Edanz Expert Scientific Review Report
Prepared by: Sample
G0000-0000-Sample
Introduction
The introduction provides a good, generalized background of the topic that quickly gives the reader an
appreciation of the wide range of applications for this technology. However, to make the introduction
Does the introduction provide sufficient
background information for readers not in
the immediate field to understand the
problem/hypotheses?
more substantial, the author may wish to provide several references to substantiate the claim made in
the first sentence (that is, provide references to other groups who do or have done research in this
area). The second sentence helpfully explains the motivation for the research to current and potential
funding agencies. However, to make the motivation clearer and to differentiate the paper some more
from other applied papers, the author may wish to provide another sentence giving examples of some
of the applications of this technology, along with appropriate references.
I think the motivations for this study need to be made clearer. In particular, the connection between (a)
the requirements of high resolution and accuracy for metrology applications, and (b) the necessity of
choosing an appropriate reconstruction technique, could be clearer. One way to demonstrate this
connection would be to cite references (if possible) that demonstrate that inappropriate reconstruction
Are the reasons for performing the study
clearly defined?
techniques can lead to inferior results.
Furthermore, after stating that the choice of reconstruction technique is important, the author offers no
explanation of why he chooses the ASM for image reconstruction in the present work. I think the
motivation for the present research would be clearer if the author could provide a more direct link
between the importance of choosing an appropriate reconstruction method and sectional image
reconstruction.
The objective is clearly defined in the last sentence of the second paragraph. However, I feel this
sentence could be modified to something like “In this publication, we show that the selective numerical
Are the study objectives clearly defined?
reconstruction method is advantageous in 3D microscopy and tomographic imaging using digital
holography.”
Methods/technical rigor
The experimental apparatus is quite standard, and is appropriate for the study, especially given that
Are the methods used appropriate to the
the main focus of the paper is not to develop a novel holographic technique, but to demonstrate the
aims of the study?
power of sectional image reconstruction.
Is sufficient information provided for a
capable researcher to reproduce the
experiments described?
Yes, although the author should probably provide more information about the spatial filter. Also, the
author may wish to mention in this section the advantage that no scanning is necessary with this
method (as opposed to scanning optical holography).
I don’t think any additional experiments are necessary to validate the results presented here, because
the results themselves are not what is important; it is the technique used to obtain these results that is
Are any additional experiments required to
important. One exception to this reasoning would be if the author could demonstrate that the results
validate the results of those that were
obtained using the present method are consistent with results obtained using a different technique.
performed?
I don’t think this is vital to the present paper (especially given the length limitations on the paper), but
it may be something that would be helpful in a longer, more detailed paper.
Are there any additional experiments that
would greatly improve the quality of this
paper?
Are appropriate references cited where
previously established methods are used?
To clearly show the advantages of the sectional reconstruction technique, would it be possible to show
images at the focus plane with and without the out-of-focus elements (that is, with and without the
numerical filtering)? This may make the advantages of the numerical filtering more obvious.
There are several instances where assertions are made that are not substantiated with references.
These have been noted in the appropriate sections of this report.
Results/statistics
Because the ASM is explained in other papers, and because the aim of the present paper is not to
Are the results clearly explained and
presented in an appropriate format?
develop a novel reconstruction technique, the author may wish to reduce the explanation of the ASM
in Section 2, and instead, provide a more in-depth discussion of the sectional images shown in
Section 3.
It seems to me that Figure 3 is not vital to the discussion presented in the paper. It is only cited in two
Do the figures and tables show essential
sentences, and in neither of those are the results that it presents discussed. I would suggest removing
data or are there any that could easily be
it and focusing the discussion on Figs. 4 and 5, which seem to me to be more relevant to the topic of
summarized in the text?
the paper (sectional reconstruction).
Are any of the data duplicated in the
graphics and/or text?
No data is duplicated, but the data in Fig. 3 does not seem vital to the paper.
Two panels in Fig. 4 have the same labels (d). Presumably this is an error. Also, it may be helpful to
Are the figures and tables easy to interpret?
Are there any additional graphics that would
add clarity to the text?
Have appropriate statistical methods been
the reader to show the coordinate z on Fig. 2.
I do not think any additional graphics are necessary. However, as noted above, I think Fig. 3 is
unnecessary and could be removed, which would allow a more in-depth discussion of Figs. 4 and 5.
This is not relevant for the present paper.
used to test the significance of the results?
Discussion
As suggested above, I think a more in-depth discussion of Figs. 4 and 5 would be helpful. I feel this is
an important result for this paper, and therefore it merits more discussion. Why is the 3D perspective
insufficient for analysis? Can the author demonstrate that the resolution of certain small particles is
less than required for a given application? More importantly; can the author show that the resolution in
Are all possible interpretations of the data
the focus plane is sufficient for analysis (which is not obvious from the images)?
considered or are there alternative
hypotheses that are consistent with the
The author may also wish to give a more detailed discussion of Fig. 5. Can the author show an out-of-
available data?
focus plane and 3D image for this specimen (this image is easier to interpret than the images of small
particles, for which it is hard to differentiate by eye the difference between focused and out-of-focus
planes)? Can the author demonstrate what biologically relevant information he can get from the infocus image that is not possible to get from the out-of-focus image?
The author may wish to mention why it is important to image small particle fields to explain the
Are the findings properly described in the
motivation for his choice of specimens, and accompany this with some references to other studies that
context of the published literature?
demonstrate this importance.
Are the limitations of the study discussed? If
not, what are the major limitations that
should be discussed?
No significant limitations are discussed. It may be worthwhile to mention the tradeoffs involved in
choosing the ASM as opposed to some other method. This may be done in Section 2 after describing
the advantages of ASM.
Conclusions
Are the conclusions of the study supported
by appropriate evidence or are the claims
exaggerated?
The conclusion says that the digital holographic method has potential biomedical imaging applications
in 3D microscopy; however, the discussion of Figs. 4 and 5 does not make this point obvious. This
conclusion would be much stronger were such a discussion provided (see Discussion section, above).
Literature cited (introduction, results, discussion)
Is the literature cited balanced or are there
important studies not cited, or other studies
disproportionately cited?
Please identify statements that are missing
any citations, or that have an insufficient
number of citations, given the strength of the
claim made.
The literature cited is relevant to the study, but there are several instances, which have been noted
above, in which the author makes assertions without substantiating them with references.
I have noted these in the sections above (see, in particular, the section on the introduction).
Significance and Novelty
As it stands, I am not sure that the results will be judged novel or important enough for publication in
Are the claims in the paper sufficiently novel
to warrant publication?
Applied Optics. However, if the author provides a more detailed analysis, in particular of Figs. 4 and 5,
I think the paper could prove to be very interesting and useful to a very large audience, possibly
making it acceptable for publication in Applied Optics.
The paper provides an excellent technique for digital holographic image reconstruction because the
Does the study represent a conceptual
technique is relatively simple to implement and yet it is quite powerful. Also, being simple, it can be
advance over previously published work?
reproduced by many who are not necessarily specialists in optics.
Journal Selection
The four major topics covered by Applied Optics (the target journal) are optical technology, information
processing, photonic devices, and biomedical optics. The manuscript deals with digital reconstruction
of holographic images, which is a technique that can be applied to biological samples and to
Is the target journal (if known) appropriate? If
not, why not?
metrology. As such, the manuscript’s topic relates to the first two and the last of the major topics of
Applied Optics (and one could also argue that it relates to the third topic). Therefore, I believe the
target journal is an appropriate forum for this article. In addition, Applied Optics is a popular and
respected journal, so publication in this journal will ensure a wide readership and good exposure for
the author’s work.
What is the likely target audience of this
Because the article discusses a method to perfect digital holographic image reconstruction, I expect it
paper? Please comment on the specific field
will draw interest from researchers who want to apply holography to their research (as opposed to
(e.g., diabetes, neurology) and activity (e.g.,
researchers doing fundamental research in optics). Therefore, I expect that the likely audience for this
article will be researchers in diverse fields (including biology, materials science, nanotechnology,
clinician, researcher).
physics, mechanical engineering, image processing…) who are looking for enabling techniques that
will allow them to progress their research.
Minor comments
Please refer to the comments in the edited manuscript file for minor comments.
Major comments
•
Regarding the figures: I recommend removing Fig. 3, shortening Section 2, and devoting the
additional space to a more detailed discussion of Figs. 4 and 5. In particular, the author may
wish to point out in-focus and out-of-focus spots in Fig. 4 (e.g., with arrows or labels), and to
provide a more thorough analysis of these features. As it stands, only the out-of-focus spots in
Fig 4(a) have earned a mention in text. It may be helpful (to show the focusing properties) to
show a 2D cross section of one or several spots from Figs. 4(a), 4(c), and 4(d), so the reader
To publish this paper in your target journal,
the following revisions are strongly advised:
can appreciate why the choice of image reconstruction plane is important.
•
I recommend strengthening and clarifying the introduction, as detailed above.
•
Because the author mentions in the abstract (and especially in the last, or concluding,
sentence) that the object-to-hologram distance can be quite small, I expected this to be a
major theme in the paper. However, very little mention of this phenomenon is made. If this is a
significant advantage of this holographic imaging technique, it should be discussed in more
detail in the paper. It could be mentioned, for example, in Section 3.
Sample Review by Macro Editor
This paper focuses on entrepreneurship of multinational companies’ (MNCs) subsidiaries
and examines the impact of corporate and local environmental contexts. This topic is
timely and important considering the increasing importance of overseas subsidiaries in
MNCs’ strategy and growth. The paper is well written and easy to follow. The existing
literature is well cited. I also applaud that the author/s measured the dependent variable
and the independent variables in different time periods, which helps address the causality
issue. Below I will discuss my main comments and suggestions, which hopefully can
help the author/s improve the study.
Theoretical Issues:
1. The role of entrepreneurship in MNC subsidiaries. This paper is based upon the
assumption that subsidiary entrepreneurship is good for MNCs and addresses the
question of what factors can affect subsidiary entrepreneurship. While I concur that
subsidiary entrepreneurship is becoming increasingly important for MNCs, not all
subsidiaries need to be entrepreneurial. It depends upon the MNC’s strategic objectives in
a particular subsidiary. For some subsidiaries, being innovative, risk-taking and proactive
are key to success while in others implementing headquarter strategy as directed may be
the most important. In other words, while examining subsidiary entrepreneurship, one
needs to consider the subsidiary’s role (or mandate) in the MNC system. This is probably
the key difference between entrepreneurship of a subsidiary and entrepreneurship of a
standalone firm. Of course, the paper examines the impact of subsidiary mandate (global
subsidiary mandate) on subsidiary entrepreneurship—which will be addressed later. But
my point is that being entrepreneurial or not (or the degree of entrepreneurship) may be
an important part of a subsidiary’s mandate instead of an outcome of its mandate.
2. The impact of local environmental contexts. The paper examines the effects of three
local environmental dimensions: dynamism, hostility, and complexity (H5-H7). The logic
leading to these hypotheses is clear. However, how are the effects of environment on
subsidiary entrepreneurship different from the effects of environment on entrepreneurship
as examined in previous studies (e.g., Zahra, 1991, 1993; Zahra and Covin, 1995)? In
other words, how does the context of MNC subsidiaries contribute to our understanding
of the relationship between environment and entrepreneurship? To make stronger
contributions rather than replicating previous studies in the context of MNC subsidiaries,
the author/s needs to more closely integrate the context of MNC subsidiaries into the
arguments. Do local environmental contexts shape the subsidiary’s mandate or do they
directly affect the subsidiary’s entrepreneurship, regardless its mandate?
3. The impact of corporate contexts. The impact of corporate contexts, or the
headquarter-subsidiary relationships, are unique to subsidiaries and thus has great
potentials to add new knowledge to our understanding of entrepreneurship. This study
examines four corporate contexts: global subsidiary mandate, autonomy, strategic
control, and financial control (H1-H4). As noted in my comment #1, the level of
entrepreneurship may be a part of a subsidiary’s mandate rather than an outcome of its
1
mandate. A similar argument can be made regarding the relationship between autonomy
and subsidiary entrepreneurship. For a subsidiary whose mandate includes a high level of
entrepreneurship, it needs a certain level of autonomy. In summary, a subsidiary’s global
mandate, autonomy, and entrepreneurship may all depend upon the MNC headquarters’
strategic objectives in the subsidiary. Therefore, even though we observe significant
relationships among them, it does not necessarily mean that there are causal relationships
between them.
Empirical Issues:
4. Overlaps in key measures. All the measures used in this study have been validated by
previous studies. However, when put together, there are overlaps between the key
measures. Specifically, innovation and new product introduction were included in the
measures of global subsidiary mandate, autonomy, and subsidiary entrepreneurship (see
the appendix on pages 45-46). The overlaps, which may partially drive the significant
relationships between them, further reinforce my concerns on the theoretical relationships
between them (as indicated in my comments #1 and #3). The author/s has tried to address
this issue by measuring the independent variable and the dependent variable (subsidiary
entrepreneurship) in two different time periods. Again, I applaud this effort. But
considering organizational inertia, there may be considerable persistence in these
dimensions over the three-year time period. It is not clear in the paper whether subsidiary
entrepreneurship was measured in both 1995 and 1999. If it was measured in both time
periods, one solution is to control for the prior level of subsidiary entrepreneurship or to
use the difference in subsidiary entrepreneurship as the dependent variable in order to
better capture the impact of environmental and corporate contexts on subsidiary
entrepreneurship.
5. The inter-rater reliability for the environmental variables is relatively low (page 23). In
supplementary analyses, measures from the second respondents (even with a reduced
sample size) may be used. If these analyses produce similar results, the concern of the
low IRR may be reduced.
6. The effect of strategic control. In Table 2 (page 42), the correlation of strategic control
and subsidiary entrepreneurship is not significant (r = 0.12, n.s.). However, in Table 3
(page 43), the coefficient of strategic control is highly significant (r = 0.35 in Model 2
and r = 0.38 in Model 3, p < 0.001). Is it because of multicollearnarity between the
independent variables? If other independent variables are not included, is strategic control
still significant?
Overall, I think that MNC subsidiary entrepreneurship is an important and interesting
topic. The author/s has clearly devoted a great deal of time and effort in this study. I hope
that these comments can help improve this study.
2
Sample Review by Macro Editor
The manuscript addresses drivers of entrepreneurship in multinational subsidiaries,
particularly taking into account how attributes of the environmental context and firm strategy
influence entrepreneurial behavior.
The topic itself is of great importance, especially because multinational firms are now
encouraging such dispersed rather than centralized entrepreneurship to allow organizational
adaptation and sourcing of new opportunities for the global market. Having professed my
general enthusiasm for the topic and its importance, I have some concerns that I feel are
tractable but require substantive effort.
Below are the major issues that I would encourage you to consider in further improving this
manuscript.
1. The framing of a global subsidiary mandate (GSM) is interesting, but appears poorly
grounded in a theoretical sense. I agree with you that the GSM gives the subsidiary greater
prominence in its activities. What I found as a gap in logic is how this translates into
entrepreneurial behavior. For instance, if a subsidiary has a mandate to develop
advancements in washing machines (for example), we would anticipate more technological
investment in that area. How does this translate to entrepreneurial behavior? Is it the
underlying mechanisms such as network centrality that provide it access to global resources?
Is it power to influence or lead decisions and investments? You will need to be more specific
on the causal mechanisms.
2. Similarly, autonomy seems disjointed in a theoretical framework. I agree with you fully
that autonomy encourages risk taking and entrepreneurial behaviors as you rightly
acknowledge. I need some scaffolding where the logic connects from the prior hypothesis.
As it stands, it feels like GSM, Autonomy, and Controls don’t really sit well together. There
isn’t an overarching sense of framework that holds them together.
3. My suggestion is for you to consider Control, Autonomy, and GSM have in common or
share. For example, can you potentially argue that entrepreneurship itself is suppressed in
corporate contexts because of structural design, and that some elements of organizational
design – let’s say, modularity -- has attributes that could create a better environment. In
which case, you could argue that your corporate predictors are really structural /
organizational design attributes? This is only a suggestion; you are much closer to the data /
logic and can come up with better arguments. My point is simple – you need a more coherent
theoretical logic for the choice of hypotheses/variables.
4. For the environmental context issues, I thought that they fit well with what was
theoretically expected. As you can see from my points 1 and 2, you may want to more tightly
connect with the theoretical logic / causal mechanisms.
5. I appreciated the significant effort in the sampling and the survey data collection waves –
this is not often done, and is commendable. The data analysis though only takes into account
the final sample of 227 firms. This could create a sampling bias and distort your analysis.
There are two suggestions that I would expect that you consider in improving the paper.
First, the sample attrition from 2743 subsidiaries to 581 respondents to the first wave causes
some concern. I wonder if you could do some kind of selection model as a two stage
econometric model? Do you have access to secondary data on the entire population? For
example, if you knew whether some subsidiaries had a GSM while others didn’t, then that
could be a good selection measure. I know that this is quite a bit of work, but just simple ttests for non-response seems a watered down way of addressing this potential issue.
Second, the data attrition from 581 firms in the first wave and 227 in the second wave with
measures of the Dependent Variable is a concern. Here, I think more could certainly be done
along the lines of a selection model. Were there other variables that were left out from the
study? Do you have secondary data sources that could be used here?
6. I found the second inter-rater reliability survey an excellent way to bolster your constructs.
You don’t seem to report the results; I would encourage you to do so. This allows the reader
to infer that your study design was more thoughtful than a simple survey with biases of crosssectional data such as validity and reliability issues.
7. I found the discussion section could be improved a lot more. First, I would encourage you
to consider effect sizes on the outcomes. This allows you to say that when GSM increases by
one standard deviation from the mean, entrepreneurship increases 8% (for instance). This is a
powerful way to describe the importance of your findings. Second, I would focus on the
theoretical contribution a lot more. The underlying causal mechanisms that I suggested in
points 1 and 2 could be an anchor to make this an interesting discussion. For instance, you
can focus on contributions to organization design and governance in multinationals. In
addition, you can enliven your discussion of what GSM means in theory and practice. How
does it work, or show up as outcomes?
Minor Points
8. Interactions – Your theory is agnostic on how the variables interact. This may not be
necessary, but something that could add nicely to your study. This could be a major point
that could change your story as well. Please consider how you can enrich your analysis and
the theoretical framing if you include interactions.
9. Table 3 – For your regression results, please report the standard errors.
10. Your discussion could include some contribution to the international business literature in
addition to the entrepreneurship literature.
Overall, I see potential for this study to contribute to the literature. The topic and the data are
interesting. The theorizing can be improved. The major work appears to be in the data
attrition and data analysis. If you could bolster the data, it would add nicely to existing
studies in this space.
Good luck!
Sample Review by Macro Editor
Thank you for the opportunity to read your study in which the determinants of international
subsidiary entrepreneurship are explored. Your work draws attention to the corporate and local
environments, and through analysis of survey data, offers insight into the effects of these
variables. Briefly, your findings suggest both the corporate mandate under which the subsidiary
operates, as well as systems put in place to effect that charge shape entrepreneurial tendencies.
So, too, do the underlying complexity and dynamism of the local environment.
There are several notable aspects to this investigation, including its international focus. Your
study not only reminds us that entrepreneurship occurs globally, but also offers important
insights of practical significance to multinational firms and their managers. Equally laudable is
the joint focus on firm and environmental-level factors. Clearly, both play an important role in
strategic outcomes, and your study’s simultaneous exploration affords a richer understanding of
entrepreneurial activity than would result from isolated examination. Finally, the considerable
care and attention devoted to survey design (pilot-testing, two-stage administration, etc.) are
impressive.
At the same time, a few questions arose in my reading of the current draft:
1. An overarching concern relates to theoretical contribution. As noted earlier, your work
demonstrates that both organizational and environmental factors affect entrepreneurial activity.
While this is an important finding, it is somewhat unoriginal. Namely, extant work (much of
which is cited in your manuscript) has ably demonstrated that these variables constitute
significant determinants of entrepreneurial activity. Can you tell us then how your work extends
established findings? Apart from original context (i.e., entrepreneurship within multinational
subsidiaries), what new insights are uncovered into the processes surrounding, and practice of,
entrepreneurship more generally?
2. I’m curious, and suspect other readers will be, as well, as to how you selected this particular
set of organizational and environmental factors. For example, with regard to environment, your
model focuses on dynamism and complexity while other potentially relevant factors such as
growth or munificence are excluded. Also, given your study’s international perspective, one
might expect that local resources, government incentives, etc. would be explored (e.g.
Birkinshaw, Strategic Management Journal, 1997; Porter, 1990, The Competitive Advantage of
Nations, New York, NY: The Free Press). The point here is not to suggest that every
conceivable environmental (or organizational) dimension be included -- not by any means.
Rather it is to say that the current draft lacks a compelling rationale for the specific variables
examined. Ideally, a conceptual model that cogently integrates the complete set would be
presented.
3. Some logical inconsistencies are also evidenced in the current draft. Perhaps the most pressing
surrounds global subsidiary mandate (H1). As stated in your theoretical section, subsidiaries
operating under a global charge are tightly integrated in the firm’s operating network.
Concomitantly, they tend to endure stronger and more direct control by headquarters (a finding
confirmed by your analyses). It is unclear, then, how these same units come to enjoy the
discretion and autonomy essential to entrepreneurial activity (H2). For that matter, many of the
characteristics and attributes attached in the manuscript to subsidiaries under global mandate are
more commonly associated with subsidiaries under local mandates (e.g., strong ties to local
customers and suppliers, local manufacture and marketing, strong sensitivity to local
environmental conditions). In sum, plainer, more systematic conceptualization of the global
subsidiary mandate would help, as would a pass-through the manuscript aimed at ensuring apt,
consistent treatment.
Similar concern extends to entrepreneurial activity. In the initial pages of the current draft,
entrepreneurship is defined as encompassing risk-taking, innovation, and proactiveness. In
subsequent passages, this emphasis is exchanged for one centered on (strategic) adaptation.
Unfortunately, the two are not synonymous – a point underscored by your own arguments.
Namely, in developing hypotheses specific to environmental factors, a reactive response is
argued -- not the proactive, innovative one suggested at the outset. Again, a pass-through aimed
at consistency would be helpful.
4. As noted earlier, the time and attention devoted to survey design, pre-testing, etc. are
admirable. A question arises, however, as to the fit of particular constructs with your theoretical
model. Namely, subsidiaries are based in the U.S. and environmental attributes are specific to the
U.S. context. Per description, however, these subsidiaries are foreign-owned and part of a large,
international network of operations. Based on a reading of the current draft, it’s unclear how it
was ascertained that reported entrepreneurial activity was in fact initiated in the U.S. – i.e.,
directly linked to conditions resident within the U.S. environment (rather than other markets or
territories in which the subsidiary operates).
A question also arises as to the levels at which variables are gauged. The subsidiary is the focal
unit. Yet performance, for instance, is measured relative to industry norms, presumably at the
firm level. (This inference is based in part on statements that financial information was often
lacking at the subsidiary level.) While firm-level factors are important (again, central to your
thesis), these constructions/levels appear at odds with your arguments.
5. Please provide additional background regarding your sample. What industries are included?
How many firms? Were individual firms repeatedly sampled – i.e., several subsidiaries of the
same firm(s) included? What efforts, if any, were taken to verify respondent’s title/knowledge of
the firm’s strategy and performance?
6. Your reporting of inter-rater reliability is appreciated. Can you also provide background on
tests of discriminant validity (e.g., results of factor analysis, preferably confirmatory)?
7. For the most part, the results for your control variables prove non-significant. This is
surprising in light of the large volume of research showing these variables tend to be significant
predictors of entrepreneurial activity. Can you speak to these findings? What explains the general
pattern of non-significance for previously-established determinants?
8. Another strength of your study is the array of headquarter countries captured. Your sample
subsidiaries/managers are affiliated with firms based in a variety of locations and cultures –
several of which prove significant predictors in hypothesis tests. Have you considered exploring
these findings further? It strikes me you have an opportunity to uncover whether entrepreneurial
advantage arises from home country factors – culture and values, expertise, or perhaps some
other factor.
In closing, again, I appreciate the opportunity to have read your work. I look forward to seeing
more of it in the future. Best of luck!
Sample Review by Macro Editor
General Comments
I think you have the potential to tell a really interesting story here. You have interesting data that
- with one major exception I will get to - was collected pretty rigorously, on a population that's
somewhat difficult to study. You are also considering a topic that has the potential to offer some
novel theoretical insights. However, at the moment I don't think your study makes a particularly
significant theoretical contribution, although it could. I also think you face a major issue in that
your theorizing about local context does not match your data collection and measurement along
this dimension. There are also some definitional issues that need to be addressed. Below I
discuss these and a few other issues in detail, and I offer some suggestions regarding how you
might address them. I hope you find my comments helpful as you continue to revise your
manuscript.
Theory
1. I'm struggling with your theoretical contribution, which as you know is a major requirement
for publication in AMJ. My understanding is essentially that your claimed contributions are to
look at the simultaneous main effects of both corporate context and local national context on the
entrepreneurial behavior of MNC subsidiaries. Although I am not an expert on MNCs and their
influence on subsidiaries, it seems to me there is quite a large literature on this subject. Similarly,
it also seems there has been a great deal of work on the effects of local country context on
corporate activities and behavior (e.g., Crossland & Hambrick, 2007). Thus, just including them
both in a single study doesn't strike me as a particularly novel theoretical contribution. Further,
while some of the actions studied in previous work may not have been labeled "entrepreneurial,"
my sense is they could be cast that way. One thought experiment I sometimes use to assess the
theoretical contributions of my own work is ask myself if I'm actually studying new processes or
new phenomena where the outcomes are likely to be different. So, in this case, I would ask: (1) If
I were to substitute other subsidiary actions for entrepreneurial actions, is the fact that corporate
and local context affect these behaviors in particular ways a new insight (i.e., are there effects on
entrepreneurial behavior different than they are on other actions)?; and (2) Even if
entrepreneurial behaviors are distinct and different from other behaviors that have been studied
(a claim you don't really make our support), are we learning anything new by studying the effects
of corporate and local context on this type of action, as opposed to all the others studied (i.e., are
we just adding one more action to the list that will be affected in the same ways as all the rest)?
Right now my sense is that the answer to both of these questions is no.
So what can be done? One option would be to look at the effects of potential interactions
between the corporate and local contexts, and develop some theory about when certain aspects of
the corporate context are more or less likely to facilitate entrepreneurial action as a function of
characteristics of the local context. For example, might certain aspects of the corporate context
matter more in complex, hostile and dynamic environments than in more stable, friendly, and
simple local environments? An alternative would be to consider if the effects are linear or nonlinear. For example, is there a point after which hostility overwhelms the effects of corporate
context on entrepreneurial behavior? Thinking in these terms instead of just linear, main effect
terms can yield some more interesting and novel theory.
2. This leads me to an important, related issue, which ultimately is another big problem for your
theorizing. As I discuss in more detail in point 1 under Methods and Data, empirically you aren't
really studying international context because you have no variance in the international location.
Although the parent MNCs are non-U.S., the national local context is only the U.S. The variance
in your sample across local contexts is principally based on industry, not nationality. The only
measure related to nationality is hostility. Thus, there is a fundamental mis-fit between your
theory and data. Although I'm guessing one of your primary interests is in international business,
I'm afraid this study has pretty limited potential to speak to this literature. Other than when
talking about the level of hostility to international competitors within different industry contexts
in another country, you can't speak to any of the issues you use in developing your local context
hypotheses. This isn't a fatal flaw, but it means you have to change what is figure and what is
ground in your framing and hypothesis development. If you focus on aspects of the local industry
context, with hostility towards international competitors as one component, I think can be okay.
3. You use terms like entrepreneurial activities and entrepreneurial intensity to describe the
actions of MNC subsidiaries, but you don't provide an explicit definition of entrepreneurship or
entrepreneurial actions. Instead you appear to essentially treat it as synonymous with
"innovation" in the introduction and theorizing. While on page 6 you equate entrepreneurial
activity with innovation, risk taking and proactiveness, this is not a definition of entrepreneurship
(although it captures some of the elements of Lumpkin & Dess's [1996 AMR] definition of
entrepreneurial orientation). You have to be very careful on this front, because definitions of
entrepreneurship can take on aspects of a religious war, particularly when you are distinguishing
between the activities of start-ups (what some would consider the only "real" entrepreneurship)
and corporate entrepreneurship, or intrapreneurship. Right now I don't think you give this issue
sufficient attention. I personally am not a big fan of the corporate entrepreneurship literature;
most of the theorizing is pretty sloppy, as it conflates different constructs (like entrepreneurship
and innovation) and is based in part on assertions that haven't been empirically supported (like
the notion that entrepreneurs are more risk taking). However, I recognize that there is a
substantial literature in this area you can draw on and within which you can situate your study.
Since you are using the Miller and Friesen (1982) measure of corporate entrepreneurship as your
DV, I suggest that you expressly present their definition of corporate entrepreneurship as the
definition you use in this study and then hew closely to it in your language throughout the paper.
Don't use innovation as a synonym, because it is in fact a distinct theoretical construct. I think
you should also note, either in a footnote or in the discussion section, that there are different
definitions of entrepreneurship, and that you have chosen this one because you are focusing
specifically on a corporate context. You should also be clear that your theorizing and conclusions
only apply to this context, and not to the entrepreneurial actions of start-ups. Then make sure you
adhere to this statement throughout the paper.
4. So to build on my first point, with respect to H1, how might differences in the local context
affect the relationship between a global strategic mandate and the level of subsidiary
entrepreneurship? Given that all subsidiaries within a firm are subject to the same corporate
context, unless you expect all subsidiaries to exhibit identical levels of entrepreneurship as a
result of a GSM, then local context has to matter in generating variations in the outcome. How
might this vary if hostility is greater, or uncertainty and complexity are higher? I'm not going to
go through each hypothesis and ask this same question, but it does apply across the board. Rather
than hypothesizing your main effects for local context, I suggest you drop these hypotheses and
develop a set of interactions (perhaps a,b,c hypotheses) for each corporate context construct.
5. If you follow my advice in point 4, I don't think the relationships would be the same for all
four corporate context dimensions. Specifically, I think financial controls would significantly
diminish corporate entrepreneurship activities when complexity, hostility and dynamism are
high, and strategic controls would enhance corporate entrepreneurship under these conditions.
Although this is more of a methods issue, if you explore these relationships you may want to
consider using a spline function (i.e., create two measures that are zero above (below) some
cutoff and have the values for the measure below (above) the cutoff) to operationalize the local
context constructs. This would allow the slope of the interaction to vary across different parts of
the range, or for the moderator to matter over only part of the range of values. Thus, for example,
you could develop theory to argue that financial controls will limit entrepreneurial activity for
high levels of dynamism, complexity and hostility, but won't have any effect for low levels of
these constructs. This would suggest an important boundary condition missing from prior work.
6. In your hypotheses you should refer to the outcome as corporate entrepreneurship. Also, are
you talking about the level, frequency or what? Be specific in the hypotheses.
Data and Methods
7. As mentioned above, in your theorizing I was led to believe that local context referred to that
nation in which the subsidiary was located, not the location of the parent. Since you surveyed
foreign subsidiaries in the U.S., there is no variance on the local context dimensions for this
direction. Thus, your sample doesn't really match your theory. I then assumed that the local
context you were referring to must be the home country, but you are controlling for country of
origin separately. Then I was really confused. How can you have variance in the local context
measures if they are all referring to the same country? Since you are using perceptual measures I
would expect a little variance, but not enough to generate significant results. Finally it occurred
to me that your effects may be based on the different industries the firms are in. However, this
means your measures don't have much to do with local national context. This is a major issue
you need to address; not only must you justify your decision, you have to make sure your
theorizing in the front end is consistent with your operationalization.
8. Other than the issue raised above, in general I think your approach to conducting the survey
was well-conceived and executed, and your sample seems reasonable. I have a couple of
clarification questions, though. First, how many different corporate parents did the 227
subsidiaries for which you had complete information represent? For the comparison of those that
responded in '95 but not '99 with those that responded both times, what measures did the chisquare test? All of the items they answered the first time? For the t-tests, it sounds like you
compared the parent companies, not the subsidiaries. Is there any way to compare the subs, since
these can vary within company in whether they would respond and why?
9. Although you don't say specifically, I’m assuming that in at least some instances you have
more than one subsidiary from a single company in your sample, or do you only have one sub
per MNC because you are only looking at subsidiaries in the U.S.? Either way you need to
clarify this issue. If you have multiple subs from the same company, then your observations
aren't independent, in which case you need to calculate robust standard errors, and/or use random
or fixed effects regression, depending on how frequently this occurs. If you only have one
subsidiary per MNC, then how are we to assess the difference between company effects in
general and subsidiary-specific characteristics with respect to the effects of local context?
Typically this would be done by seeing if subsidiaries of the same company in two different
countries behaved differently. Either way, you need to provide some more discussion of this
issue.
10. How frequently did the same person who completed the survey in '95 also complete it in '99?
For those cases where it was the same person, was there consistency in their response on
SUBENT? How about for those that were completed by different people across the two time
periods?
11. Simple correlations are not typically considered adequate for assessing interrater reliability,
because they do not take agreement based on random chance into account. Other measures that
do, and that are typically employed to assess interrater reliability include Cohen's Kappa (Cohen,
1968), Krippendorff's alpha (Krippendorff, 2004) and Intraclass correlation coefficients
(McGraw & Wong, 1996). You need to employ one of these other measures in your assessments.
12. You never address the fact that all of your data are perceptual measures based on self-reports.
The fact that you collected your IVs and DVs at different points in time certainly helps, as does
your getting multiple raters for both periods. I personally think that the limitations associated
with these kinds of data are frequently way-overblown. Nonetheless, you should acknowledge
this issue as a potential limitation in your discussion section, and explain why it is not likely to
be a problem here.
References
Cohen J. 1968. Weighted Kappa: Nominal Scale Agreement with Provision for Scaled
Disagreement or Partial Credit. Psychological Bulletin, 70: 213-220.
Crossland, C. & Hambrick, D.C. 2007. How national systems differ in their constraints on
corporate executives: A study of CEO effects in three countries. Strategic Management
Journal, 28: 767-789.
Krippendorff, K. 2004. Content analysis: An introduction to its methodology, Second edition.
Thousand Oaks, CA: Sage.
Lumpkin, G.T. & Dess, G.G. 1996. Clarifying the entrepreneurial orientation construct and
linking it to performance. Academy of Management Review, 21: 135-172.
McGraw, K.O & Wong, S.P. 1996. Forming inferences about some intraclass correlation
coefficients. Psychological Methods, 1: 30-46.
Miller, D. & Friesen, P. 1982. Innovation in conservative and entrepreneurial firms: two models
of strategic momentum. Strategic Management Journal, 3: 1-25.
Exploring homophily and peer-consumer behavior in an accommodation sharing
economy platform
Abstract
Though research attention on the sharing economy is growing, we still lack systematic
understanding and empirical investigations of peer-users’ interaction dynamics on sharing
economy platforms. Employing homophily as a social organizing principle, this study investigates
its impact on peer-consumer behavior in an accommodation sharing economy platform. We also
explore which homophily drivers are salient in the sharing economy context. Our results suggest
that homophily does contribute to peer-users’ consumption intention through trust and attitude,
and that ethnicity and gender, two strong homophily factors identified in the literature, do not
have statistically significant effects on homophily in this context. However, qualitative analysis of
open-ended questions suggests that acquired or value homophily factors may be more influential
in the accommodation sharing economy context. We believe that this study makes a meaningful
contribution to the sharing economy literature by expanding our understanding of peer-users’
interaction dynamics in the sharing economy context.
Keywords: accommodation sharing economy; homophily; trust; Airbnb
1. Introduction
The sharing economy, an umbrella term with a range of meanings, is commonly used to refer to
diverse Internet platforms that enable people to engage in peer-to-peer transactions (Cusumano,
2015; Hamari, Sjöklint, & Ukkonen, 2016; Sundararajan, 2016). Such platforms enable individuals
to transact underutilized individual resources/assets such as labor, spare time, automobiles, extra
bedrooms, and other household items or tools (Cusumano, 2015) by bringing together or creating
market places for individuals who are willing to share the resource (peer-providers) and others
who are willing to use/rent/purchase it (peer-consumers).
The sharing economy is drawing attention from academia and practitioners with its
phenomenal growth and potential. For example, 33.9 million adults used sharing economy services
such as Airbnb and Uber in the United States in 2017, and its number is expected to increase to
45.6 million by 2022 (Statistica, 2018). The European sharing economy market including the five
sectors of accommodation, transportation, buying or selling of goods, sharing or renting of goods,
and on-demand personal services facilitated €27.9 billion worth of transactions between May 2015
and May 2016, with an estimated 191 million Europeans engaging in at least one transaction
involving payment (Hausemer et al., 2017).
While there exist high expectations about the sharing economy due to its potential to
democratize economic activity, increase inter-personal interaction, and provide more sustainable
and environmentally friendly options in the market (Botsman & Rogers, 2010; Hamari et al., 2016;
Hellwig, Morhart, Girardin, & Hauser, 2015), we lack systematic understanding of peer-users’
interaction dynamics in the sharing economy. According to Azevedo and Weyl (2016), participants
in a marketplace care more about who they transact with, i.e., their peer-users in the marketplace,
rather than the marketplace itself. In addition, an industry expert pointed out the importance of
trust among peer-users in the peer-to-peer marketplace such as sharing economy platforms (Ufford,
2015), which has been supported by recent studies on the sharing economy (Cho, Park, & Kim,
2019; Ert, Fleischer, & Magen, 2016; Milanova & Maas, 2017). However, there have been fewer
empirical studies on what can drive trust in peer-users’ interactions on the sharing economy
platforms.
In this study, we employ a well-known social organizing principle of homophily and
empirically investigate the role of homophily in the peer-users’ interaction dynamics in the sharing
economy platform, as the literature suggests that contacts between similar people occur at a higher
rate than among dissimilar people (McPherson, Smith-Lovin, & Cook, 2001). More specifically,
the objective of this study is to empirically examine how homophily contributes to peer-users’
behavioral intention through trust and attitude in an accommodation sharing economy platform,
and to explore which factors can drive peer-users’ perception of homophily.
In the following, we review the theoretical background and introduce the research model and
hypotheses. Then we present our research method, analysis, and results. Finally, we discuss the
implications, limitations, and future research directions.
2. Theoretical background and hypotheses development
2.1. Conceptualizing sharing economy as exchanges, a type of multi-sided markets
Peer-to-peer platforms which enable people to collaboratively make use of their underutilized
resources are known as the “sharing economy” collectively. However, the definition and scope of
the term is still evolving as it is referenced differently: ‘sharing economy’ (Cohen & Kietzmann,
2014; Hamari et al., 2016), ‘collaborative economy’ (Botsman, 2014), ‘collaborative consumption’
(Botsman & Rogers, 2010), ‘access-based consumption’ (Bardhi & Eckhardt, 2012), and ‘crowdbased capitalism’ (Sundararajan, 2016), ‘two-sided markets/multi-sided platforms’ (Evans,
Schmalensee, Noel, Chang, & Garcia-Swartz, 2011), ‘matching markets’ (Azevedo & Weyl, 2016).
Botsman and Rogers (2010) use the term ‘collaborative consumption’ as an encompassing concept
including swap trading, time banks, local exchange trading systems, bartering, social lending, peerto-peer currencies, tool exchanges, land share, clothing swaps, toy sharing, shared workspaces,
cohousing, co-working, car sharing, crowdfunding, bike sharing, ride sharing, food co-ops,
walking school buses, shared micro crèches, and peer-to-peer rental. Prior research also offers a
different set of terms to cover such a broad spectrum. Belk (2010, 2014) notes that sharing and
collaborative consumption may not be used interchangeably. He defines sharing, mainly
nonreciprocal pro-social behavior, as the act and process of distributing what is ours to others for
their use and/or the act and process of receiving or taking something from others for our use, and
argues that sharing is different from either gift giving or marketplace exchange. Belk (2014) also
defines ‘collaborative consumption’ as “people coordinating the acquisition and distribution of a
resource for a fee or other compensation which encompasses bartering, trading, and swapping.”
Sundararajan (2016) states that the sharing economy covers a wide variety of economic forms with
gift economies at one end and market economies at the other end of the spectrum and suggests that
“crowd-based capitalism” be used in place of the ‘sharing economy’.
Despite a variety of terms being used as this phenomenon socially and technologically evolves,
we still lack a theoretical underpinning of the phenomenon. Is the sharing economy a truly new
transaction form? Or is it another name of C2C e-commerce? Or is it a sophisticated and
technology-mediated version of traditional marketplaces where individuals buy and sell their
products and services? If it is a genuinely new market form, what are unique characteristics that
distinguish it from others? From an extensive literature review, we notice the theory of two-sided
markets/multi-sided platforms from the field of economics to provide theoretical ground to
contextualize the sharing economy. Such a viewpoint is shared among those, who began to use
‘platform economy’ to refer to sharing economy platforms (Farrell, Greig, & Hamoudi, 2018;
Kenney & Zysman, 2016)
A two-sided market is a platform that provides goods or services to two distinct groups of
customers who need each other in some way and who rely on the platform to intermediate
transactions between them. Evans et al. (2011) suggested that “multi-sided platforms” are more
appropriate as markets here refer to businesses and some platforms often serve and support more
than two interdependent types of customers. We subscribe to this view and use ‘platforms’ in this
paper. Two-sided platforms enable parties to realize gains from trade or other interactions by
reducing the transaction costs of finding and interacting with each other, and they perform three
core functions of matchmaking, building audiences, and minimizing costs (Evans & Schmalensee,
2011).
Evans (2011) categorizes multi-sided platforms into four types: exchanges, advertisersupported media, transaction devices, and software platforms. Exchanges help buyers and sellers
search for feasible contracts or mutually advantageous transactions. Examples of exchanges
include NASDAQ, many business-to-business, and consumer-to-consumer eCommerce sites,
various kinds of brokers and financial exchanges. Another type of multi-sided market is
advertising-supported media where a platform provides content and attracts viewers and
advertisers. Examples are magazines, newspapers, free television, yellow pages, and many Internet
portals. Some advertising-supported media platforms are multi-sided. For example, YouTube
serves content generators, content consumers, and advertisers. The third category is transaction
systems such as card associations (e.g., American Express, Discover, MasterCard, and Visa). Most
card associations serve two-sided markets with an agreement/contract of a percentage of the
transaction with merchants and annual fees and/or various rewards with cardholders. The fourth is
software platforms. A software platform (e.g., iPhone platform) provides services for application
developers and helps them obtain access to the hardware for the computing device in question.
The other party of this platform is users who use these applications only if they have the same
software platform that developers have relied on in writing their applications (Evans, 2011).
We subscribe to the view that considers sharing economy platforms as two-sided/multi-sided
platforms for two major reasons. First, most sharing economy platforms perform three core
functions of two-sided platforms such as matchmaking, building audiences, and minimizing
transaction costs as Evans (2011) suggested. Sharing economy platforms serve as matchmakers to
facilitate exchange by making it easier for peer-consumers to find peer-providers (e.g., Airbnb and
Uber). They need to build audiences because accumulating a critical mass of peer-users (i.e., peerconsumers and peer-providers) makes it more likely that those peer-users will find a suitable match.
They also provide resources to reduce the cost of providing services to both peer-users. For
example, Airbnb and Uber provide insurance for homeowners and drivers to attract more potential
service providers. In addition, sharing economy platforms share the defining characteristic of twosided platforms, which is the “symbiotic relationship between the two sides” meaning the
platforms must cater to two different customer groups (i.e., peer-consumers and peer-providers)
simultaneously to be successful (Evans, 2011). Second, sharing economy platforms sit on such a
wide spectrum of business models including exchanges, advertiser-supported media, transaction
devices, and software platforms. The multi-sided platform theory provides a comprehensiveenough theoretical framing to cover such diversity. Out of the four categories of two-sided
platforms, we suggest that the majority of sharing economy platforms belong to “exchanges.”
Exchanges provide participants with the ability to search participants on the other side, and cover
various match-making activities, which include traditional exchanges such as auction houses,
Internet sites for business-to-business, person-to-business, and person-to-person transactions,
various kinds of brokers (e.g., insurance and real estate), and financial exchanges for securities
and futures contracts (Evans & Schmalensee, 2011).
Sharing economy platforms are distinct from traditional exchanges. Sharing economy
platforms are technology-based, and information technologies enable transaction scalability with
increased speed, reduced cost, and enhanced convenience, connecting highly distributed and
decentralized individuals, markets, or networks. Matching-making on large scale among
distributed users would not have been possible without such technology-based platforms. Sharing
economy platforms cater to private individuals called “peers” not business entities. Although we
had transactions among individuals in the past, the scale and scope of transactions cannot be
compared with those in sharing economy platforms. What adds to the newness of sharing economy
platforms is the wide variety of services and products being transacted (e.g., lodging, driving
services, and personal services). This combination of product/service types, providers as private
individuals, and the scale of transactions enabled by technology has been unprecedented. It is an
agreeable statement that sharing economy platforms enabled by Internet technologies have
unleashed multi-sided markets for peer-to-peer transactions of underutilized individual resources
on large scales.
2.2. Homophily
Homophily is a social organizing principle suggesting that contacts between similar people occur
at a higher rate than among dissimilar people (McPherson et al., 2001). Homophily structures
network ties of every type, including marriage, friendship, work, support, information transfer,
exchange, and other types of relationships. Forming of social ties based on homophily has been
studied in various contexts such as social networks, voluntary associations, social capital, social
movements, culture, organizations, and a variety of substantive topics that are affected by network
processes (Ibid).
Homophily is categorized into two types: status homophily and value homophily (Lazarsfeld
& Merton, 1954). In status homophily, the similarity is based on “ascribed” characteristics
including the major sociodemographic dimensions such as race, ethnicity, sex, or age, or “acquired”
characteristics such as religion, education, occupation, and other behavioral patterns (McPherson
et al., 2001). Value homophily is based on individuals’ values, attitudes, and beliefs. Various
homophily factors have been studied in the sociology and communications literature. In their metastudy on homophily dimensions, McPherson et al. (2001) report that salient dimensions of
homophily include ethnicity, gender, age, religion, education, occupation, social class, network
positions, behavior, attitudes, abilities, beliefs, and aspirations. We adopt ethnicity and gender,
two well-established “ascribed” factors of status homophily in the existing social science literature
and propose that such factors would take the same effect on homophily in peer-users’ interactions
on sharing economy platforms. Thus, we state the following hypotheses.
H1: A peer-consumer perceives higher levels of homophily for a peer-provider with the same
ethnicity.
H2: A peer-consumer perceives higher levels of homophily for a peer-provider with the same
gender.
Some recent IS studies have explored the effects of homophily in various contexts. Gu et al.
(2014) studied the extent to which social media participants exhibit homophily in the context of
virtual investment communities and found that investors are more likely to respond to those threads
that echo their own opinions (i.e., value homophily). Kordzadeh and colleagues (2014)
investigated the impact of reciprocity and homophily (similarity of user characteristics such as age,
gender, and tenure) on user participation in virtual health communities and found that gender
homophily is positively associated with posting supporting messages for peers. In their analysis of
Facebook data, Han and colleagues (2015) found that people tend to exhibit more similar
tastes/interests if they have similar demographic characteristics (e.g., age, location). More
interestingly, the effect of homophily on trust has been explored and studied in the context of ecommerce transactions. Gaskin and Oakley (2010) found that customers who feel similar to a
product reviewer are more likely to trust the person and purchase the product. Lin and Xu (2017)
found that perceived similarity in ethnicity has a significant effect on perceived reviewer
trustworthiness in online consumer reviews. Thus, we propose that the sense of homophily affects
trust in peer-to-peer interactions in the context of sharing economy platforms and state the
following hypothesis.
H3: A peer-consumer’s perceived homophily for a peer-provider is positively associated with
his/her perceived trust in the peer-provider.
Homophily is known to have powerful implications for the attitudes individuals form and the
interactions they experience (McPherson et al., 2001). In their study of the consumer-generated
media adoption, Ayeh and his colleagues (2013) argued that perceived homophily would influence
both credibility and attitude, and found that homophily has a direct impact on attitude and an
indirect impact through trustworthiness. In addition, Kim et al. (2018) introduced homophily in
their study on the social influence of online reviews and empirically found that consumers having
strong homophily with a website have positive attitudes towards the website and the information
on it. Based on the literature supporting the impact of homophily on attitude, we propose the
following hypothesis.
H4: A peer-consumer’s perceived homophily for a peer-provider is positively associated with
his/her attitude towards the peer-provider.
2.3. Trust, attitude, and consumer behavior
While trust has received research attention in the information systems literature, it has been
extensively discussed in the context of e-commerce (Kelton, Fleischmann, & Wallace, 2008;
McKnight, Choudhury, & Kacmar, 2002; McKnight, Cummings, & Chervany, 1998; Pavlou &
Gefen, 2004). Previous studies argue that trust is an essential element in e-commerce because the
risks of opportunistic behaviors such as violations of privacy, conveying inaccurate information,
unauthorized tracking of transactions, fraudulent transactions, unfair pricing, and unauthorized use
of credit and private information are much higher in the e-commerce environment than in the
traditional commerce environment with actual human interaction and physical inspection (Gefen
& Straub, 2003, 2004).
Prior research on trust has explored and discussed the effects of trust on human behavior. For
example, trust has been identified and studied as a salient factor that affects perceived uncertainty
(Pavlou, Liang, & Xue, 2007), intended use of the system (Gefen, Karahanna, & Straub, 2003;
Vance, Elie-Dit-Cosaque, & Straub, 2008), customer attitude (Elliot, Li, & Choi, 2013; Pan &
Chiou, 2011; Teo & Liu, 2007), and purchase/repurchase intention (Chiu, Wang, Fang, & Huang,
2014; Cho et al., 2019; Fang et al., 2014; Ou, Pavlou, & Davison, 2014). More specifically in the
e-commerce context, Ayeh and his colleagues (2013) examined online travelers’ perceptions of
trustworthiness on the user-generated sources and found that such perceptions influence their
attitudes towards the source in the travel planning process. In their study on Airbnb, Ert and
colleagues (2016) found that peer-users’ purchase decision is influenced by trustworthiness of
hosts. Drawing on prior research on trust and its effects on attitude and behavioral intention in the
e-commerce context, we examine the impact of a peer-consumer’s perceived trust on his/her
attitude and behavioral intention in sharing economy platforms. Thus, we state the following
hypotheses.
H5: Perceived trust is positively associated with a peer-consumer’s attitude towards peerproviders.
H6: Perceived trust is positively associated with a peer-consumer’s consumption intention.
Theories of reasoned action and planned behavior have been among the most influential
theories in explaining and predicting a wide range of behaviors (Ajzen, 1985; Fishbein & Ajzen,
1980; Sheppard, Hartwick, & Warshaw, 1988). According to those theories (Ajzen, 1991), the
proximal determinant of behavior is the behavioral intention, and attitude is considered as one
important determinant of behavioral intention. The link between attitude and behavioral intention
has been well-established theoretically and empirically in previous studies on e-commerce (Elliot
et al., 2013; Teo & Liu, 2007). For example, Chang et al. (2005) developed reference models of
online shopping adoption from an extensive literature search and review, where they found that
six empirical papers confirmed the significant positive impact of online shoppers’ attitude on
online shopping intention and behavior. Hence, we state the following hypothesis.
H7: A peer-consumer’s attitude is positively associated with his/her consumption intention.
Figure 1 shows a schematic diagram of the research model, including all seven hypotheses.
Figure 1. Research model.
3. Research method, analysis, and results
3.1. Instrument development, procedure, and analysis
The survey method was used for this study. The survey items were used to test the hypotheses in
the research model, and open-ended questions were asked to explore homophily factors not
hypothesized in the model. For the survey instrument, we adopted the established multi-item
measures for homophily, trust, attitude, and consumption intention. We adapted four items of
homophily from McCroskey et al. (1975), three items of trust from Gefen and Straub (2003), two
items of attitude from Pavlou and Fygenson (2006), and three items of consumption intention from
Gefen and Straub (2003) and Cho et al. (2019). Those items were measured on a seven-point Likert
scale ranging from “strongly disagree” to “strongly agree” or “very unlikely” to “very likely”. Two
single-item dichotomous measures were created to ask participants if they have a host of the same
or different gender, and the same or different ethnicity. Appendix shows the constructs and
measures used in this study.
For data collection, we conducted a structured and open-ended online survey about Airbnb,
one of the fastest-growing sharing economy platforms. Airbnb provides a peer-to-peer lodging
service where peer providers voluntarily share their identity information such as gender, ethnicity,
and age with a typed profile and a photo. Hence, we believe that Airbnb is suited to test our research
model hypothesizing the antecedents (e.g., gender and ethnicity) and the consequences (e.g., trust,
attitude, and consumption intention) of homophily on sharing economy platforms.
At the beginning of the survey, participants were asked to read a short scenario explaining the
survey context where they plan to visit New York for vacation and have already searched for
candidate lodging places on Airbnb. After reading the scenario, they were guided to click a host
profile page link embedded in the online survey, which was a randomly generated link from the
pool of pre-designed host profiles. The pool of host profiles included six combinations of two
different gender (e.g., female and male) and three different ethnic groups (e.g., black, Hispanic,
and white). One host profile was randomly selected from the pool and assigned to each participant.
While all the other information including the travel destination (i.e., New York), job, and selfintroduction was controlled in the survey to preclude any effect of different preferences, each
participant was randomly assigned to a host with a 50% chance of the same gender and a 33%
chance of the same ethnicity. Each participant was given time to read the profile information
including the photo of his/her potential host, and then was asked to answer the survey questions.
A total of 419 participants were recruited in two public universities in the United States. Thirtyfour responses were dropped due to incomplete questionnaires, and the response rate is 91.8% with
the completed responses of 385. Eighty-five out of 385 participants have used Airbnb or other
peer-to-peer lodging services before. We ran the path analysis separately for each of two
participant groups with and without prior peer-to-peer lodging service experience and found no
difference between the two groups. Based on this preliminary analysis results, we decided to
combine the data for further analysis.
Partial least squares (PLS) analysis with SmartPLS 3 was used as the primary analysis tool to
validate the measurement and estimate the structural paths in the research model (Hair Jr, Hult,
Ringle, & Sarstedt, 2013). As an extension of the multiple linear regression model (Marcoulides
& Saunders, 2006), PLS first computes loadings of indicators on each construct in the
measurement model, and then iteratively estimate causal relationships among construct in the
structural model (Fornell & Bookstein, 1982). PLS is considered preferable to other traditional
methods such as factor analysis and regression because it assesses both measure and structural
models (Gefen, Straub, & Boudreau, 2000). We prefer to use PLS analysis because we can analyze
the complex research model with six constructs in one unified process and this study is more
exploratory nature of research attempting to understand the variation in the dependent variables
explained by the independent variables in the proposed model (Gefen et al., 2000; Petter, 2018).
3.2. Measurement model
In the measurement model of PLS analysis, we evaluate convergent and discriminant validity by
examining the psychometric properties of the construct measures. As the assessment of convergent
validity, we calculated and examined the standardized loadings for each factor model. The
standardized loading is the shared variance between each item and its associated construct, which
should be higher than 0.707. Table 1 shows that the standardized loadings of all measurement
items are 0.86 or higher. Therefore, we retain all the indicators for subsequent analysis.
In order to assess the internal consistency for each block of measures in each construct, we
examine the Cronbach’s alpha, composite reliability, and average variance extracted (AVE). While
the threshold for Cronbach’s alpha and composite reliability is not absolute, it is suggested that
0.70 indicates extensive evidence of reliability and 0.80 or higher provides exemplary evidence
(Bearden & Netemeyer, 1999). As shown in Table 1, all the constructs in the measurement model
exhibit Cronbach’s alpha of 0.79 or higher and the composite reliability of 0.90 or higher,
indicating exemplary reliability. AVE has been suggested as another measure of construct validity
(Fornell & Larcker, 1981), which compares the amount of variance obtained from indicators with
variance due to measurement error. The acceptable value for the AVE is 0.5 or higher, indicating
that 50% or more variance of the indicators is accounted for. Table 1 shows that all the AVEs are
0.78 or higher. Thus, our evaluations of standardized loadings, Cronbach’s alpha, composite
reliability, and AVE indicate that the construct reliability of all measurement items in this research
has been established satisfactorily.
Table 1. Item loadings and reliability.
We conducted two tests to evaluate the discriminant validity. First, we calculated and
compared each indicator’s loading on its own construct and its cross-loadings on all other
constructs. Table 2 shows that each indicator has a higher loading with its own construct than its
cross-loadings with any other constructs. For example, HOM1 loads higher on Homophily (0.906)
than on Trust (0.544), Attitude (0.407) or Consumption Intention (0.423). Moreover, all indicators
for their intended construct form a block with similar loadings, which are higher collectively than
the loadings of other bocks in each column. These results show that discriminant validity is
established.
Table 2. Construct loadings and cross loadings.
As the second test of discriminant validity, we compared AVE of each construct with the
shared variance between all possible pairs of constructs (Fornell & Larcker, 1981). As shown in
Table 3, AVE for each construct is higher than the squared correlation between the construct pairs.
It means that more variance is shared between the latent construct and its block of indicators than
with other constructs representing different blocks of indicators. Thus, it also supports discriminant
validity.
Table 3. AVEs versus squares of correlations between constructs.
3.3. Structural model
We can assess the structural model by examining path coefficients, their significance levels, and
the R2 values of dependent variables. First, we computed the path coefficients with the entire
sample of 385 and employed the bootstrapping method with 500 resamples to obtain the t-values
corresponding to each path. The acceptable t-values for two-tailed tests are 1.96 and 2.58 at the
significance levels of 0.05 and 0.01, respectively.
As shown in Figure 2, while four out of seven hypothesized paths have been supported, three
of them are not supported. First, both ethnicity and gender did not have a significant effect on
homophily (β = 0.059, n.s.; β = 0.067, n.s.; respectively) and therefore H1 and H2 are not supported.
It means that neither ethnicity nor gender is a significant factor influencing a consumer’s
perception of homophily. Homophily positively affects trust (β = 0.584, p < 0.01), supporting H3.
While trust has a direct positive effect on attitude (β = 0.537, p < 0.01, supporting H5), homophily
does not have a direct, significant effect on attitude (β = 0.088, n.s., not supporting H4). These
results imply that homophily has an indirect effect on attitude via trust. Both trust and attitude
positively affect consumption intention (β = 0.309, p < 0.01; β = 0.539, p < 0.01; respectively),
supporting H6 and H7.
Another important analysis in the structural model is to evaluate the explanatory power by
examining the R2 value of the final dependent variable. The final dependent variable, consumption
intention, had an R2 value of 0.58, which indicates that our research model accounts for 58% of
the variance in the dependent variable. This R2 value is sufficiently high to indicate that trust and
attitude have a reasonable power to explain the consumption intention in sharing economy
platforms such as Airbnb. In addition, this R2 value for the final dependent variable is comparable
to the results obtained in previous studies that examined other factors influencing purchase or
consumption intention in the e-commerce context. For example, Ou et al. (2014) reported an R2 of
50% for purchase intention in their research model that investigated the effects of communication
technologies on purchase in online market places, and Cho et al. (2019) reported an R2 of 57% for
a model that investigated the effects of social presence and trust on consumption intention in a
sharing economy platform. It is also instructive to examine the R2 values for the intermediate
variables in the research model. The R2 values for attitude, trust, and homophily are 0.351, 0.341,
and 0.008, respectively. The first two R2 values for attitude and trust are also high enough to make
meaningful interpretations of the path coefficients from one independent variable of interest (i.e.,
homophily) to the intermediate variables (i.e., trust and attitude) in the research model. However,
the R2 value for homophily is not high enough to make meaningful interpretation of the path
coefficients from two independent variables (i.e., ethnicity and gender). It may not be a surprise
since both path coefficients are not significant in the hypothesis testing.
Figure 2. Structural model.
3.4. Post hoc analysis on the effects of ethnicity and gender on homophily
While the homophily literature has identified ethnicity and gender as strong factors affecting
homophily, our analysis in the structural model shows that their effects on homophily are not
significant. In order to better understand the relationship between homophily and the two proposed
antecedents, ethnicity and gender, we conducted a two-way factorial analysis of variance
(ANOVA) with two independent variables (ethnicity and gender) on one dependent variable
(homophily). A benefit of the factorial ANOVA is to test an interaction effect of independent
variables on the dependent variable. The results of ANOVA show that the interaction of ethnicity
and gender (ethnicity × gender) does not have a significant effect on homophily (F(3,381) = 0.042,
p > 0.05, n.s.).
While there is no interaction effect from ANOVA results, we notice that somewhat large
difference of means between two extreme groups in terms of ethnicity and gender, which are a
group with same ethnicity and same gender and a group with different ethnicity and different
gender (see and compare the two groups with bold-faced mean values in Table 4). To test if the
homophily of the group with same ethnicity and same gender is significantly larger than that of
the other group with different ethnic and different gender, we ran one-tailed, two-sample t-test and
found the t value of +1.760, which is significant at p
Purchase answer to see full
attachment