NSF reviews

Report as inappropriate Your report has been received

Shared by: HC120929192059

Tags

Stats

views:: 3
posted:: 9/29/2012
language:: Unknown
pages:: 16

Public Domain

Document Sample

Panel Summary #1

Proposal Number: 1224087

Panel Summary:
NSF Sociology Program
PANEL SUMMARY

SES-1224087
PI Jennifer Lundquist
U Mass-Amherst

The Sociology Advisory Panel met on April 10-11, 2012 and evaluated a total of 83 projects.
The 83 projects included all regular research projects. Because 13 of the 83 projects were
collaborative proposals, the review panel actually considered 96 research proposals. But 13 of
these 96 regular research proposals did not require independent review because they were
included as parts of collaborative proposals (i.e., proposals simultaneously submitted by
principal investigators from different institutions to complete work on a unified research project).
Other NSF Programs participated in the joint review of 54 (65 percent) of the 83 projects.

Brief Description of Project:

This proposal is a national study that looks at how race/ethnic identity affects partner selection
processes. It accesses an online dating site. The question is: how does race/ethnicity intersect
with other factors in driving mate selection? The goal is to access over 4 million requests for
dates and responses to requests, and a pilot study is underway.

Intellectual Merit

Strengths:

The study is innovative in being a national study with millions of users. It seems innovative
methodologically. Many studies that look at mate selection look at marriage, but the pre-
marriage part of mate selection is also interesting.

Status exchange theory and intersectionality theory ground the study. The PI is well qualified
and has a proven record of NSF funding and top publications.

The PI is very well qualified to carry out this research.

Weaknesses:

The literature review is incomplete. The proposal doesn't reference prior studies. The PI
speculates about interracial dating among Asians, for instance, but doesn't cite studies. The
theory of mate availability is disregarded, and propinquity is conflated with opportunity.
The work of Gary Gates needs to be cited for his work on gay and lesbian populations. See also
Guttentag & Secord.

The panel expressed concerns about the literature review, and, on method, the reliance on a
single web site for data.

Broader Impacts

Strengths:

Dissemination to academic, media, and policy audiences. Training and mentoring of graduate
students.

Weaknesses:

Concerns were expressed about the limited broader impacts given the complications in third-
party access to data.

Data Management Plan: Conditions for third-party access are not given.

Panel Recommendation: Do Not Fund

REVIEWS

Proposal Number: 1224087
NSF Program: Sociology
Principal Investigator: Lundquist, Jennifer
Proposal Title: Mate selection in cyberspace: The intersection of race, gender and
sexual orientation
Rating: Excellent

REVIEW:

What is the intellectual merit of the proposed activity?

How does ethnic identity intersect with a host of other variables (gender, sexual orientation,
religion, nationality, etc.) in shaping partner selection and dating preferences? Data from one of
the largest dating websites in the US will be used to answer this question. The research will
access data for 4.2 million cases of invitations and responses posted on the website and analyze
the data using multi-level logistic regression. An initial pilot study is underway. This study is
grounded in partner selection theories (status exchange theory, homophily theory, utility
maximization theory) but stands to challenge/expand those theories and promises to further
enrich the intersectionality framework. It will look at racial hierarchies as they affect dating
patterns. It addresses dating patterns on a national level, but will also look at local racial
dynamics as they may vary by region. It includes romantic pairing and the partner selection
process for all ethnic groups. The author is well-qualified to conduct the study, with previous
NSF funding and publications in major sociological journals (American Sociological Review,
Social Problems, Social Forces).

What are the broader impacts of the proposed activity?

The popularity of internet dating has soared, with 74% of singles reporting they have used
websites. This study will add to our understanding of the role that race plays in shaping partner
selection among a very broad racially diverse sample of people of various sexual orientations,
religions, etc. This study is very well designed and piloted and, as the first interactional study to
analyze national level dating patterns (looking at initiation of and responses to dating requests)
and it is methodologically innovative. It is grounded in status exchange theory, but preliminary
findings from the pilot study are already contracting some of those assumptions.
Underperesented students will be involved in the research.

Summary Statement

This national level study of dating and partner selection is methodologically innovative and will
utilize data from a sample of millions of Americans. It will illuminate how race matters in
partner selection and how it intersects with a host of other variables.

REVIEW:

What is the intellectual merit of the proposed activity?

I [Material redacted per PAM Chapter XI G.2]recommended that it be funded because of the
valuable new insights it will provide into the patterning of racial preferences in relationship
formation. The [Material redacted per PAM Chapter XI G.2]proposal [Material redacted per
PAM Chapter XI G.2]clarifying the theoretical contributions of the project and demonstrating
the rich qualities of the dataset for testing the researchers' hypotheses. The dataset that the
research team will use to examine the structure of racial preferences in romantic relationships has
distinctive features that overcome important methodological limitations that undermined the
validity of previous investigations of this topic. For example, this study will examine relationship
initiation and reciprocation behaviors directly, and thus it is not compromised by the self-report
biases involved in survey research on this topic. The unique characteristics of this dataset also
provide the opportunity to examine new research questions. For example, to my knowledge this
will be the first study that uses behavioral data to examine the differing roles that racial
preferences may play in contact initiation versus contact reciprocation. Given its large sample the
study has ample power to investigate whether there are interesting differences in the patterns of
racial preferences in these two roles or whether an individual's gender or other identity
characteristics might interact with their role to predict the patterns of their racial preferences. I
believe that just this feature of the study alone will make it a groundbreaking examination of
racial preferences in relationships. This study also has the potential to test competing
explanations of racial patterns in relationship formation. For example, as the author points out,
this study will be able to isolate racial homogeneity preferences from propinquity as an
explanation for racial homogamy in relationship formation. This is an important contribution
because propinquity has often been a plausible alternative explanation of racial homogamy
patterns in other studies. Finally, the proposed study will test a number of important hypotheses
about interactions between gender, race, and sexual orientation that have potential to enrich
theories of identity intersectionality. The size of this dataset will provide the statistical power to
test the intersectional effects of these identity variables more reliably than they have been in
most previous investigations.

Given the unique strengths of the proposed study I believe it has the potential to be seen as the
most definitive investigation of racial preferences in relationship formation in the literature.
Research papers produced by this project are likely to be published in top-ranked
interdisciplinary journals and widely cited in the literature. The project's research findings are
also likely to be referenced in legal and policy documents.

In her past research the PI has successfully used quantitative analyses of other uniquely rich
datasets to produce new contributions to theories of racial stratification. Given this track record
and the well-planned description of the data analysis strategy for the proposed study it is clear
that the PI is highly qualified to successfully carry out the planned study. The [Material redacted
per PAM Chapter XI G.2]proposal have further sharpened the analytic plan and the pilot
analyses demonstrate the considerable strengths of this dataset for testing hypotheses. Also, the
efforts that the PI has made to provide other investigators access to this uniquely rich dataset will
be of considerable benefit to the broader research community.

What are the broader impacts of the proposed activity?

Racial homogamy in romantic relationships is an important factor contributing to the persistence
of racially segregated communities and racially stratified outcomes. The proposed project will
use a unique dataset to provide new insights into the nature and scope of this important social
problem. The proposed study is thus likely to be of considerable interest not only to behavioral
scientists but also to policymakers, social activists, community leaders, and the general public.

The proposed study has excellent potential to contribute to graduate student training. Graduate
research assistants will acquire valuable skills in analyzing networked data and managing a large
and complexly structured dataset. The statistical and research methods that graduate students will
acquire through their participation in this project will be transferable to other research projects.
Also, as potential co-authors of high profile research articles stemming from the project graduate
students are likely to acquire recognition within the research community and will be well-
positioned to apply for jobs as academic researchers or expert consultants. This project thus has
excellent potential to advance the professional development of graduate research assistants.

The proposed project will involve underrepresented groups in two important ways. First, the
project itself will study groups that are underrepresented in studies of romantic relationships.
Second, the PI will make efforts to recruit members of underrepresented groups for positions as
graduate research assistants.

In previous grant funded projects the PI has been effective in communicating research findings to
the broader public through major media outlets (e.g., Newsweek, NPR) and journals aimed at
popular audiences (e.g., Contexts). Given the high interest value and societal relevance of the
proposed study and the PI's dissemination plans, there is every reason to expect that the project's
results will gain widespread attention and influence. The proposed research is also likely to
receive widespread attention within the international scientific community and has the potential
to promote new interdisciplinary research collaborations.

Summary Statement

I believe that this project deserves the highest level of consideration for funding for several
reasons: 1) the project is conceptually and methodologically sophisticated, 2) the research
questions the project examines have important theoretical and practical implications, 3) the
project will develop and utilize a uniquely rich dataset that will allow the research team to
address longstanding research questions in a novel and more definitive way, 4) the PI has an
excellent track record of success in research projects of similar scope and complexity, 5) the
research findings will be of interest not only to other social and behavioral scientists but also to
policy-makers, and 6) graduate student assistants will acquire valuable, transferable skills in
research methodology and data analysis.

REVIEW:

What is the intellectual merit of the proposed activity?

The research proposed here is interesting. PI Lundquist proposes to study a dataset of millions of
records of profiles and contacts from an unidentified US web dating site. The PI already has the
data in hand, and has done some very interesting and provocative preliminary analyses. The PI
argues, very reasonably, that the data from the web dating site will allow an analysis of the
process of dating and mate selection, whereas most previous studies have had to rely on the
patterns of established couples to infer things about the mating and dating process. Thorough
analysis of the web dating data could provide important new insights about love, marriage,
sexuality, and race in America. The intellectual merit of the proposal is strong. Nonetheless,
there are aspects of this proposal that require improvement.
First, the kind of study the PI proposes has been done before. The proposal ought to cite Hitch,
G³nter J., Ali Hortaþsu, and Dan Ariely. 2010. "Matching and Sorting in Online Dating."
American Economic Review 100:130-163, because Hitch et al have already done many of the
kinds of analyses that the PI proposes to do here, with the same kind of data (profile and inter-
user contact data from a web dating site). Prior work in the area need not detract from the
intellectual merit of this project, but the PI needs to do a better job of explaining what is new.

The examination of both same-sex couples and heterosexual couples is a strength of the project.
Given the importance of the availability of data on sexual minorities to the intellectual merit of
the proposal, I was disappointed to see that the PI plans to exclude bisexuals from the analysis.
This is unfortunate because quite a lot of the people (especially the women) who end up with
same-sex partners have a bisexual orientation, or a fluid sexual orientation (see Diamond, Lisa
M. 2008. Sexual Fluidity: Understanding Women's Love and Desire. Cambridge, MA: Harvard
University Press). I think that if the project is going to make an important contribution to
scholarship about sexual minorities, the bisexuals need to be included (though of course how the
bisexuals would be included would depend on the analysis being performed).

In the same vein of questioning the PI's self-imposed data usage limitations, I am not convinced
that limiting analyses to people who live in the 20 largest metropolitan areas (as the PI proposes
to do) is the most sensible use of the data. While I understand the most people are looking for
partners in their local area, one of the unique characteristics of the Internet is the Internet's ability
to overcome the physical constraints of geography. If one is looking for a partner whose
characteristics are unusual enough (the hindi-speaking, mountain-climbing, banjo-playing
lawyer, for example), one might have to look beyond one's own metropolitan area to find such a
partner. Since finding unusual things that are beyond one's immediate view is one of the
strengths of the Internet, I think an Internet dating study ought to contemplate the ways in which
geography might and might not matter in Internet dating.

The intellectual merit of this proposal hinges, in part, on the PI's ability to use the unique aspects
of the online dating data to analyze the opportunity structure of the dating market for each
individual. The utility of the data for determining opportunity structure depends upon how
widely used and how representative the particular web dating site is of the dating market as a
whole. Representativity is particularly important because different groups tend to use different
dating websites. Black users may prefer blackplanet.com. Gay who men who are looking to hook
up are on adam4adam.com. The Jews have Jdate, and in fact every dating niche has web sites
that cater to that niche specifically. One of the ways that the Internet is unique is that the Internet
allows for niche marketing to niches that were never marketed to before. So the PI needs to
carefully explain how the data from the one web dating site at her disposal will properly
represent the dating market for not only white heterosexuals (who presumably predominate), but
all of the racial, ethnic, religious, and sexual minorities who have their own specific websites.
One way the PI could begin to make the case for representativity is simply to compare the profile
of users of in the web dating dataset to the known profile of adults in the US. How similar are the
two profiles? If you examine the Hitch et al paper, their first tables address this question of
representativity.

The PI has an impressive record of accomplishment, especially given the short time since her
PhD degree. The literature review is professional and thoughtful, and the preliminary data
analyses are promising and provocative. It is clear that the PI has the capacity to undertake the
work described, and I believe that the work will be useful and interesting to the general
sociological community. Nonetheless, the budget for this project seems to me to be too high. The
data are already at the PI's disposal, and are already being analyzed. It is not entirely clear why
the NSF ought to pay for 3 months of the PI's time or 2 years of full-time graduate student
support simply for further data analysis. A more modest budget would, in my view, be easier to
justify.

What are the broader impacts of the proposed activity?

Aside from PI Lundquist's mentoring of graduate students, which may be excellent, the most
important broader impact that this project could have would be the sharing of the data from the
dating website. These data are important. Other scholars would benefit by having access to these
data. Footnote 3 of page 6 of the proposal and the Data Management Plan both suggest that the
data could possibly be shared with other researchers under certain insufficiently defined
conditions, but the Data Management Plan lacks specific directions for how other researchers
will get access to the data. Note the following statement from the NSF's instructions regarding
Data Management Plans:

Investigators are expected to share with other researchers, at no more than incremental cost and
within a reasonable time, the primary data à created or gathered in the course of work under NSF
grants. (http://www.nsf.gov/pubs/policydocs/pappguide/nsf11001/aag_index.jsp From the Award
and Administration Guide, Section VI.D.4.b)

I realize that the online dating data may not have been created or gathered with NSF funding, but
I think the principle still holds: the PI should find a way to formalize her agreement with the data
producer, so that the data can be hosted by a third party and other users may, if they meet
reasonable conditions, gain access to the data.

It is not enough for the PI to share the results of her findings with other scholars through journal
publications. In order for the scholarly community to really benefit from this project, the PI must
institute a procedure through which other scholars can get access to the data. This procedure
would have to be cleared with the owners of the website who are the original source of the data.
One possibility would be to entrust ICPSR to hold the data, and allow other scholars to access
the data when they meet appropriate criteria (which could include the promise to share the results
of their work).

If the proposal were to include a Data Management Plan that included a concrete, timely, and
realistic plan for sharing the data with other scholars, then the project would have much broader
impact on the scholarly community, and the $200,000 budget would be easier to justify.

Summary Statement

An interesting and innovative proposal.

REVIEW:

What is the intellectual merit of the proposed activity?

The proposed research uses interesting data from an unnamed online dating service to answer
questions about romantic partnering and racial boundaries, namely, how do we understand
preferences and behaviors in dating as they relate the interracial and interethnic relationships?
Dr. Lundquist makes a compelling case that existing research focuses on interracial
relationships by focusing on marriage without adequate attention to interracial dating or dating
preference. She rightly points out that much existing research asks for self-reports of
preference, rather than capturing actual behaviors in dating, which may be more informative.
We know little also about who initiates contacts, which could be useful to understanding
interactional characteristics between individuals. Lundquist also quite convincingly suggests
that the focus on marriage has excluded gays and lesbians from examination and that her data
make analyses of same sex couples possible.

I am compelled that this dataset is a novel way to explore these questions. While I am not
qualified to speak to the specific models nested for analysis, I appreciate the design which aims
to tease out interactional effects of variables that are not racial. It is also impressive that we
may understand regional differences alongside other characteristics like income, education
level, parental status, or religion. It is not clear how pilot analyses of rural areas, which will
otherwise be excluded from the models, will be used or might be useful. Although I appreciate
the offer to analyze these regions to see if they look different than urban areas, this is not an
impressive approach. If they do behave different, one would think that would make for rich
opportunity. And if they do behave similarly, it remains unclear why the analytical decision
was made to only analyze the top 20 metropolitan areas.

To explain these possible interactional effects, the proposal describes the ways variables will be
controlled or nested, but sadly, describes little about the implications of these data themselves.
We are offered pilot analysis of contact initiation and contact reciprocation, with many caveats.
While these are interesting and potentially informative, it is not entirely clear how these
processes help us understand the likelihood of actual meetings between online daters in the
offline world or how we understand these gestures as good measures of romantic interest. What
do we know about the relationship between online contact and authentic interest, particularly
when only 2-5% of messages are returned? It would be useful to have a better sense of the
relative generalizability to offline research questions, particularly as the project is situated in
dialog with offline studies and theories.

Dr. Lundquist has a proven track record of innovative strategies for research design and is
clearly well-qualified to conduct this research. She appears to have a good record as a mentor
and will likely yield interesting results from this study. She has also secured unusual data that
should make for interesting analyses.

What are the broader impacts of the proposed activity?

We cannot know fully how the use of data from a reportedly large (but unnamed) online dating
website gets us closer to knowing the answers to how these issues matter in the offline world.
Lundquist acknowledges that this sample is not representative, but explains that use of such
sites is increasing as people search for partners. (Even the compelling statistic that 74% of
internet users have tried a dating site does not explain how many actually use the internet and
what patterns of use look like for whom.) Lundquist promises a test of representativeness, but
acknowledges that this group likely slants towards younger and better educated.

The inclusion of sexual orientation is a significant contribution to a literature that is heavily
heterocentric. The use of mixed race individuals as well as smaller racial and ethnic groups is
also important, once they are broken down from the larger categories in the initial analysis.
This research also does not take whites as the center of interracial relationships to ask the trite
question of when and how white marry out. In fact, this proposal potentially tells us a great deal
about interracial relationshipsùin dialog with other characteristicsùacross and between many
groups. This is a major potential impact.

The data are stripped of identifying information and it appears the data were purchased. Other
than a promise to conceal the company's name, it is not entirely clear whether Dr. Lundquist
owes the corporation who gave her these data any power to review findings or has an
obligation to share information with them. Agreements with this companyùvague in description
hereùclearly affect the possibility of sharing or depositing data in the future. But more
immediately, they raise questions about how this research might impact the social worlds
studied.

This is a well-written proposal that asks important questions. The dissemination planùtop
sociology journals and three conferencesùseems fairly straightforward, albeit unimaginative.
However, this is a well-qualified researcher and the study will likely yield interesting results.

Summary Statement

REVIEW:

What is the intellectual merit of the proposed activity?

First, let me say that I was very impressed with this proposal. This is an area that is vastly
unexplored and this project will fill that hole.

Knowledge about interracial and interethnic relationships is heavily dependent upon Census
data which is fraught with weaknesses. This project extends to dating across racial and ethnic
lines, which is so important because it gets at the origin of mixed relationships. There is little
research that even addresses dating among mixed relationships. The proposal is also
ambitious in that it seeks to understand these same dynamics in gay and lesbian relationships.
Again, there is a dearth of literature in this area. I am especially impressed with the dataset
and design of the study given how important and commonplace online dating has become.

Jennifer Lundquist is highly qualified to do this research and I have no doubt that her and her
team will do an excellent job with this research endeavor.

In my opinion, this research will literally transform the way we think about "intermarriage."
First, this project distinguishs between interracial and interethnic relationships which is
crucial. If this is all it did it would be worthy of the grant. Second, it distinguish between
homosexual and heterosexual relationships to see if there are differences between the rates of
mixed relationships. Finally, and most important, the researchers will be able to measure
preferences, opportunities (both in terms of numbers and geographic regions), and behaviors
in regards to mixed relationships. Most studies are only able to measure only one aspect (two
at best), so to be able to distinguish between these vital aspects of a relationship will help us
to know what is driving mixed relationships. I am greatly anticipating the results of this
research!

The proposal is well conceived and organized. The researchers have ample access to
resources to complete this project.

What are the broader impacts of the proposed activity?

This project is well poised to address key issues relating to underrepresented groups. The
researcher addresses race, ethnicity, religion, gender, and sexual orientation in the delicate
area of intimate relationships. This helps us to better understand how these underrepresented
groups are gaining a foothold in mainstream America and how mainstream America is
changing as a result of these mixed relationships.

Making this dataset available to the public will be a great service to other researchers (such as
myself) who study these issues. I can imagine dozens and dozens of research reports coming
from this dataset once it is available. What a unique opportunity to gain access to such a
difficult-to-access dataset.

In sum, I think the results of this study will be crucial to understanding how the United States
of America is changing through the mixing of various racial and ethnic groups through
intimate relationships. This helps us to understand how minority groups are incorporating into
our society and will help us to better understand how immigration is transforming our society.

Summary Statement

As with most projects or prosals, there is room for improvement. I have several suggestions to
improve the proposal, and it is for this reason that I did not rate the proposal as "Excellent"
although I think this prosal should absolutely be funded. My suggestions are therefore just
intended to help the researchers to improve this research.

First, I feel like relgion could be a more important aspect of this study. Few sudies address
religion in terms of mixed relationship precisely because this information is not available in
Census data (which is mostly used for studies on intermarriage) and relgion has the potential
to be a critical line drawn in mixed relationships. As currently proposed, religion is conflated
with ethnicity and while certain religions overlap with ethnic groups (white and Jewish),
others do not. In other words, I think it would be helpful to think of religion as an indepent
variable in addition to a dependent variable.

Second, is there any way to break down the Hispanic/Latino category into specific national-
origin groups such as Mexican, Puerto Rican, Cuban, etc? This would greatly enhance the
ethnic breakdown as you are able to break down the Asian group into various ethnic groups
based on language. In addition, I do agree that language is a decent proxy for acculturation,
but care needs to be taken with the assumption that those that speak Spanish are first
generaiton, while those that do not are second generation. While this assumption generally
holds for Asian ethnic groups, it is much more problematic for Latinos, esepcially Mexicans
who are third or fourth generation, not to mention a greater number of second generation
Latinos who are bilingual.

Finally, I question whether the assumptions about intial messages (first full paragraph on page
8) are justified. The assumption that initial messages and responses are meaningful is a big
assumption to make and in my views makes or breaks this research. In other words, if this
assumption proves to be incorrect it has the potential invalidate much of the results. This
applies to all of the assumptions mentioned in the first four sentences of this above-mentioned
paragraph. I am not saying that I think these assumptions are incorrect, rather I think it is
important to dig a little deeper into these assumptions. In fact, isn't it possible to test these
assumptions within the dataset? I realize that the researchers do not have access to the actual
messages, but it would be easy enough to go onto the website and examine the messages in
sufficient detail to determine whether or not the above assumptions are valid. In other words,
a minor qualitative componenet of the messages for this particular website would add a
crucial validity check for this proposal.

What is the intellectual merit of the proposed activity?

[Material redacted per PAM Chapter XI G.2][Material redacted per PAM Chapter XI G.2]"The
PI needs to work considerably harder on developing a driver for this study that has implications
for a broader literature and future work on race, mate selection, desegregation process,
submerged bias, etc. in order to warrant this level of support for the project."

In the initial reading of the project summary, one does not find that this problem has been
rectified. A series of empirical findings from the pilot study are listed, but there really is no
mention of a theoretical agenda here whatsoever. There is a big statement about "the meaning of
race in contemporary romantic interaction" but we are [Material redacted per PAM Chapter XI
G.2]lacking a theoretical contribution that would cause interest by others who work in related
areas. The intellectual merits of the proposal focus solely on methodological and some empirical
elements.

[Material redacted per PAM Chapter XI G.2]the fact that the core critique is left unanswered
does not recommend support. The project needs a theoretical driver. Not just a series of
empirical questions and potential broader impacts. There are certainly some theoretical issues at
stake in this general terrain, and the author bring some of those up in the literature review, but
those questions are not specifically examined in the actual study. As an example I point to the
issue of status exchange among interracial marriage partners.

I do think the methodological contribution here is real, and does get more directly at the kinds of
behaviors that are at stake. However, that's not enough all by itself. It's true that that's not all
that's there, and that there are some theoretical ideas expressed in the proposal. But it is not clear
to me that these are terribly compelling or applicable to other domains. Other reviewers in the
panels may see the importance of these questions differently.

On the plus side, the investigator has done a great deal of pilot work with the data. On the
negative side of that same coin, it is not clear to me why exactly NSF would fund this proposal
given that it seems to me that it can be and will be completed as a part of the PI's normal research
routine.

What are the broader impacts of the proposed activity?

The data management plan seems to me to be inadequate. The problem is that the data itself is
proprietary. It sounds like other researchers may be able to access the data with permission from
the company that owns them. However, it's not clear what the conditions of permission would
be, or how likely the company would be to allow others to use the data. If NSF is going to
provide support for this project, the data needs to be more widely available than that. Or, at least
the process under which another researcher could access the data must be spelled out in much
greater detail.

Summary Statement

REVIEW:

What is the intellectual merit of the proposed activity?

1. The proposed study will investigate racial/ethnic preferences in dating partners, and by
implication, group boundaries along racial/ethnic lines. The dataset, which came from an online
dating site, features messages to initiate contact and any responses to those messages. The
dataset has data on straight, gay, and lesbian members.

2. The proposed study has several strengths.
ò The idea is excellent.
ò The large dataset affords a unique opportunity to investigate racial preferences in romantic
partners among multiple racial/ethnic groups.
ò The PI has thought carefully about the data analysis.
ò The PI is well qualified to conduct the research.
ò In addition, she has recruited a statistician at her university as a consultant to assist on the
project. Both have access to necessary resources.

3. The proposal needs to be organized around the research questions, and the questions
themselves need to be clearly stated. Based on the discussion on p. 2, I thought the project was
going to test exchange theory and/or gendered racial formation theory. However, status
exchange theory is not mentioned again in the proposal. Gendered racial formation theory is
mentioned on p. 5, under the heading "Gender, Sexual Orientation, and Local Contexts?" I am
not sure what research question #3 is.

4. The proposal confuses propinquity and opportunity. Research has found that people select
partners who live nearby (propinquity). That is different from opportunity, which relates to the
supply of potential partners.

5. A serious, related problem is that the proposal ignores previous work on how marriage/partner
markets influence marriage patterns, which also would influence interracial dating. Two
important elements are group size and sex ratios (more generally, mate availability). The PI
speculates that group size influences interracial dating among Asians (pp. 5-6), but cites no
literature. Look at Hwang et al. (1997) and see who has cited it.

Theory going back to Guttentag and Secord's book Too Many Women? suggests ways that sex
ratios may influence a host of outcomes pertinent to the present study. In this case, think about
mate availability more generally. It can be measured by partner availability on the dating site (the
study will include only messages exchanged from partners in the same metro area) and in the
metro area. (The PI plans to compare characteristics of online daters with those of the general
population [p. 12]. Include measures of group size and mate availability for the metro areas.)

6. The proposed analyses, too, should link to the research questions. The research questions
discussed on pp. 4-6 do not have clear parallels with the Analytical Strategy.

7. The list of references cited is short, incomplete, and not current. For example, it does not
include Qian and Lichter's (2011) recent article on interracial marriage or the work of
demographer Gary Gates. Schwartz and Graf (2009) have looked at assortative mating among
same-sex and different-sex couples. Also see the recent review of online dating research by
Finkel et al.ùa must-read, as it discusses partner preferences.

8. The proposal mentions incorporating users' ratings of other users into the analyses. These
ratings are voluntary. How much missing data is there? Did enough users rate other users to
make this measure useful? I didn't understand the description of this measure. "If one received
fewer ratings, we will impute the average of his or her gender orientation group as their
attractiveness weighted by their own rankings.5" Their own rankings of whom?

What are the broader impacts of the proposed activity?

The results of the study will be broadly disseminated to academic audiences via conference
presentations and articles, and also to media and policy outlets. The project will train and mentor
a graduate research assistant, and the PI will make extra effort to recruit a student from an
underrepresented group.

Data Management Plan (DMP):
The DMP satisfactorily addresses the criteria listed in the reviewer guidelines. In the dataset
provided to the PI by the online dating site, users are anonymous. The online dating site granted
access to the dataset on the condition that PI does not identify the site. The PI will maintain
security of the data. Other researchers potentially can collaborate with the PI on projects using
the data if they obtain written permission from the online dating site.

Summary Statement

The proposed research takes a creative approach (using data from online dating sites) to studying
patterns of interracial dating. However, the research questions need to be informed by previous
theory and research, clearly stated, and linked to the research design. The PI is qualified to do the
research and has sufficient access to resources. The findings would be broadly disseminated, and
the project would train and mentor a graduate student, likely from an underrepresented group.
My rating is based more on the intellectual merit than on the broader impacts.