Sociological Methods & Research

The credibility of quantitative social science benefits from policies that increase confidence that results reported by one researcher can be verified by others. Concerns about replicability have increased as the scale and sophistication of analyses increase the possible dependence of results on subtle analytic decisions and decrease the extent to which published articles contain full descriptions of methods. The author argues that sociology should adopt standards regarding replication that minimize its conceptualization as an ethical and individualistic matter and advocates for a policy in which authors use independent online archives to deposit the maximum possible information for replicating published results at the time of publication and are explicit about the conditions of availability for any necessary materials that are not provided. The author responds to several objections that might be raised to increasing the transparency of quantitative sociology in this way and offers a candidate replication policy for sociology.

The author introduces a set of integrated developments in Web application software, networking, data citation standards, and statistical methods designed to increase scholarly recognition for data contributions; to put some of the universe of data and data-sharing practices on firmer ground; and to facilitate the public distribution of persistent, authorized, and verifiable data, with powerful and easy-to-use technology, even when the data are confidential or proprietary. The goal is to solve some of the political and sociological problems of data sharing via technological means, with the result intended to benefit both the scientific community and the sometimes apparently contradictory goals of individual researchers.

Jeremy Freese makes the case for data sharing as a condition of publication for quantitative research in sociology, and Gary King tells us of a Dataverse Network under construction that is designed to routinize the process of posting and storing such data sets. No matter how user-friendly that network turns out to be, it is clear that no system is entirely cost-free, either for researchers or for journal editors. It is important, then, to determine whether the benefits of mandatory data sharing (or ``data relinquishment,'' as Herrnson calls it) would outweigh the costs. In this comment, the author discusses the issue from his vantage point as a former editor and concludes that the benefits of such a requirement most likely would exceed the costs.

This comment argues that although replication will and should gain ground in sociology, that process will be complicated by issues of ownership, mechanics, and security. Replicationism will also change the economy of peer review. Ironically, it could also reveal that sociologists have less agreement on methodological issues than we think.

Commentators appreciate the benefits of developing improved standards for replicating quantitative results in sociology. Nonetheless, reservations remain, and the author addresses several of them and explains why improved replication standards do not endanger participant confidentiality, do not undermine incentives for collecting data, do not need to wait for greater standardization of data formats, need not require that editors assign a reviewer to actually replicate results, and do not diminish methodological diversity in any positive sense. The author concludes by encouraging sociologists to find a way of moving beyond intermittent discussions of replication standards to collective action.

Least absolute deviation (LAD) is a well-known criterion to fit statistical models, but little is known about LAD estimation in structural equation modeling (SEM). To address this gap, the authors use the LAD criterion in SEM by minimizing the sum of the absolute deviations between the observed and the model-implied covariance matrices. Using Monte Carlo simulations, the authors compare the performance of this LAD estimator along several dimensions (bias, efficiency, convergence, frequencies of improper solutions, and absolute percentage deviation) to the full information maximum likelihood (ML) and unweighted least squares (ULS) estimators in structural equation modeling. The results for LAD are mixed: There are special conditions under which the LAD estimator outperforms ML and ULS, but the simulation evidence does not support a general claim that LAD is superior to ML and ULS in small samples.

Randomized response (RR) is an interview technique designed to eliminate response bias when sensitive questions are asked. In RR the answer depends partly on the true status of the respondent and partly on the outcome of a randomizing device. Although RR elicits more honest answers than direct questions do, it is susceptible to self-protective response behavior; that is, the respondent gives an evasive answer irrespective of the outcome of the randomizing device. The authors present a log-linear RR model that accounts for this kind of self-protection (SP). The main results of this SP model are estimates of (1) the probability of SP, (2) the log-linear parameters describing the associations between the sensitive characteristics, and (3) the prevalence of the sensitive characteristics that are corrected for SP. The model is illustrated with two examples from a Dutch survey measuring noncompliance with social welfare rules.