations.
guess as to how many people are needed, such as 10% of
the population. They can ask as many people to partici-
pate as possible within the resources and time that both
researchers and participants can provide. They can select a
number that satisfies different types of statistical proce-
SAMPLING ERROR FORMULA
dures, such as approximately 30 scores for each variable in
a correlational analysis. A sampling error formula is often used in survey or corre-
A more rigorous approach than any of these is to sys- lational research (see Fink & Kosekoff, 1985; Fowler, 2008)
tematically identify the number of participants based on when investigators seek to generalize results from a sam-
sample size tables available in published texts. To under- ple to a population. A sampling error formula is a
stand these tables, you need to understand the fundamen- calculation for determining size of a sample based on the
tals of the formulas used. Two formulas will be explained chance (or proportion) that the sample will be evenly
here: the sampling error formula and the power analysis divided on a question, sampling error, and a confidence
formula. interval.
This discussion about the two formulas builds on ear- The formula is based on the proportion of the sample
lier comments about the importance of systematically cal- that will have the desired characteristic that you are trying
culating sample size using formulas in quantitative to estimate. For example, when parents are polled to deter-
research. This importance was introduced in chapter 5 in mine whether they would vote “yes” or “no” on a school
the “Sample Size” section and reinforced in chapter 10 on bond issue, there is a 50/50 chance that they will vote
experimental designs under “Step 3. Select an Experimental “yes.” Selecting a proportion of 50/50 means that the pop-
Unit and Identify Study Participants.” It was also mentioned ulation will be evenly split, and this proportion yields the
in chapter 11 on correlational designs, although sample largest size for your sample.
size is often dictated by the required size for making The sampling error formula is also based on stating
assumptions about normality of the distribution of scores. the amount of sampling error you are willing to tolerate.
(See chapter 11 discussion about size in correlation studies Recall that sampling error (stated as a percent, such as
in “Step 2. Identify Individuals to Study.”) However, a sam- from 4% of the time) is the difference between your sam-
pling error formula may be used when the intent of the ple mean and the true population mean. This error results
correlational study is to generalize from a sample to a pop- because samples are randomly selected from the popula-
ulation. Using a formula was also encouraged again in tion and may not represent the true characteristics of the
chapter 12 on survey designs in “Step 3. Identify the population (see chapter 5). Finally, the formula also
Population, the Sampling Frame, and the Sample.” Here includes identifying a confidence interval, such as a 95%
609
610 APPENDIX B Determine Size Using Sample Size Tables

TABLE B.1
Fowler’s (1988, p. 42) Sample Size Table: Confidence Ranges for Variability
Due to Sampling*
Percentage of Sample with Characteristic

Sample Size 5/95 10/90 20/80 30/70 50/50

35 7 10 14 15 17
50 6 8 11 13 14
75 5 7 9 11 12
100 4 6 8 9 10
200 3 4 6 6 7
300 3 3 5 5 6
500 2 3 4 4 4
1000 1 2 3 3 3
1500 1 2 2 2 2

Note: Chances are 95 in 100 that the real population figure lies in the range defined by +/– number indi-
cated in table, given percentage of sample reporting characteristics and number of sample cases on which
the percentage is based.
* This table describes variability due to sampling. Errors due to nonresponse or reporting errors are not
reflected in this table. In addition, this table assumes simple random sample. Estimates may be subject to
more variability than this table indicates due to the sample design or the influence of interviewers on the
answers they obtained; stratification might reduce the sampling errors below those indicated here.
Source: Fowler, F. J. Survey Research Methods, p. 41. Copyright © 2009. Reprinted by permission of Sage
Publications, Inc.

confidence interval. Recall that a confidence interval indi- population mean. Finally, in the left column we see the
cates the upper and lower values that are likely to contain sample size recommendation that you would use as a
the actual population mean (see chapter 6). guide to the minimum sample size for the sample in your
Understanding these three factors helps you interpret study.
a sample size formula table, such as Table B.1 by Fowler Let’s apply information from this table to a study to
(1988, p. 42). The first row in this table shows the percent- see how it works. Assume that you need to determine the
age of the sample with the desired characteristic, ranging size of your sample in a study of Native American students
from 5/95 (small chance) to 50/50 (equally split chance). in a school district. You want to survey students in high
To maximally divide the sample, researchers typically schools in a large metropolitan district to determine if stu-
select 50/50 as the proportion of the sample with the char- dents plan to enroll in advanced placement courses (“yes”
acteristic they are trying to estimate. In terms of the confi- they plan to enroll, “no” they do not plan to enroll). The
dence interval, this table reports only a 95% confidence procedure needs to be as rigorous as possible, so you
interval (default), which means that 95 out of 100 times the decide to use Fowler’s (1988) table to calculate the sample
sample mean will fall within the upper and lower limits, or size for your study.
range, of the population mean. This is a rigorous standard A given in Fowler’s (1988) table is that you will use a
to use. In the columns under the heading “Percentage of rigorous confidence interval standard—a 95% confidence
Sample with Characteristic,” we see values, such as 17, 14, interval (95 out of 100 times your sample value will fall
and 12, under the 50/50 column. These values are the within the range of the population mean). You assume that
amount of sampling error we are willing to tolerate. students have a 50/50 chance of participating in these
Typically, researchers set a small error that they are willing courses. Based on this information, you select the column
to tolerate, such as 4% or 6%. This means that only 4% or 50/50. Further, you want a low error rate—a small percent-
6% of the time the sample mean will differ from the true age of the time your sample mean will differ from the
 APPENDIX B Determine Size Using Sample Size Tables 611

population mean. You select an error of 4% (4 out of 100 appropriate sample size for group comparisons by taking
times). into consideration the level of statistical significance
To identify the appropriate sample size you need in (alpha), the amount of power desired in a study, and the
your study, you look at the last column (50/50), go down effect size. By determining these three factors, you can
the column to “4” (4%), and then look across the row and look up the adequate size for each comparison group in an
find that the ideal sample size is 500 Native American stu- experiment and use tables available in published texts
dents. This number, based on the sample size formula, will (e.g., Lipsey, 1990). The process works this way as shown
ensure that 95 out of 100 times (95% confidence interval) in Table B.2:
your sample mean will have an equal chance (50/50 split)
◆ First identify the statistical level of significance to
of differentiating among the students 96% of the time (or
use in testing your group comparison hypothesis,
an error of 4%).
typically set at p = .05 or p = .01. (See chapter 6
discussions on hypothesis testing.)
◆ Next identify the power needed to reject the
POWER ANALYSIS FORMULA hypothesis when it is false, typically set at .80.
(See chapter 6 discussions on types of outcomes of
hypothesis testing.)
In many experiments, the size of the overall number of ◆ Determine the effect size, which is the expected
participants (and participants per group) is dictated by difference in the means between the control and
practical issues related to the number of volunteers who experimental groups expressed in standard deviation
enroll for the study or the individuals who are available to units. This effect size is often based on expectations
the researcher. Researchers can also use statistics to ana- drawn from past research and is typically set at .5 for
lyze the data, and these statistics call for minimum num- much educational research (Murphy & Myors, 1998).
bers of participants for each group when group compari- (See chapter 6.)
sons are made. ◆ Go to a table for calculating the size of the sample
A rigorous, systematic approach is to use a power given these parameters, and identify the size of
analysis. A power analysis is a means of identifying each group in the experiment. This size becomes

TABLE B.2
Lipsey’s (1990, p. 137) Sample Size Table: Approximate Sample Size per Experimental
Group Needed to Attain Various Criterion Levels of Power for a Range of Effect Sizes
at Alpha = .05
Power Criterion

Effect Size .80 .90 .95

.10 1570 2100 2600
.20 395 525 650
.30 175 235 290
.40 100 130 165
.50 65 85 105
.60 45 60 75
.70 35 45 55
.80 25 35 45
.90 20 30 35
1.00 20 25 30

Source: Lipsey, M. W. Design Sensitivity: Statistical Power for Experimental Research, p. 137. Copyright ©
1990 by Sage Publications. Reprinted by permission of Sage Publications, Inc.
612 APPENDIX B Determine Size Using Sample Size Tables

the number of participants you need for each group In other words, we need an experiment with sufficient
in your sample. The approximate sample size per size to have power (see Table 6.8 on possible outcomes in
experimental group with an alpha set at .05 is given hypothesis testing). We turn to Lipsey’s (1990) power anal-
in Table B.2 (Lipsey, 1990, p. 137). ysis table for assistance in determining the appropriate
sample size for our experimental and control groups. This
Let’s take an example to apply the power formula and table will indicate the size needed given a confidence
Lipsey’s table. Assume that elementary education children level, the amount of power desired, and the effect size.
identified as gifted in a school district are assigned to one Examining Lipsey’s table, we find that the significance
of two groups. One group receives an enrichment program level for the table is set at an alpha = .05.
(the experimental group), and the other group receives tra- We use a rigorous standard for power, such as .80
ditional instruction (the control group). At the end of the (80% of the time we will reject the null when it is false),
semester both groups are tested for creativity. How many and select the column “.80.” Then we look at the column
gifted students are needed for both the experimental and for effect size and choose “.50” as the standard for differ-
control groups? We might use the number of students avail- ences in the means (in standard deviation units) that we
able and equally assign them to groups. Although such an will expect between the two groups. Using this informa-
experiment could be made, we want a sufficient number tion, and going down the column of .80 to the row .50, we
of students in our groups so that we can be confident in find that 65 students are needed for each of our two
our hypothesis test of no differences between the control groups, the experimental and control groups, in our study
and experimental groups. of gifted children.
Commonly Used Statistics
in Educational Research
C
A P P E N D I X

Statistics
Name Symbol Use Example Application
t test t To test for a difference between 2 groups Compare boys and girls on time spent reading
in terms of 1 dependent variable
Analysis of F To test for a difference among 2 or more Compare four groups (freshman, sophomores,
variance groups in terms of 1 dependent variable juniors, and seniors) on time spent studying
(ANOVA)
Chi square x To test for a difference among groups in Compare men and women in terms of their
terms of a categorical dependent variable political party affiliation (democrat, republican,
independent)
Pearson R To test for a relationship between two Determine whether time spent studying is
correlation variables related to grade point average
Multiple R To determine the degree to which 2 or Determine whether grade point average,
regression more independent variables are related to SAT score, and depression combined predict
(or predict) 1 dependent variable retention in the first year of college

Source: Adapted from Plano Clark, V. L., & Creswell, J. W. (2010).

613
Nonnormal Distribution D
A P P E N D I X

We cannot collect sample after sample of individuals to flat picture (positive kurtosis). In a skewed distribution,
study (e.g., collect their depression scores), but we can the scores tend to pile up toward one end of the scale and
plot the scores of our sample and see what shape the dis- then taper off slowly at the other end (Gravetter & Wallnau,
tribution of their scores might be. On the one hand, they 2000). In a kurtosis distribution, the scores pile up at the
may reflect a normal curve. On the other hand, they may middle or spread out to the sides. Most distributions are
not be normal. Examine Figure D.1. Here we see a non- not normal and will exhibit some skewness or kurtosis.
normal distribution of scores in which the actual shape Whether the distribution of scores from your sample are
of a frequency distribution of scores may be negatively normal or nonnormal is one factor that determines what
skewed to the right, positively skewed to the left, concen- statistical test you will use to analyze your comparative or
trated around the mean (negative kurtosis), or resemble a relationship questions or hypotheses.

FIGURE D.1
Skewness and Kurtosis

4 4
Percentages
Percentages

3 3
2 2
1 1
0 0
Negative Skewness Negative Kurtosis

4 4
Percentages
Percentages

3 3

2 2

1 1

0 0
Positive Skewness Positive Kurtosis

614
Strategies for Defending
a Research Proposal
E
A P P E N D I X

In doctoral programs and in many master’s programs, stu- phases of your research? Can you substitute fac-
dents are required to complete a formal research study as ulty on your committee to obtain needed faculty
part of their requirements for graduation. This requirement resources at this point in your program? An ideal
is included in programs so students can demonstrate their committee consists of a statistician or research meth-
knowledge and application of research. As a first step in odologist who can help with the technical aspects
preparing the formal research study, a student develops a of a study, an editor who can help you shape your
proposal or plan for the study. The proposal is distributed writing, a content specialist who knows the litera-
to the student’s faculty. Then, a short time later, the student ture about your topic, and a supporter who can give
presents the proposal to the faculty in a meeting called a positive feedback throughout the process.
proposal defense. The intent of this meeting is for faculty to ◆ Talk with your advisor about the degree of formal-
have a chance to react and provide constructive comments ity required during the meeting. A formal meeting
about the study before the student begins the project. It consists of a brief, organized presentation by the
also serves to provide an opportunity for the student to student followed by faculty questions. It may also
carefully review the entire project, practice giving an oral include a slide presentation to present the proposal
presentation of research, and improve the study based on for the research study. An informal meeting may be
faculty feedback. a discussion or conversation about your proposed
The proposal defense meeting varies in length. A typi- project without the use of a slide presentation or an
cal meeting might last about 1 hour. During this meeting, organized, formal presentation.
the student presents an overview of the study, and faculty ◆ If possible, attend proposal defenses held by other
ask questions that arose when they read the proposal. students so that you can gain some experience with
Some defense meetings are open to all faculty and stu- the procedures of this meeting.
dents, whereas others are closed to all participants except ◆ Visit with your faculty committee members before
for the student’s faculty committee. To best prepare for this the meeting to gauge their feedback. They will likely
meeting and to leave with productive ideas, several strate- have suggestions for improving the study before the
gies are identified here that students might use before, dur- defense meeting.
ing, and after the meeting. ◆ Consider how you will build rapport with your
faculty committee prior to the meeting and during
your program. Rapport might be established infor-
PRIOR TO THE MEETING mally through social interaction or formally through
collaboration on projects or by serving as a teacher’s
assistant.
◆ Look over the composition of your committee. ◆ Discuss with your advisor her or his role during the
Do you have individuals who can assist you in all meeting. Advisors assume different roles, such as

615
616 APPENDIX E Strategies for Defending a Research Proposal

advocate, supporter, arbitrator, and adversary. The Legitimacy questions:

better you understand his or her role, the better you ◆ Are there journals that will publish this type of
can anticipate how your advisor will respond during research?
a defense meeting. ◆ Is this type of research rigorous enough?
◆ Anticipate questions that faculty will ask during a ◆ Will your involvement in the study influence the
proposal meeting. Sample questions during a meet- findings?
ing for a quantitative proposal might be:

Research design questions: DURING THE MEETING

◆ Is the problem trivial?
◆ What is the research question being addressed?
◆ Is the theory a good predictor to use? ◆ Keep your presentation brief—15 minutes or less.
◆ Is the study internally valid? ◆ Be open, negotiable, and responsive to faculty
questions.
Data collection and analysis questions: ◆ Restate questions in your own words to indicate that
◆ Are the methods detailed? you have correctly heard them.
◆ Is the best data collection approach used? ◆ Be willing to admit that you have not considered the
◆ Was the most appropriate statistic chosen? question being raised.
◆ Are the constructs valid? ◆ Have your advisor intercede if the questions become
◆ Do items on the instrument need refinement? unfair or unrelated to your topic.
◆ Was a good instrument chosen to be used in the ◆ Listen for constructive suggestions for improv-
study? ing your project. Keep notes of these suggestions
or have your advisor take notes for you so that
◆ Sample questions during a meeting for a qualitative later there will not be any question about changes
study will likely range from research design to the needed.
legitimacy of the design as compared with quantita-
tive research.

Design questions:
AFTER THE MEETING
◆ Why did you choose a qualitative study (instead
of a quantitative study)? ◆ Review the events of the meeting with your advisor.
◆ What is the type of design you are using (e.g., ◆ Note suggestions that need to be made to improve
ethnography, grounded theory)? or change your study. Negotiate with your advisor
◆ Is the study valid? about what committee suggestions need to be imple-
mented and what suggestions need to be revisited
Quantitative-type questions that require knowledge with members of the committee.
about qualitative research: ◆ Visit with faculty on your committee, if needed, to
◆ How can you generalize from this study? clarify their suggestions for improvement.
◆ Are your variables measurable and valid? ◆ Talk with your advisor about the next step in the
◆ Do you plan to use theory in this study? process—what chapters you should begin to write
◆ Why did you conduct a brief literature review? and when.