Abstract
For several decades now, a new inductive method, meta-analysis, is all the rage in social and medical sciences. Meta-analyses, that is, statistical reviews of the results of primary studies concerning a test hypothesis, set new standards of excellence on what counts as strong evidence. In the current prevailing mood in medical and behavioural sciences, it is only a properly conducted, up-to-date meta-analysis that licenses detachment of hypotheses from the host of evidential claims made in individual studies, which claims may be inconclusive or contradictory with each other. My goal in this chapter is to see the extent to which judgment aggregation methods subsume meta-analytic ones. To this end, I derive a generalized version of the classical Condorcet Jury Theorem, and I contend that one can model at least some meta-analytic procedures using this theorem.
This research was funded by BU research funds BAP6021.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
See, for instance, (Fiske 1983).
- 2.
In the EBM hierarchy of types of evidence, a systematic review (SR) of randomized trials is ranked topmost—“SRs, by combining all relevant randomized trials, further reduce both bias and random error and thus provide the highest level of evidence currently achievable about the effects of health care.” (Sackett et al. 2000, 134).
- 3.
- 4.
As Nancy Cartwright observes, “In physics there is a rich network of knowledge and a great deal of connectedness so that any one hypothesis will have a large number of different consequences by different routes to which it is answerable. This is generally not true of hypotheses in the social sciences. In social sciences, we need techniques to export conclusions from where they are confirmed to across the board” (Cartwright 2007, 74).
- 5.
I discuss the implications of this condition further below.
- 6.
- 7.
These results follow from the law of large numbers, as shown in the appendix.
- 8.
Even though A n also depends on c, so that a better notation would be A n,c , I keep that dependence implicit in order to simplify the notation.
- 9.
See appendix for the proof.
- 10.
This follows from the fact that A n is unbiased if and only if in the long run it indicates with nearly unit probability that H is (not) the case given H is (not) the case. The latter is equivalent to requiring that the results (i) and (iv) hold, while the results (ii) and (iii) do not.
- 11.
An exception is the work of Haenni and Hartmann (2006), where one of the model types in their typology of partially reliable information sources, namely the model (CD) in section 4.3, includes a special case of the generalized CJT here, namely when the reliabilities and the powers are each constant, but not necessarily identical with each other, and when c = 0 or c = 1.
- 12.
That a rational agent should do so can be argued on the basis of Lewis’s Principal Principle (Lewis 1980). Of course, not being a theorem of the probability theory, the Principal Principle is an additional assumption available to those who seek to establish a plausible connection between degrees of belief and chances.
- 13.
Another way to model this situation is suggested in (Dietrich and List 2004). Instead of modeling the meta-analytic procedure as an evaluation of the unlimited potential evidence the world discloses to the researchers, we can model it as an evaluation of the body of possible evidence with constraints deriving from paradigmatic commitments. The difference amounts to determining whether the world was the immediate cause of the results S i’s or whether the evidential framework E (including the background assumptions B) preempts the world’s input. If the latter is the case, then it can be inserted as a new random variable E between the state of the world {H, ¬H} and the study outcomes S i ’s in such a way that E screens off the former in S i ’s ancestry. In this case, the convergence results mentioned in my model should be modified so that they reveal in the limit the misleading role of E.
- 14.
See (Hunt 1997, 43).
- 15.
See (Hawthorne 2001) for a similar point about jury interaction in the applications of CJT.
- 16.
See, for instance (Higgins 2009).
- 17.
This situation is quite frequently the case in clinical research. As (Ioannidis 2005) notes, 16% of highly cited primary research was plainly contradicted by subsequent research, and another 16% was found to have exaggerated the effects of medical interventions.
- 18.
Cooper (1994, 24) distinguishes between “review-generated evidence,” and the “study-generated evidence” to highlight the significance of the former. Review-generated evidence may include the gender of the researchers, the publication dates (of primary studies), study design, etc.
- 19.
Tukey’s project was precisely to de-emphasize formal mathematics in data analysis. See (Donoho 2000) for an assessment of Tukey and the future of data analysis.
References
Cartwright, Nancy. 2007. Hunting causes and using them. Cambridge: Cambridge University Press.
Cooper, Harris, and L.V. Hedges, eds. 1994. The handbook of research synthesis. New York: Russell Sage Foundation.
Dietrich, Franz, and Christian List. 2004. A model of jury decisions where all jurors have the same evidence. Synthese 142(2): 175–202.
Donoho, David L. 2000. High-dimensional data analysis: The curses and blessings of dimensionality. Lecture delivered at the American Mathematical Society Conference, “Mathematical Challenges of the Twenty-first Century”. http://www-stat.stanford.edu/~donoho/Lectures/AMS2000/AMS2000.html
Duval, S., and R. Tweedie. 2000. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 56: 455–463.
Fiske, Donald W. 1983. The meta-analytic revolution in outcome research. Journal of Consulting and Clinical Psychology 51(1): 65–70.
Haenni, R., and S. Hartmann. 2006. Modeling partially reliable information sources: A general approach based on Dempster-Shafer theory. Information Fusion 7: 361–379.
Hammersley, Martyn. 2001. On ‘systematic’ reviews of research literatures: A ‘narrative’ response to Evans&Benefield. British Educational Research Journal 27(5): 543–554.
Hawthorne, James, unpublished manuscript (circulated beginning 2001). Voting in search of the public good: The probabilistic logic of majority judgments. http://faculty-staff.ou.edu/H/James.A.Hawthorne-1/Hawthorne-{}-Jury-Theorems.pdf
Hedges, Larry V., and Ingram Olkin. 1985. Statistical methods for meta-analysis. San Diego, CA: Academic.
Higgins, J.P.T., and S. Green, eds. 2009. Cochrane handbook for systematic reviews of interventions Version 5.0.2 [updated September 2009]. The Cochrane Collaboration, 2009. http://www.cochrane-handbook.org.
Hunt, Morton. 1997. How science takes stock: The story of meta-analysis. New York: Russell Sage Foundation.
Hunter, J.E., and F.L. Schmidt. 1990. Methods of meta-analysis. Newbury Park: Sage.
Ioannidis, John P.A. 2005. Contradicted and initially stronger effects in highly cited clinical research. JAMA 294(2): 218–228.
Lewis, D. 1980. A subjectivist’s guide to objective chance. In Studies in inductive logic and probability, Vol. II, ed. Richard C. Jeffrey. Berkeley, CA, and Los Angeles: University of California Press.
Light, Richard J., and David B. Pillemer. 1984. Summing up: The science of reviewing research. Cambridge, MA, and London: Harvard University Press.
Resnick, Sidney I. 1998. A probability path. Basel: Birkhauser Verlag AG.
Rubin, Donald B. 1992. Meta-analysis: Literature synthesis or effect-size surface estimation? Journal of Educational and Behavioral Statistics 17: 363–374.
Sackett, David L. et al. 2000. Evidence-based medicine. Edinburgh: Churchill Livingstone.
Schmidt, Frank L. 1996. Statistical significance testing and cumulative knowledge in psychology: Implications for training of researchers. Psychological Methods 1(2): 115–129.
Smith, Mary L., and Gene V. Glass. 1977. Meta-analysis of psychotherapy outcome studies. American Psychologist 32(9): 752–760.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Appendix
Proof of (A). This result amounts to showing that \(\frac{{S(n)}}{n} - r\) converges to 0 in probability as n tends to infinity, given H. One may show it by applying a general weak law of large numbers, such as Theorem 7.2.1 in (Resnick 1998, 205), to the special setup of this result. One may also prove it as follows, noting that
Since we assume that \(r = \lim _{n \to \infty } \frac{{r_1 + \ldots + r_n }}{n}\) exists, the second term of the above sum converges to 0 as n goes to infinity. The first term can be shown to converge in probability to zero, since using Chebychev’s inequality, we have:
Here we have used the assumption that Si’s are independent from each other, as well as the fact that this implies that the (Si–ri)’s are also independent from each other, given H. Now, it is easy to see that for each i, \({\textrm{Var(S}}_{\textrm{i}} - {\textrm{r}}_{\textrm{i}} {)} = {\textrm{r}}_{\textrm{i}} (1 - {\textrm{r}}_{\textrm{i}} ) \le 1\), so that the above probability is less than or equal to \(\frac{n}{{\varepsilon ^2 n^2 }} = \frac{1}{{\varepsilon ^2 n}}\), and this quantity approaches to 0 as tends to infinity. This shows that \(\frac{{S(n)}}{n} - r\) can be written as the sum of two random variables each of which converges to 0 in probability as n tends to infinity, proving the same for the sum.
Proof of (i): Note that when c < r, we have the following:
When in addition (1) holds, then the result (A), the weak law of large numbers with \(\varepsilon = r - c > 0\) applies, and thus the last expression converges to 1 when n tends to infinity. This establishes the same for \({\textrm{P(A}}_{\textrm{n}} {\textrm{ = 1|H)}}\).
The proofs of the remaining clauses are very similar.
Rights and permissions
Copyright information
© 2012 Springer Science+Business Media B.V.
About this paper
Cite this paper
Kilinc, B. (2012). Meta-analysis as Judgment Aggregation. In: de Regt, H., Hartmann, S., Okasha, S. (eds) EPSA Philosophy of Science: Amsterdam 2009. The European Philosophy of Science Association Proceedings, vol 1. Springer, Dordrecht. https://doi.org/10.1007/978-94-007-2404-4_12
Download citation
DOI: https://doi.org/10.1007/978-94-007-2404-4_12
Published:
Publisher Name: Springer, Dordrecht
Print ISBN: 978-94-007-2403-7
Online ISBN: 978-94-007-2404-4
eBook Packages: Humanities, Social Sciences and LawPhilosophy and Religion (R0)