The problem of inference from curves based on group data.

By W. Estes. 1956

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Can you Trust the Trend?: Discovering Simpson's Paradoxes in Social Data (2018, background): “Data heterogeneity also impacts statistical analysis of data [7] and causal inference [21].”
How algorithmic popularity bias hinders or promotes quality (2017, neutral): “This model could accurately capture aggregate behavior even if individuals followed different selection schemes [8].”
Belief and recollection of autobiographical memories (2003, methodology): “For the within-subjects analyses, regressions were done for each subject separately, and the parameters of the resulting equations were averaged (see Estes, 1956, for a discussion of averaging such linear equations).”
A cognitive latent variable model for the simultaneous analysis of behavioral and personality data. (2014, background): “As has been pointed out by Estes (1956, 2002), Hamaker (2012), and Heathcote, Brown, and Mewhort (2000), averaging artifacts can lead to biased estimates and errors in inference.”
On the shape of the probability weighting function. (1999, background): “If there is substantial individual variability, then some of the properties that have been observed in previous studies at the aggregate level may not hold at the level of individuals (see Estes, 1956, for a related discussion).”
An introduction to Bayesian hierarchical models with an application in the theory of signal detection (2005, background): “For example , Estes (1956) warned about aggregating learning curves across participants.”
The power law repealed: The case for an exponential law of practice (2000, result): “Our findings reinforce long standing analytic results that the learning curve for an arithmetic average need not have the same form as the functions contributing to the average (e.g., Estes, 1956; Sidman, 1952).”
One Hundred Years of Forgetting : A Quantitative Description of Retention (1996, background): “It could be argued that there is some theoretical (Estes, 1956) or practical (Newell & Rosenbloom, 1981) reason to transform the data or not to (Slamecka & McElree, 1983), but transformations do not make for a fair comparison among different functions.”
Methodological and empirical developments for the Ratcliff diffusion model of response times and accuracy (2009, background): “…As Bill Estes and others have shown, however, individual differences that are ignored can lead to so-called averaging artifacts in which the data that are averaged over participants are no longer representative for any of the participants (Estes, 1956, 2002; Heathcote, Brown, & Mewhort, 2000).”
Facets of personality and the phenomenology of autobiographical memory (2004, methodology): “These regressions were done over the 15 memories for each subject separately, and the parameters of the resulting equations were averaged (see Estes, 1956, for a discussion of averaging such linear equations).”
The Science of the Individual (2013, background): “At worst, they are statistical artifacts that represent nobody (Estes, 1956; von Eye, 2009)!”
Dynamic Development of Action and Thought (2007, background): “Often the deviations are ignored, as researchers mistakenly use methods that sum across individuals, activities, and contexts and treat true variations in level as errors of measurement (Estes, 1956; Fischer, Knight, & Van Parys, 1993; Skinner, 1938; van Geert & van Dijk, 2002).”
Response Errors Explain the Failure of Independent-Channels Models of Perception of Temporal Order (2012, background): “…goodness-of-fit analyses because there is no reason that the model should hold for a given sample as a whole (or just for 95% of the samples when the Type-I error rate is 0.05) but also because analyzing data aggregated across observers poses serious problems (Estes, 1956; Estes and Maddox, 2005).”
Default Bayes factors for ANOVA designs (2012, background): “Estimating learning curves has been a particularly vexing problem; although averaging across participants seems attractive to reduce noise, the functional form of the averaged data may not accurately reflect the functional form of individuals (Estes, 1956).”
Repeated Measures Correlation (2017, background): “This aggregation may resolve the issue of non-independence but can produce misleading results if there are meaningful individual differences (Estes, 1956; Myung et al., 2000).”
Hierarchical Bayesian parameter estimation for cumulative prospect theory (2011, background): “In particular, Estes and others have shown that individual differences,when ignored, can induce averaging artifacts inwhich the data that are averaged over participants are no longer representative for any of them (Estes, 1956, 2002; Heathcote, Brown, & Mewhort, 2000).”
Person-Oriented and Variable-Oriented Research: Concepts, Results, and Development (2006, background): “The term ecological fallacy is used to describe the situation in which “aggregated data are analyzed and the results are assumed to apply to relationships at the individual level” (Everitt, 1998, p. 110). One well-known example of the ecological fallacy was first published by Robinson (1950). The au”