Experimenting with Networks

Arun G. Chandrasekhar; Matthew O. Jackson

Experimenting with Networks

T0 review · 0 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read An experiment on a networked population is identifiable only to the extent that many independent copies of each treatment exposure exist.

desk verdict A solid, useful review chapter on designing network experiments; the central 'many copies' principle is asserted rather than sharply delimited, but this is a review, not a theory paper. read the letter →

arxiv 2506.11313 v1 pith:2EZY7PSK submitted 2025-06-12 econ.GN q-fin.EC

Arun G. Chandrasekhar , Matthew O. Jackson This is my paper · ORCID

classification econ.GNq-fin.EC

keywords networkexperimentsinterferencespilloversexposuremapsfieldlabnaturalexperimentaldesign

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is a survey chapter, but it argues a single organizing thesis: an experiment on a networked population is identifiable and powered only to the extent that the researcher can obtain many statistically independent copies of how each node is treated, directly and through its neighbors. This can come from many separate networks, or from one large network in which spillovers decay quickly enough with distance. The authors use exposure maps to turn this thesis into practical guidance on design: whether to randomize individuals or clusters, how much network data to collect and at what precision, whether panels help, and which settings favor lab, field, or natural experiments. If the thesis is right, it gives applied researchers a usable test for when causal inference on networks is feasible and when it is not.

What carries the argument

The central object is the exposure map, $d_i = f_i(D_{1:n}, G)$: a function that reduces the full treatment vector and network to the exposure status that determines node $i$'s potential outcome $Y_i(d_i)$. The related structural causal map plays the same role for causal parameters. In the paper's running example, the exposure structure is generated by best responses in a linear-quadratic network game, $Y_i = \varepsilon_i + \beta \sum_j G_{ij}Y_j$, whose iterated solution makes the influence of far-away nodes decay as powers of $\beta$ times walks in $G$. This object carries the argument because it tells the researcher whether exposures are effectively independent: if $d_i$ and $d_j$ are systematically highly correlated for most $i,j$, treatment is an aggregate shock and no amount of data rescues identification.

What would settle it

Run or simulate a large connected network experiment in which the treatment deliberately shifts outcomes at every distance, so far-apart neighborhoods are not independent, then apply the paper's neighborhood-averaging estimator and increase $n$; if the estimated treatment effect keeps a bias that does not shrink, the limited-spillover premise is false.

Watch

Extended reading notes

Core claim

The paper's central claim is that when the stable unit treatment value assumption fails because outcomes depend on neighbors' treatments and on network structure, causal inference still goes through if the dependence can be compressed by an exposure map into a limited set of exposure types and each type has many nearly independent replications. In the linear-quadratic example, the influence of node $j$ on node $i$ is $\sum_{t\ge 1}\beta^t G^t_{ij}$, which vanishes with network distance when $\beta$ is small enough; this is the concrete mechanism by which a single large network can supply the needed independence. The authors conclude that researchers should either collect many independent networks, or restrict attention to settings with limited and well-measured spillovers, or impose a parsimonious parametric model that lets the joint distribution of outcomes carry the identification.

Load-bearing premise

The prescription relies on the assumption that spillovers fade with network distance quickly enough for far-apart neighborhoods to be nearly independent; if interference is long-range, dense, or hidden in unmeasured relationships, the recommended designs and power calculations lose their warrant.

Editorial extensions

If this is right

If spillovers decay with distance and the network is large enough, consistent estimation and inference are possible from a single network; otherwise many independent networks are required.
Cluster designs beat Bernoulli randomization when spillovers are nontrivial, with worst-case bias tied to the share of cross-cluster links and worst-case variance tied to cluster-size imbalance.
With partial or noisy network data, model-based design that estimates a generative graph model and optimizes treatment allocation over coarse groups can outperform full network data paired with naive randomization.
Exact finite-sample randomization tests can be built for nonsharp null hypotheses such as no spillovers or no second-order spillovers by constructing an artificial experiment with focal nodes and re-randomization.
Factorial designs followed by data-driven pooling let researchers study high-dimensional treatment combinations without needing a prohibitive number of independent networks.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper's 'many copies' principle implies an effective-sample-size diagnostic: a researcher could report the number of nearly independent exposure neighborhoods, much as cluster designs report effective cluster counts, as a standard accompaniment to power calculations.
The same logic suggests adaptive designs should oversample rare exposure types in pilot waves rather than only central nodes, since the binding constraint is replications of each exposure configuration, not average connectivity.
If spillovers are suspected to be long-range, the framework points to a robustness exercise: re-estimate effects under several assumed interference radii and check how much the conclusions move; the paper's own examples stop at finite neighborhoods, but the logic invites this stress test.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague, this is a handbook chapter, not a research paper, and it should be read that way. It offers no new estimator or theorem, but it does something valuable: it gives the literature on network experiments a coherent organizing vocabulary (exposure maps) and a practical walk-through of design choices across field, lab, and natural experiments. The authors know this material from the inside, and it shows. The sections on many independent networks versus a single large network, on link elicitation, and on partial network data (ARD, RDS, star sampling) are concrete and useful. The factorialization discussion makes a good, non-obvious point: pre-emptively collapsing treatment arms is equivalent to deciding you already know the 23 cells you aren't testing.

The soft spots are few but real. There's a mechanical typo in Section 2.5.1 ("Causal Clustering is . Given..."). The 80% budget saving from ARD is repeated without enough detail to check. More substantively, the stress-test note is right that the chapter's load-bearing idea—that you can identify and power a network experiment when you have many approximately independent copies of a treatment neighborhood—is illustrated with the linear-quadratic game, but no general condition or diagnostic is given for when spillovers are local enough. The chapter acknowledges the perfect-diffusion extreme, and it points to the CLT and estimation papers for the technical conditions, but a practitioner is left to guess whether their setting is in the identifiable regime. That is a real gap, though it is a gap in the field, not just this chapter.

Who is this for? A grad student planning a field experiment, or anyone wanting a single map of the literature. It is not for theorists seeking new results. I'd send it to peer review as a handbook chapter; the authors should fix the typos and add a short paragraph flagging the conditions under which the 'many independent neighborhoods' strategy is justified. Verdict: solid review, worth engaging.

Referee Report

0 major / 7 minor

Summary. This is a review chapter on the design and analysis of experiments when outcomes are subject to network interference. The authors develop a vocabulary centered on exposure maps, argue that an experiment is identified and powered only if the researcher can find 'many copies' of treatment exposures with enough statistical independence, and apply this principle to field, lab, and natural experiments. The chapter covers data collection and link elicitation, the choice between many independent networks and a single large network, cluster and adaptive designs, partial network data, measurement error, panel data, lab experiments with designed networks, and natural experiments exploiting random assignment to network positions. It emphasizes trade-offs between parametric structure and nonparametric generality, and it points to exact randomization tests for non-sharp null hypotheses as a way to learn about the extent of spillovers.

Significance. As a survey, the chapter succeeds in organizing a large literature around a practically relevant principle: network experiments are only as strong as the availability of approximately independent copies of the treatment exposure of interest. Its strengths include a clear exposition of exposure maps, a balanced treatment of many-versus-one networks, a careful account of partial network data and its use in optimal design (Reeves et al. 2024), and the inclusion of exact finite-sample tests for non-sharp null hypotheses. The authors are appropriately explicit that the principle requires limits on spillovers and that parametric assumptions may be necessary. The stress-test concern about distance decay is partly mitigated by these explicit caveats and by the §1.4 tests, which provide a practical diagnostic for the extent of spillovers; however, the chapter could more clearly state that the formal decay argument in §1.2 is model-specific. Overall, the framework is sound and useful for guiding design choices, even though it deliberately offers no new theorem.

minor comments (7)

[2.5.1] The sentence 'The practical implementation of Viviano et al. (2023)'s Causal Clustering is.' is broken; it should be completed, for example, 'The practical implementation of Viviano et al. (2023)'s Causal Clustering is described by the authors.'
[2.3] The claim that ARD would have saved 80% of the budget while yielding the same conclusions is a strong quantitative statement attributed to unpublished J-PAL South Asia calculations; the chapter should cite the precise appendix of Breza et al. (2020) where these calculations appear, or attribute them more cautiously.
[1.3] The expression 'Yi(Dj :j∈C i,G|Ci)' appears to be missing notation; it should be written as Yi((Dj)_{j∈Ci}, G|_{Ci}) to indicate dependence on treatments within the cluster and the induced subgraph on Ci.
[1.6] The text says Figure 1 presents the case of 'two arms each varying from intensities 1-3, omitting the third arm,' but the example has three factors; please clarify that Figure 1 is a two-dimensional slice of the three-factor design.
[References] There are duplicate and inconsistent reference entries: Banerjee et al. (2013a) and (2013b) are the same Science article under different author formats; Banerjee et al. (2023) and (2024c) appear to be the same Econometrica paper; and Chandrasekhar and Jackson (2024a) and (2024b) are identical entries. These should be consolidated.
[2] In the opening paragraph, 'or in in some combination' should be 'or in some combination'; likewise, §3.2.1 has 'uncertain about others' .' where a word such as 'connections' has been omitted.
[1.2] The decay statement following Eq. (1.4) is true for the linear-quadratic game but is not a general property of network interference. For clarity, the chapter should explicitly say that for non-linear processes (e.g., threshold or complex contagion) the 'many copies' approach requires the researcher to verify limited spillovers, and should refer readers to the §1.4 tests for that purpose.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the chapter is a survey whose 'many copies' principle is a stated framework, not a result derived from its own outputs.

full rationale

No circularity found. The chapter is an overview of methods for network experiments, not a derivation of a new result from first principles. The central organizing claim—that an experiment in a network setting is identified and powered only to the extent that one can find many approximately independent copies of treatment exposures—is presented as a conceptual condition on exposure maps, and it is used to structure the survey rather than derived from an equation that already contains the conclusion. Equation (1.4) is an illustrative linear-quadratic example, credited to Ballester et al., showing how influence between nodes decays with network distance under the stated eigenvalue condition; the surrounding text explicitly flags this as one model ('This is a variation on the linear-quadratic network game'), not as a general theorem. The chapter also offers a second, explicitly parametric route to progress when the 'many copies' condition fails, which further shows that the advice is not forced by a single definitional equivalence. Section 1.3 labels the cluster-independence approximation as an assumption ('represent the key case that applies to many settings'), not as a conclusion derived from the linear-quadratic model. The paper's self-citations (e.g., Chandrasekhar et al. 2023b for a central limit theorem; Chandrasekhar and Jackson 2024a for consistency under limited interference) are used as supporting references for general statistical results and as examples from the authors' own research agenda, but the chapter's design guidance does not reduce to those citations; it also independently draws on Athey et al. 2018, Aronow and Samii 2017, Viviano et al. 2023, and others. The skeptical concern that distance decay is not guaranteed outside the linear-quadratic model is a substantive assumption or correctness risk about the scope of the framework, not a circularity in the paper's own argument. The chapter does not fit any parameter and then rename that fit as a prediction, and it does not invoke a uniqueness theorem imported from the authors' prior work to forbid alternatives. Therefore the honest finding is no significant circularity, with score 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The chapter introduces no new parameters or entities. Its prescriptions rest on domain assumptions about interference structure (SUTVA violation, exposure-map reduction, and decay of spillovers) and on a standard result in network game theory.

assumptions (4)

domain assumption SUTVA is violated whenever outcomes depend on others' treatments and the network G (Section 1.1).
This is the starting point, cited to Rubin (1974, 1980). The review does not prove it; it is standard in causal inference with interference.
domain assumption Exposure maps d_i = f_i(D_{1:n}, G) fully summarize how treatments and network map into outcomes (Section 1.1).
The chapter assumes this reduction is valid for design and identification; prior work motivates it, but it is not proven in this text.
domain assumption Spillovers decay sufficiently fast with network distance so that many approximately independent neighborhoods exist (Section 1.2, condition (i)).
The central power and identification guidance depends on this; when it fails (e.g., dense or long-range interference), the recommended approach collapses.
standard math The linear-quadratic game (1.1) has a unique equilibrium when beta is less than the reciprocal of the magnitude of the first eigenvalue of G (Section 1.2).
This follows from Ballester et al. (2006a); the chapter uses it to illustrate exposure effects.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Experimenting with Networks." pith.science (2026). https://pith.science/paper/2EZY7PSK

@misc{pith2026250611313,
  author       = {Pith},
  title        = {Pith review of: Experimenting with Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2EZY7PSK}},
  note         = {Machine review of arXiv:2506.11313}
}

read the original abstract

We provide an overview of methods for designing and implementing experiments (field, lab, hybrid, and natural) when there are networks of interactions between subjects.

Figures

Figures reproduced from arXiv: 2506.11313 by the authors.

**Figure 1.** E[Yi | (2, 3)] = E[Yi | (3, 3)] (Left) and E[Yi | (2, 3)] ̸= E[Yi | (3, 3)] (Right) imagine that a medium and high amount of advertizing have the same impact, but that they always have a higher impact than a low amount. There could also be an interaction effect where the higher levels of dosage of both have a much greater impact than the sum of the effects. This is pictured in the left-hand panel of Figure [PITH_FU… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 7 canonical work pages

[1]

Diffusion in networks and the virtue of burstiness,

Akbarpour, M. and M. O. Jackson(2018): “Diffusion in networks and the virtue of burstiness,”Proceedings of the National Academy of Sciences, 115, E6996–E7004. Angrist, J. D.(2014): “The perils of peer effects,”Labour Economics, 30, 98–108. Angrist, J. D., G. W. Imbens, and D. B. Rubin(1996): “Identification of causal effects using instrumental variables,”...

arXiv 2018
[4]

Randomization tests for peer effects in group formation experiments,

Basse, G., P. Ding, A. Feller, and P. Toulis(2024): “Randomization tests for peer effects in group formation experiments,”Econometrica, 92, 567–590. Beaman, L. A.(2012): “Social Networks and the Dynamics of Labour Market Outcomes: Evidence from Refugees Resettled in the U.S. ”Review of Economic Studies, 79 (1), 128–161. 55 Bramoull´e, Y., H. Djebbari, and...

arXiv 2024
[11]

The 1/d law of giving,

Goeree, J. K., M. A. McConnell, T. Mitchell, T. Tromp, and L. Yariv(2010): “The 1/d law of giving,”American Economic Journal: Microeconomics, 2, 183–203. 58 Goeree, J. K., A. Riedl, and A. Ule(2009): “In search of stars: Network formation among heterogeneous agents,”Games and Economic Behavior, 67, 445–466. Granovetter, M.(1973): “The strength of weak tie...

work page 2010
[163]

The Doughboys Network: Social Interactions and Labor Market Outcomes of World War I Veterans,

Laschever, R. A.(2011): “The Doughboys Network: Social Interactions and Labor Market Outcomes of World War I Veterans,”SSRN discussion paper 1205543. Leider, S., M. M. M ¨obius, T. Rosenblat, and Q.-A. Do(2009): “Directed altruism and enforced reciprocity in social networks,”The Quarterly Journal of Economics, 124, 1815–

work page 2011
[399]

Testing unilateral and bilateral link formation,

57 Comola, M. and M. Fafchamps(2014): “Testing unilateral and bilateral link formation,” The Economic Journal, 124, 954–976. Comola, M. and S. Prina(2021): “Treatment effect accounting for network changes,” Review of Economics and Statistics, 103, 597–604. Corbae, D. and J. Duffy(2008): “Experiments with network formation,”Games and Economic Behavior, 64,...

work page 2014
[582]

Growing stars: A laboratory analysis of network forma- tion,

Rong, R. and D. Houser(2015): “Growing stars: A laboratory analysis of network forma- tion,”Journal of Economic Behavior & Organization, 117, 380–394. Rubin, D. B.(1974): “Estimating causal effects of treatments in randomized and nonrandom- ized studies. ”Journal of Educational Psychology, 66,

work page 2015
[608]

Learning in networks: An experimental study,

Choi, S., D. Gale, and S. Kariv(2004): “Learning in networks: An experimental study,” Levine’s bibliography, UCLA Department of Economics. ——— (2009): “Social Learning in Networks: A Quantal Response Equilibrium Analysis of Experimental Data,”WP, May. ——— (2012): “Social learning in networks: a quantal response equilibrium analysis of exper- imental data,...

work page 2004
[688]

Randomization analysis of experimental data: The Fisher randomization test comment,

——— (1980): “Randomization analysis of experimental data: The Fisher randomization test comment,”Journal of the American statistical association, 75, 591–593. Sacerdote, B.(2001): “Peer effects with random assignment: Results for Dartmouth room- mates,”The Quarterly journal of economics, 116, 681–704. 61 Sheng, S.(2020): “A structural econometric analysis...

arXiv 1980

Show all 13 references

[704]

An experimental study of the coloring problem on human subject networks,

Kearns, M., S. Suri, and N. Montfort(2006): “An experimental study of the coloring problem on human subject networks,” inProceedings of the 7th ACM conference on Electronic commerce, 213–222. Kirchsteiger, G., M. Mantovani, A. Mauleon, and V. Vannetelbosch(2016): “Limited fars...

2006
[1360]

Using smartphones to collect behavioral data in psychological science: Op- portunities, practical considerations, and challenges,

Harari, G. M., N. D. Lane, R. Wang, B. S. Crosier, A. T. Campbell, and S. D. Gosling(2016): “Using smartphones to collect behavioral data in psychological science: Op- portunities, practical considerations, and challenges,”Perspectives on Psychological Science, 11, 838–854. He...

2016
[1417]

Changes in social network structure in response to exposure to formal credit markets,

54 Banerjee, A., E. Breza, A. G. Chandrasekhar, E. Duflo, M. O. Jackson, and C. Kinnan(2024a): “Changes in social network structure in response to exposure to formal credit markets,”Review of Economic Studies, 91, 1331–1372. Banerjee, A., E. Breza, A. G. Chandrasekhar, and B. ...

2024
[1851]

A weak law for moments of pairwise stable networks,

Leung, M. P.(2019): “A weak law for moments of pairwise stable networks,”Journal of Econometrics. Lewbel, A., X. Qu, and X. Tang(2024): “Ignoring measurement errors in social networks,” The Econometrics Journal, 27, 171–187. Ligon, E. and L. Schechter(2012): “Motives for shari...

2019 arXiv
[2013]

Selecting the Most Effective Nudge: Evidence from a Large-Scale Experiment on Immuniza- tion,

Banerjee, A., A. G. Chandrasekhar, S. Dalpath, E. Duflo, J. Floretta, M. O. Jackson, H. Kannan, F. Loza, A. Sankar, A. Schrimpf, and M. Shrestha(2023): “Selecting the Most Effective Nudge: Evidence from a Large-Scale Experiment on Immuniza- tion,”Econometrica, forthcoming. Ban...

2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Diffusion in networks and the virtue of burstiness,

Akbarpour, M. and M. O. Jackson(2018): “Diffusion in networks and the virtue of burstiness,”Proceedings of the National Academy of Sciences, 115, E6996–E7004. Angrist, J. D.(2014): “The perils of peer effects,”Labour Economics, 30, 98–108. Angrist, J. D., G. W. Imbens, and D. B. Rubin(1996): “Identification of causal effects using instrumental variables,”...

arXiv 2018

[2] [4]

Randomization tests for peer effects in group formation experiments,

Basse, G., P. Ding, A. Feller, and P. Toulis(2024): “Randomization tests for peer effects in group formation experiments,”Econometrica, 92, 567–590. Beaman, L. A.(2012): “Social Networks and the Dynamics of Labour Market Outcomes: Evidence from Refugees Resettled in the U.S. ”Review of Economic Studies, 79 (1), 128–161. 55 Bramoull´e, Y., H. Djebbari, and...

arXiv 2024

[3] [11]

The 1/d law of giving,

Goeree, J. K., M. A. McConnell, T. Mitchell, T. Tromp, and L. Yariv(2010): “The 1/d law of giving,”American Economic Journal: Microeconomics, 2, 183–203. 58 Goeree, J. K., A. Riedl, and A. Ule(2009): “In search of stars: Network formation among heterogeneous agents,”Games and Economic Behavior, 67, 445–466. Granovetter, M.(1973): “The strength of weak tie...

work page 2010

[4] [163]

The Doughboys Network: Social Interactions and Labor Market Outcomes of World War I Veterans,

Laschever, R. A.(2011): “The Doughboys Network: Social Interactions and Labor Market Outcomes of World War I Veterans,”SSRN discussion paper 1205543. Leider, S., M. M. M ¨obius, T. Rosenblat, and Q.-A. Do(2009): “Directed altruism and enforced reciprocity in social networks,”The Quarterly Journal of Economics, 124, 1815–

work page 2011

[5] [399]

Testing unilateral and bilateral link formation,

57 Comola, M. and M. Fafchamps(2014): “Testing unilateral and bilateral link formation,” The Economic Journal, 124, 954–976. Comola, M. and S. Prina(2021): “Treatment effect accounting for network changes,” Review of Economics and Statistics, 103, 597–604. Corbae, D. and J. Duffy(2008): “Experiments with network formation,”Games and Economic Behavior, 64,...

work page 2014

[6] [582]

Growing stars: A laboratory analysis of network forma- tion,

Rong, R. and D. Houser(2015): “Growing stars: A laboratory analysis of network forma- tion,”Journal of Economic Behavior & Organization, 117, 380–394. Rubin, D. B.(1974): “Estimating causal effects of treatments in randomized and nonrandom- ized studies. ”Journal of Educational Psychology, 66,

work page 2015

[7] [608]

Learning in networks: An experimental study,

Choi, S., D. Gale, and S. Kariv(2004): “Learning in networks: An experimental study,” Levine’s bibliography, UCLA Department of Economics. ——— (2009): “Social Learning in Networks: A Quantal Response Equilibrium Analysis of Experimental Data,”WP, May. ——— (2012): “Social learning in networks: a quantal response equilibrium analysis of exper- imental data,...

work page 2004

[8] [688]

Randomization analysis of experimental data: The Fisher randomization test comment,

——— (1980): “Randomization analysis of experimental data: The Fisher randomization test comment,”Journal of the American statistical association, 75, 591–593. Sacerdote, B.(2001): “Peer effects with random assignment: Results for Dartmouth room- mates,”The Quarterly journal of economics, 116, 681–704. 61 Sheng, S.(2020): “A structural econometric analysis...

arXiv 1980