A Guide to Quantitative Research Proposals: Aligning Questions

November 202

A Guide to Quantitative

Research Proposals:

Aligning Questions,

Goals, Research

Design, and Analysis

Elizabeth Tipton, Ph.D.

Contents

1.0 Overview 03

2.0 Articulate Clear Research Questions 04

3.0 Set Clear Goals 05

4.0 Shape the Study Design from the Research

Question(s) and Goal(s) 07

4.1 Sample: Recruitment approaches 07

4.2 Measures: Operationalization of outcomes and variables 08

4.3 Interventions: Explanatory studies and causality 09

5.0 Conclusions, Samples, Estimates, and Uncertainty 10

6.0 Writing a Strong Quantitative Proposal 12

7.0 Appendix: Further Resources 13

A Guide to Quantitative Research Proposals

3A Guide to Quantitative Research Proposals

The purpose of this guide is to provide a broader

framework for thinking about and approaching

quantitative research grant proposal writing. It is divided

into four sections. The ﬁrst section highlights the types

of questions that can be answered with quantitative

research. The second section provides further guidance

in specifying the questions and goals of a study, including

the population and how conclusions will be drawn. The

third section focuses directly on the research design,

including how the sample will be recruited, the use of

appropriate measures, and questions of causality. The

fourth section focuses on the modeling of outcomes and

uncertainty due to sample size, providing guidelines and

questions to ensure that the analyses are appropriate for

the question type and goals of the study.

Importantly, although this guide presents an approach

for thinking about research and a variety of examples, we

do not rigidly specify how a researcher should convey this

information in a proposal. There are a variety of formats

and organizations possible, and depending upon the

research question and type, proposals may need to spend

more space on one part versus another.

• What are the possible conclusions or implications of the

research once it is ﬁnished?

• Be sure to deﬁne the population to whom the results of

your study are intended to apply, and conversely, where

they may not.

• The parameter(s) of focus (i.e., to be estimated) should

be clearly deﬁned.

• It is important that you identify the types of conclusions

you intend to draw and the criterion through which you

will arrive at these decisions.

• Be sure to clearly explain the research design and how

this design will enable you to estimate the parameters of

focus.

• For a given goal and conclusion, you should explain

how you arrived at the sample size that is necessary to

warrant these conclusions.

1.0

Overview

In this essay, noted scholarElizabeth Tiptonelaborates onhow to

best articulateresearch design in grant proposals. Thisessayis

a companion piece to our “A Guide to Writing Successful Field-

Initiated Research Grant Proposals,” which provides general

information about the elements of grant writing. We also have

guides for proposals focused on qualitative methods and

research practice partnerships. These essays offer some insights

for grant-proposal crafting as a writing genre. They are not

intended to bestand-alone guides for writing grant proposals,

receiving a Spencer grant, or conducting research.We recognize

that there are numerous valuable approaches to conducting

research and urge scholars to consider the specific concerns in

their respective disciplinary and methodological fields.

Question Type Example Possible claims

Descriptive: Who, What, When,

Why, Where, How?

What types of students dropped out

of high school over the past 20 years?

There is a problem/ gap/ inequality/

process/ phenomenon.

Explanatory: Does a change in X lead

to an increase/ decrease in outcome Y?

Does dropping out of high school

reduce future earnings?

A policy/ practice should be implemented

/expanded/ removed/ changed.

Predictive: Does this model /

algorithm improve our ability to

predict future outcomes?

Can we identify in advance the

students likely to drop out

This algorithm/ model should be

implemented.

Well-written proposals offer clear and appropriate

alignments from the guiding research question(s) through

the study design, data, sample, analysis, and its implications.

Quantitative research—and proposals—typically involve

one or more of three types of questions: descriptive,

explanatory, and predictive. Descriptive questions

focus on understanding and describing a problem, gap,

procedure, or practice. Explanatory questions – also called

‘causal’ questions – move beyond describing a problem,

focusing instead on possible interventions, practices, or

policies that are able to cause changes in outcomes or

practices. Importantly, both descriptive and explanatory

questions tend to focus on broad populations or samples

and aggregate analyses across these in the past or

present. Predictive questions, in comparison, focus on the

development of algorithms and models to identify and

predict future outcomes, often for individuals.

Determining if a question is descriptive, explanatory, or

predictive is often more complicated than it might seem,

because the same methods (e.g., regression) can often be

used to answer all three questions. One helpful approach in

determining the question type is to focus on the possible

claims that can be made once the study is ﬁnished, i.e., the

implications of the study. If the ﬁndings will be interpreted

as evidence supporting the use of a policy or practice, then

the question is likely explanatory.

If the implication is that a new algorithm or model should

be used to identify students or schools in need of an

intervention, then the question is predictive. Finally, if the

conclusion is that a form of inequality, social structure, or

problem is now better understood, the question is likely

descriptive. Table 1 provides examples for each of these

question types to help determine the kind of question

being asked.

The type of question asked should be in clear alignment

with the choice of research design, analysis plan, and

potential conclusions drawn. For example, when asking

a question that is inherently explanatory (e.g., if a policy or

intervention improves outcomes for students),

the research

design and analysis plan should be appropriate for research

(e.g., a design that appropriately addresses causality).

Without clear alignment between question and design,

the conclusions and goals of the proposed study will not

be met using the design and analysis strategy provided.

Importantly, this is true even if the method used in the study is not a

randomized experiment. Furthermore, the fact that a change is suggested

means that the intervention under study is something that can be

manipulated and is not an attribute of an individual (e.g., race, gender)

or group (e.g., urbanicity).

2.0 Articulate Clear Research Questions

4A Guide to Quantitative Research Proposals

Table 1. Three general kinds of quantitative questions

Often the goal of the study is not merely to report a

statistic—e.g., an estimate or conﬁdence interval—but to

also draw a conclusion or make a decision based upon this

estimate. For example, in an explanatory question, the study

may conclude that an intervention is successful if there is

(a) evidence that the sample effect size estimate is > 0.20

standard deviations (or another threshold), and (b) that

there is evidence the effect is non-zero in the population

(e.g., a hypothesis test based on a Type I error of 0.05). In a

descriptive study, there may not be a decision per se, but

instead, the goal may be to precisely estimate a measure

(e.g., of poverty or achievement). In this case, a margin of

error might be speciﬁed. Finally, in a predictive study, a

new method or algorithm might be considered useful if it

has a lower mean-squared-error than an existing measure

or reaches a target accuracy metric. Importantly, these

criteria given here are only examples. Others might be more

appropriate depending on the case. What is important,

however, is that criteria are stated in advance. Table 2

below provides examples of these for each of the types of

questions.

Clear questions and goals shape research design and

analytic strategies. When misaligned, research projects fall

apart in their logic. For example, if a proposed explanatory

study seeks to determine if an intervention can improve

student learning broadly for all students, but then indicates

that the schools in the study are all found in a single large

urban school district, the results will not be able to meet the

proposed goal. Persuasive research proposals are careful,

therefore, to ensure that the questions asked are answerable

and stated inquiry goals are feasible using the proposed

research design and analysis methods.

3.0 Set Clear Goals

Offer a clearly deﬁned and speciﬁc research question and

goals for the study, once the question type is identiﬁed.

This speciﬁcation will ultimately narrow the question from

a broader class of interest for study—e.g., about inequality

in all schools—to the narrower class that can be feasibly

answered in the proposed study, given the budget and

real-world constraints. For example, while inequality may

be a broad concern, in a particular study one might focus

speciﬁcally on school funding inequality in North Carolina

elementary schools in 2018-2019. This specificity does not

make the research any less important—but it does help in

deﬁning the scope of the claims that can be made from the

study.

In framing a study, it is important to deﬁne the population

to whom the results are intended to apply, and conversely,

where they may not. It is often helpful to turn to population-

level data, such as the Common Core of Data (an annual

census of schools), the American Community Survey, the U.S.

Census, or national probability survey data like the

Early Childhood Longitudinal Study (ECLS), to deﬁne the

characteristics of this population. For example, it may be that

the study focuses on rural students of color in the American

South, or that the study focuses on schools in Texas. Keep in

mind that the population deﬁnition is broader than simply

reporting summary statistics on the sample in hand (unless

the population of interest is the sample, which is possible).

Distinguishing between a sample and population often

involves asking questions like, “What is this sample a case

of?” and “Where might results of this study not apply?”

Once a target population is deﬁned, offer a clear deﬁnition

of the parameter(s) of focus in the study. This requires

narrowing the question and anticipating the kinds of

‘output’ for focus in the quantitative analyses. For an

explanatory question, the goal may be to estimate the

average treatment effect, subgroup treatment effects, or

treatment effect differences. For a descriptive question,

the goal may be to provide summary statistics (e.g., means,

standard deviations, correlations), or the amount of variation

in an outcome explained by different subsets of variables.

For a predictive question, the goal may be to develop an

algorithm that can predict well (e.g., 95% accuracy) an

outcome for each student or school in the population. Again,

it is important to distinguish here between the parameter,

which is unknown in the population, and the estimator,

which is based on the model and data in the study.

A Guide to Quantitative Research Proposals

Topic Questions Descriptive Explanatory Predictive

Population What population is this

study about? What is it

not about? (What are

inclusion/ exclusion

criteria?)

High school students

in public schools in the

U.S. that were included

in NELS.

Public, regular,

elementary schools

serving > 50% students

on free-or reduced-

priced lunch in

Alabama.

Elementary school

students in Texas in

public schools captured

in the state longitudinal

data system.

Parameter What is the population

quantity of interest?

(There may be more

than one).

What is the average

and variation of X? How

correlated is X with

Y? How much of X is

explained by Y and Z?

What is the average

treatment effect of X

on Y?

What is the predicted

value of Y given observed

characteristics?

Criteria Is there a decision to be

made? If so, what is the

criterion – how will you

know there is “enough”

evidence or that the

model is “good”?

The estimates of the

mean will have a margin

of error within X units,

giving conﬁdence

intervals with a width of

2X.

The intervention is

effective if there is

sufﬁcient evidence (p <

.05) that the treatment

effect is non-zero in the

population and if the

estimated effect size is

medium in size (e.g., ≥

0.20 SDs).

The algorithm is effective

if it predicts Y better (e.g.,

has lower MSE) than

other available methods.

Table 2. Goals, Questions, and Examples

6A Guide to Quantitative Research Proposals

A researcher might plan to collect data themself, for

example, in a survey or experiment. An important

consideration is how to approach recruiting a sample

that represents the population of interest. A probabilistic

selection plan may be ideal, but not often very feasible,

particularly with uncommon subgroups. However,

researchers must contend with this question even when

the sample will be recruited based mainly on convenience—

e.g., using Facebook or MTurk or by recruiting schools in

neighboring school districts. Importantly, this may mean

that budgetary and resource constraints of the study will not

allow for generalizability to a target population. In this case,

the target population speciﬁed may need to be narrowed,

and the logics made clear in a proposal.

Just as importantly, to make claims about how a sample is

related to a population, data collection is needed. In a study

about college students, for example, it may be helpful to

know the demographics of students in the population and

to collect similar demographics for those in the sample.

This will allow an assessment for reporting the degree of

similarity between the sample and target population. It is

important for a proposal of this type to offer a clear plan for

data collection and analyses.

Finally, it is important to note that all studies have

limitations, particularly based on the type of sample

recruited (or data available) compared to the population

more broadly. There is no perfect study, and reviewers

understand this. Convincing proposals, therefore, anticipate

problems, carefully examine possible approaches, and

acknowledge study limitations.

4.0

Shape the Study Design from the

Research Question(s) and Goal(s)

Research design connects the speciﬁc research question to

the study’s claim or conclusions. The design operationalizes

how the sample will be collected, how outcomes will be

measured, and if relevant, how causality will be determined.

Importantly, there may be multiple ways to design a study.

This is a place where creativity and ingenuity show. However,

some approaches are particularly well suited to speciﬁc

question types. This section gives a broad overview of these

problems and common approaches.

4.1 Sample: Recruitment approaches

Regardless of question type, nearly all studies focus on

a sample, not the full population. As a result, how this

sample is (or was) recruited has broad implications for

generalizability to the target population (deﬁned in Section

3.0). In some studies, these data may have been previously

collected. For example, descriptive and predictive research

often turn to large national probability samples, like the Early

Childhood Longitudinal Survey (ECLS). The fact that these

studies involved the random selection of units (e.g., schools)

into the survey enables clear generalizations to

a target population. Here the documentation to these

surveys can help determine any particular concerns for

generalization. For example, they may have excluded

particular parts of the population (e.g., private schools), they

were collected only during a speciﬁc period of time, or there

may be concerns with non-response.

In other studies, the analyses focus on existing data that

were not collected probabilistically. This type of data

is particularly common for predictive and descriptive

questions, where rich data on subgroups or processes are

available only for those that completed a survey or who

interacted with a particular product. For example, data

may have been collected by a publisher on how students

interacted with an online educational game or via an Mturk

or Facebook survey regarding attitudes and beliefs about

inequality. In these cases, the prevailing design concern

is with how those that are in the data (e.g., completed the

survey) might or might not represent the target population.

How might this impact study results if those motivated

to complete the survey on inequality are those with the

strongest feelings? Or what if those schools that have

provided data on an educational software platform tend to

be highly concentrated in private schools in San Francisco?

These external validity bias concerns should be considered.

A Guide to Quantitative Research Proposals

4.2 Measures: Operationalization of

outcomes and variables

Regardless of question type, all studies require careful

consideration of how important outcomes and variables

will be operationalized and measured within a study. For

example, at a broad level, a study might focus on the effect

of socio-economic status (SES) on student achievement

outcomes, and particularly a sample of schools in North

Carolina. But how should socio-economic status and

achievement be measured? There are a variety of ways SES

has been conceptualized in the literature (e.g., free- or

reduced-priced lunch indicators, based on mother’s income,

based on mother’s income and occupational prestige, etc.).

Similarly, achievement can be measured in many ways—

based on a proximal measure specific to the study in hand

(e.g., knowledge of fractions), a researcher-developed

measure, a more distal measure available in extant data (e.g.,

NC state second-grade math score), and so on. The

relationship between SES and achievement depends upon

the operationalization of each.

One concern here is with regard to construct validity—how

well an item measures what it says it measures. This is a

high-level question and requires one to think carefully about

(a) what exactly the broader construct is, (b) what possible

candidate operationalizations (i.e., items) there are available,

and (c) how to tease apart empirically how well these

candidates actually meet the study goal.

At a more specific level, there are concerns with

measurement validity, which has to do with the level of

measurement, the variable encoding, and the reliability of

this measurement. The level of measurement might be, for

example, at a particular time-point for a person, a general

person-specific measure, at the classroom or school level. To

continue this example, the relationship between SES and

achievement could be measured at the student level or the

school level – and again, these relationships may differ.

Furthermore, SES could be encoded as a dichotomous

variable (e.g., low-SES versus not), as a continuous variable, or

as an ordinal variable.

A Guide to Quantitative Research Proposals

Reliability has to do with the ‘consistency’ or ‘repeatability’

of selected measures. Reliability statistics (e.g., Cronbach’s

alpha) are often provided for existing measures, particularly

test scores and other psychological constructs. However,

these statistics are speciﬁc to the populations and samples

studied. In this example, an achievement test (or items from

it) might have been shown to be highly reliable for 8

-grade

students, but it may not be for 4

-grade students. Similarly,

the entire achievement test may have high reliability, but the

sub-score focused only on fractions may not.

Just as with sample recruitment, there is no perfect study

with respect to measurement. Persuasive proposals make

clear how constructs will be measured and encoded,

available information on the reliability of these measures,

and possible limitations to these approaches. Compelling

proposals anticipate problems, carefully examine possible

strategies, and acknowledge study limitations.

Just as with sample recruitment and measurement, there

is no perfect study with respect to causality. For a particular

study, the most compelling proposals identify the most

robust possible design given the budget and resource

constraints. Persuasive proposals anticipate potential

threats to causality in the study and carefully plan for

approaches to mitigate these. For example, this might

include planned robustness or sensitivity analyses, as well

as clear statements about assumptions required and

possible study limitations.

4.3 Interventions: Explanatory studies

and causality

In an explanatory study, the goal is to make a causal claim,

by estimating the effect of an intervention. In the literature,

this is referred to as

internal validity. Here the ‘gold standard’

design is a randomized experiment. In such an experiment,

students, classrooms, teachers, or schools might be

randomized to receive a new intervention. It is important

here to carefully think through questions like: What is the

comparison condition? What are the components of the

intervention? How will the intervention be implemented? Is

there any chance the intervention might be shared with the

comparison condition?

Given the difﬁculties of implementing randomized

experiments, researchers often turn to quasi-experiments

instead to answer causal questions. These research designs

include regression discontinuity, instrumental variables,

interrupted time-series, and non-equivalent control group

designs. In all of these cases, since the intervention is

not randomized, it can be challenging to tease apart the

effect of the intervention itself from the selection of the

intervention. In other words, those that implement an

intervention may also be different in a variety of ways

from those that do not implement the intervention. For

example, students that are retained for an additional year

of Kindergarten typically differ in other ways (e.g., age, SES)

from those who are not, making it difficult to tease apart the

“effect” of grade retention from other experiences and

characteristics.

A Guide to Quantitative Research Proposals

Question Type Example Models and Methods

Descriptive Geographic (e.g., GIS). Networks. Summary statistics. Correlations. Regression.

Explanatory Average difference in group means (randomized experiment). Regression with

controls. Difference-in-difference. Propensity scores. Instrumental variables.

Two-stage least squares. Non-parametric regression (e.g., regression discontinuity).

Predictive Regression. Machine learning. Artiﬁcial intelligence. Supervised learning.

Given this estimator, the next concern is with uncertainty

and the ability of the sample in this study to answer the

question and meet the goals of the study deﬁned previously.

All estimators have a measure of uncertainty associated

with them, and these measures of uncertainty depend

on the research design. For example, if schools are the

unit of sampling or random assignment and outcomes

are measured on students, then standard errors should

take this clustering into account. Similarly, clustering and

selection probabilities should be considered in the standard

errors calculated using analyses of large national probability

surveys (e.g., ECLS).

5.0

Conclusions, Samples, Estimates,

and Uncertainty

Once the question type, goals, and study design are

clear, the proposal should clearly deﬁne the models and

methods that will be used to estimate the parameters

of focus and draw conclusions based upon the speciﬁed

criterion. Depending upon the question, the models

and methods used might range from very simple (e.g.,

means, correlations) to very complex (e.g., multilevel

models). Models should be chosen to estimate the target

parameters (Section 3.0), as well as to appropriately

account for sources of uncertainty. Table 3a presents a

possible, but not exhaustive, sampling of methods and

models. As highlighted before, the same methods (e.g.,

regression) can be used for all three question types.

For a given model, one should clearly identify the

parameter estimator. For example, in a descriptive

study, the sample correlation may be reported, while in

an explanatory study, the analytic focus may be on the

second-stage regression coefﬁcient associated with an

instrument in an instrumental variable analysis. Or, more

commonly, a regression model may be used with many

control variables to estimate the average treatment effect

of an intervention in a quasi-experiment; in which case, the

coefﬁcient of focus is that associated with the intervention,

not those associated with the controls.

Table 3a. Example models and methods

10A Guide to Quantitative Research Proposals

Topic Questions Descriptive Explanatory Predictive

Estimator What is the estimation /

prediction strategy (e.g.,

boosted regression)?

Summary statistics –

means and SDs – will

be calculated for each

question. When there are

multiple items regarding

the same construct,

latent variables will be

estimated.

Dummy-variable

regression coefﬁcient

in a hierarchical linear

model that accounts for

nesting of students in

schools (unit of random

assignment).

Bayesian predictive

models using multivariate

hierarchical models with

X speciﬁcations will be

used to predict X. Half of

the data will be set aside

for validation.

Uncertainty /

Design

Sensitivity

What is the measure of

certainty (e.g., standard

error, predictive validity,

reliability)? How do

design parameters

affect this measure?

When estimating the

mean, the margin of

error is a function of

sample size and the

residual variation.

Standard error: a

function of the intra-class

correlation, sample

size, and effect size.

Prediction intervals are

a function of the sample

size, residual variation

in the population, and

covariate variation in the

sample.

Sample size Given the goals

and estimators and

uncertainty, what sample

size is required?

For a margin of error of

X, a sample size of X is

required.

The minimum

detectable effect size

(MDES) for power of

80% with α = .05 is X.

To classify students with

X% error, a sample size of

X is required.

Table 3

b provides a handful of examples for each of these

three concerns—estimator, uncertainty, and sample size.

Persuasive proposals clearly convey the degree of certainty

(precision) or uncertainty expected in the estimates, as a

result of sample size and budgetary and other constraints,

and how this degree of certainty will affect any conclusions

the researcher seeks to draw. Here, the alignment between

what is desired (the goal) and what is possible (the

uncertainty) is important. For example, a proposed study

that seeks to determine if an intervention will improve

student achievement (a hypothesis test question) but has

a very small sample with limited statistical power (e.g.,

0.20) is unlikely to be funded. This is not to say that small

studies, exploratory projects, or pilot studies cannot be

considered strong and fundable proposals, but simply that

the conclusions that such studies can possibly draw should

be tempered by the scope of the project.

11A Guide to Quantitative Research Proposals

In most studies, to some degree the degree of uncertainty

can be anticipated in advance through a power

analysis.

Measures of uncertainty can be written as functions of

design parameters (i.e., sample characteristics that are

knowable before a study is started). For example, the

standard error of a regression coefﬁcient is a function

of the sample size (n), the variation in the covariate in

the sample (s

), the residual variation (σ

) and the degree

of multicollinearity (VIF). Or, in a group-randomized

experiment, the uncertainty is a function of the intra-class

correlation (ρ, the degree of clustering), as well as the

number of student (n) and school (J) observations. Even in

prediction models, the ability to classify well is a function

of covariate features and sample size. In all cases, some or

all of these characteristics are knowable in advance based

on previous studies or extant data, and software is available

to help conduct these analyses. Thus, given the decisions

and conclusions that will be made from a study and the

estimation strategy used, researchers should be able to

determine the sample size that is necessary to warrant

these conclusions.

Table 3b. Estimators, Uncertainty and Sample Size Examples

In conclusion, no amount of fancy statistical modelling

can make up for a poor study design. Compelling research

proposals take this to heart, clearly deﬁning the logic of the

research process from beginning to end. As outlined in this

guide, this involves:

• Accurately determining the type of question being asked.

• Cleary stating the speciﬁc population and parameters

under study and how conclusions will be drawn from the

study.

• Carefully describing the research design, including sample

recruitment, measurements, and, when required, strategy

for identifying causality.

• Appropriately selecting a model for estimation and

indicating that the sample size of the study will be

appropriate for drawing the stated conclusions and

implications

Perhaps more important than any one of these steps,

however, is the alignment of these steps throughout

the proposal. The alignment between the question,

goal, design, and analysis are of utmost importance.

For additional resources on quantitative research design,

please refer to the appendix, which includes articles,

books, software, and training opportunities.

6.0 Writing a Strong Quantitative

Proposal

12A Guide to Quantitative Research Proposals

13A Guide to Quantitative Research Proposals

7.0

Appendix: Further

Resources

Question

Descriptive research questions:

1. Loeb, S., Dynarski, S., McFarland, D., Morris, P., Reardon, S., & Reber, S. (2017).

Descriptive analysis in education: A guide for researchers. (NCEE 2017–4023).

Washington, DC: U.S. Department of Education, Institute of Education

Sciences, National Center for Education Evaluation and Regional Assistance.

Explanatory versus predictive questions:

2. Schmueli, G. (2010) To explain or to predict? Statistical Science, 25(3): 289-310.

Explanatory versus predictive questions:

3. Murnane, R. J., & Willett, J. B. (2010).Methods matter: Improving causal

inference in educational and social science research. Oxford University Press.

4. Morgan, S. L., & Winship, C. (2015).Counterfactuals and causal inference.

Cambridge University Press.

Goals

Pre-registration:

1. A best practice is to pre-register hypotheses, allowing conﬁrmatory

and exploratory tests to be clearly deﬁned. In education, the Registry of

Educational Effectiveness Studies (free) provides formats for pre-registering

both experiments and quasi-experiments.

General overview and guide:

2. Duran, R. P., Eisenhart, M. A., Erickson, F. D., Grant, C. A., Green, J. L., Hedges,

L. V., & Schneider, B. L. (2006). Standards for reporting on empirical social

science research in AERA publications: American Educational Research

Association.Educational Researcher,35(6), 33-40.

Design

Populations:

3. The Generalizer is a free webtool that can be used to help deﬁne a target

population of schools (both K-12 and higher education) in the United States,

and to develop a recruitment plan to represent this population.

Measurement and Causality:

4. Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002).Experimental and quasi-

experimental designs for generalized causal inference. Boston: Houghton

Mifﬂin,.

Analysis

1. The Society for Research on Educational Effectiveness maintains a

professional development library (with videos of prior workshops) as well as a

list of software useful to researchers.

2. There are several options available for conducting analyses of statistical

power including:

• G*Power (both PC and Mac, free) – for laboratory experiments, t-tests, ANOVA,

and individually randomized trials.

• Optimal Design (PC based, free) – for cluster-randomized designs (e.g., school

randomization), randomized block designs (e.g., classrooms within schools

randomized), and longitudinal growth models.

• PowerUp! (Excel-based, free) – cluster randomized designs, randomized

block designs, as well as mediators and moderators of treatment effects in

these designs.

3. There is a wide range of workshops available each year funded by IES and NSF

intended to build methodological capacity for education researchers. Those

interested may want to sign up to be on the IES and NSF listservs to receive

announcements of future workshop opportunities. As of February 2020, there

are ﬁve available to researchers:

• NSF funded workshops (focus on STEM education):

o Study Design (Summers 2021 – 2023)

o Introductory Meta-Analysis (Summers 2021 – 2023)

o Quantitative Methods and Measurement Intro

(Summers 2020 – 2022)

• IES funded workshops:

o Advanced Meta-Analysis (Summers 2018 – 2021)

o Cluster Randomized Trials (Summers 2007 – 2021)

14A Guide to Quantitative Research Proposals

Spencer Foundation

625 North Michigan Avenue

Suite 1600, Chicago, IL 60611

spencer.org