Autocorrelated Errors in Experimental Data in the Language Sciences: Some Solutions Offered by Generalized Additive Mixed Models

Baayen, R. Harald; Rij, Jacolien van; Cat, Cécile De; Wood, Simon N.

doi:10.1007/978-3-319-69830-4_4

Cited by 95 publications

(114 citation statements)

References 14 publications

Supporting

Mentioning

112

Contrasting

Unclassified

Order By: Relevance

“…They emerge not only in the present data set, but have been observed for visual lexical decision (Mulder et al, 2014) as well as for word naming and for eeg data (Baayen et al, 2016b). If this interpretation is correct, penalized factor smooths are the appropriate statistical tool to use.…”

Section: The Kkl Datasetsupporting

confidence: 61%

“…Different words enjoy different popularity across the genders (see also Baayen et al, 2016b), and adjusting by-word intercepts for gender results in a tighter model fit. With respect to subject, we included by-subject factor smooths for session.…”

Section: The Baldey Datasetmentioning

confidence: 99%

“…As discussed in further detail in Baayen et al (2016b), since different subjects typically emerge with different degrees of autocorrelation, one would want to adjust the ρ parameter for each individual subject. Unfortunately, it is at present not known how to achieve this mathematically within the framework of the generalized additive mixed model.…”

Section: The Poems Datasetmentioning

confidence: 99%

See 2 more Smart Citations

The cave of shadows: Addressing the human factor with generalized additive mixed models

Baayen

Vasishth

Kliegl

et al. 2017

Journal of Memory and Language

197

201

View full text Add to dashboard Cite

Generalized additive mixed models are introduced as an extension of the generalized linear mixed model which makes it possible to deal with temporal autocorrelational structure in experimental data. This autocorrelational structure is likely to be a consequence of learning, fatigue, or the ebb and flow of attention within an experiment (the 'human factor'). Unlike molecules or plots of barley, subjects in psycholinguistic experiments are intelligent beings that depend for their survival on constant adaptation to their environment, including the environment of an experiment. Three data sets illustrate that the human factor may interact with predictors of interest, both factorial and metric. We also show that, especially within the framework of the generalized additive model, in the nonlinear world, fitting maximally complex models that take every possible contingency into account is ill-advised as a modeling strategy. Alternative modeling strategies are discussed for both confirmatory and exploratory data analysis.

show abstract

Section: The Kkl Datasetsupporting

confidence: 61%

Section: The Baldey Datasetmentioning

confidence: 99%

Section: The Poems Datasetmentioning

confidence: 99%

See 1 more Smart Citation

The cave of shadows: Addressing the human factor with generalized additive mixed models

Baayen

Vasishth

Kliegl

et al. 2017

Journal of Memory and Language

197

201

View full text Add to dashboard Cite

show abstract

“…In [15] it is recommended to evaluate the quality of gam models by analyzing the autocorrelation in the prediction error. Although this may work well when predicting time series of reaction times, which are indeed characterized by a first-order auto-regressive process that should -and can-be accounted for by a linear mixed effects model [16,17], the situation with EEG signals is very different.…”

Section: Mid-sentence Datamentioning

confidence: 99%

Comparing Different Methods for Analyzing ERP Signals

Mulder¹,

Bosch²,

Boves³

2016

Interspeech 2016

View full text Add to dashboard Cite

Event-Related Potential (ERP) signals obtained from EEG recordings are widely used for studying cognitive processes in spoken language processing. The computation of ERPs involves averaging over multiple participants and multiple stimuli. Especially with speech stimuli, which also evoke substantial exogenous excitation, even averaging within conditions results in pooling many sources of variance. This raises questions about the statistical processing needed to uncover reliable differences between conditions. In this study we investigate differences between ERPs when participants listened to full and reduced pronunciations of verb forms in Dutch, in isolation and in mid-sentence position. Conventional statistical analysis uncovers some (but not all) differences between full and reduced forms in isolation, but not in mid-sentence position. In this paper, we show that linear mixed models (lmer) and generalized additive models (gam), which are able to account for participant-and stimulus-related variance may uncover more effects than conventional statistical models. However, depending on the complexity of the data, lmer and gam models may not be able to fit the data closely enough to warrant blind interpretation of the summary output. We discuss opportunities and threats of these approaches to analyzing ERP signals.

show abstract

“…It has been observed that the residual error in sequences of RTs in mixed effects models may contain a substantial autocorrelation [18]. That may even be the case if the model has the RT to the previous stimulus (RT previous), which is supposed to capture what is known as 'local speed effects' [19] in RT sequences, as one of the predictors.…”

Section: Simulationmentioning

confidence: 99%

Combining Data-Oriented and Process-Oriented Approaches to Modeling Reaction Time Data

Bosch¹,

Boves²,

Ernestus

2016

Interspeech 2016

View full text Add to dashboard Cite

This paper combines two different approaches to modeling reaction time data from lexical decision experiments, viz. a dataoriented statistical analysis by means of a linear mixed effects model, and a process-oriented computational model of human speech comprehension.The linear mixed effect model is implemented by lmer in R. As computational model we apply DIANA, an end-to-end computational model which aims at modeling the cognitive processes underlying speech comprehension. DIANA takes as input the speech signal, and provides as output the orthographic transcription of the stimulus, a word/non-word judgment and the associated reaction time. Previous studies have shown that DIANA shows good results for large-scale lexical decision experiments in Dutch and North-American English.We investigate whether predictors that appear significant in an lmer analysis and processes implemented in DIANA can be related and inform both approaches. Predictors such as 'previous reaction time' can be related to a process description; other predictors, such as 'lexical neighborhood' are hard-coded in lmer and emergent in DIANA. The analysis focuses on the interaction between subject variables and task variables in lmer, and the ways in which these interactions can be implemented in DIANA.

show abstract

Autocorrelated Errors in Experimental Data in the Language Sciences: Some Solutions Offered by Generalized Additive Mixed Models

Cited by 95 publications

References 14 publications

The cave of shadows: Addressing the human factor with generalized additive mixed models

The cave of shadows: Addressing the human factor with generalized additive mixed models

Comparing Different Methods for Analyzing ERP Signals

Combining Data-Oriented and Process-Oriented Approaches to Modeling Reaction Time Data

Contact Info

Product

Resources

About