Estimation for Better Inference in Neuroscience

Calin-Jageman, Robert; Cumming, Geoff

doi:10.1523/eneuro.0205-19.2019

Cited by 119 publications

(90 citation statements)

References 83 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For example, a retrospective power analysis from the social isolation data obtained, using Glimmpse 48 , estimated a power of 0.83 to detect a genotype x drug interaction at p < 0.05 by ANOVA. Further, in accordance with recent and evolving recommendations 38,39,49 , we include information on the confidence intervals for the main effects, and also some preliminary Bayesian analysis. The use of Cauchy priors for the Bayesian analysis might be seen as over-conservative, considering the previous reports of hyperactivity and hyposociability phenotypes in 16p11.2 DEL mice 21,30,31,50 .…”

Section: Discussionmentioning

confidence: 99%

Drug-responsive autism phenotypes in the 16p11.2 deletion mouse model: a central role for gene-environment interactions

Mitchell

Thomson

Openshaw

et al. 2020

Sci Rep

View full text Add to dashboard Cite

There are no current treatments for autism, despite its high prevalence. Deletions of chromosome 16p11.2 dramatically increase risk for autism, suggesting that mice with an equivalent genetic rearrangement may offer a valuable model for the testing of novel classes of therapeutic drug. 16p11.2 deletion (16p11.2 DEL) mice and wild-type controls were assessed using an ethological approach, with 24 h monitoring of activity and social interaction of groups of mice in a home-cage environment. The ability of the excitation/inhibition modulator N-acetyl cysteine (NAC) and the 5-HT 1B/1D/1F receptor agonist eletriptan to normalise the behavioural deficits observed was tested. 16p11.2 DEL mice exhibited largely normal behaviours, but, following the stress of an injection, showed hyperlocomotion, reduced sociability, and a strong anxiolytic phenotype. The hyperactivity and reduced sociability, but not the suppressed anxiety, were effectively attenuated by both NAC and eletriptan. The data suggest that 16p11.2 DEL mice show an autism-relevant phenotype that becomes overt after an acute stressor, emphasising the importance of gene-environmental interactions in phenotypic analysis. Further, they add to an emerging view that NAC, or 5-HT 1B/1D/1F receptor agonist treatment, may be a promising strategy for further investigation as a future treatment. Autism is extremely common, affecting males more than females (estimated to affect roughly 4/1,000 boys and 1/1,000 girls) 1,2 , and characterised by communication difficulties, social dysfunction, and repetitive or restricted behaviour patterns, with a high rate of comorbid conditions such as anxiety. There are no available drug treatments for autism. Development of improved treatments will only be enabled by increased understanding of the causes of the disease and how they impact on neurobiology, informed by better preclinical models of facets of the disease. The genetic architecture of autism is complex 3. While a large number of common sequence variations increase disease risk, each has only a very small effect individually, and it is the cumulative burden of a range of risk, and protective, gene variants that underlies the aetiological mechanisms ultimately resulting in the manifestation of the common, sporadic disease. However, it is now clear that very rare copy number variants (CNVs), where small numbers of genes are present in one or three, rather than two, copies, substantially increase disease risk. For example, carriers of the deletions of the 16p11.2 locus have dramatically increased risk of autism-spectrum disorders (ASD) and also intellectual disability (8-40x) 4-6. Interestingly, a high proportion of carriers of the corresponding 16p11.2 duplication develop schizophrenia, suggesting that studying the neurobiological impact of CNVs at this locus may be particularly informative. The 16p11.2 deletion is one of the most powerful genetic risk factors for autism 3. Various drug classes have been tested in mouse models relevant to ASD, for efficacy in reversing behavioural...

show abstract

Section: Discussionmentioning

confidence: 99%

Drug-responsive autism phenotypes in the 16p11.2 deletion mouse model: a central role for gene-environment interactions

Mitchell

Thomson

Openshaw

et al. 2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…We need your feedback. Finally, there is another dimension to the estimation approach: the way we plan our research, which I leave you to discover in the Commentary (Calin-Jageman and Cumming, 2019) or the book.…”

Section: How To Move Forwardmentioning

confidence: 99%

“…It is linked to a Commentary published simultaneously (Calin-Jageman and Cumming, 2019), which describes the benefits of the estimation approach. When I was thinking to propose this change in the journal policy, I had a paper under revision at eNeuro (Manouze et al, 2019). I seized the opportunity to change the way I was reporting the results, using the estimation approach.…”

mentioning

confidence: 99%

Changing the Way We Report, Interpret, and Discuss Our Results to Rebuild Trust in Our Research

Bernard

2019

eNeuro

View full text Add to dashboard Cite

eNeuro is moving forward with a new initiative asking authors to present their results with estimation statistics and not to rely solely on p values. In this editorial, I would like to introduce to you the concept of this new statistics while first discussing my evaluation of the present situation and my own experience with using statistics to interpret results, then I will propose a solution and how we will move forward in the journal. I have also included my own experience using these new statistics and provided a list of resources. This new initiative will not change what is already acceptable for statistics in the journal; it is to encourage a simple addition of using estimation statistics.

show abstract

“…If there is no basis for justifying an effect size, consider if one only need estimate (e.g. derive means and confidence or credibility intervals) to draw the inferences needed (Calin-Jageman, & Cumming, 2019;Cumming & Calin-Jageman, 2017;Gelman, Carlin, Stern, Dunson, et al, 2013;McElreath, 2016;Rothman & Greenland, 2018;Wagenmakers, Marsman, Jamil, Ly et al, 2018). Remember that if estimation is used, it gives no grounds for asserting H0; one can only say the effect is (or is probably) between certain bounds.…”

Section: The Rough Scale Of Effect Predictedmentioning

confidence: 99%

Obtaining evidence for no effect

Dienes¹

2020

Preprint

View full text Add to dashboard Cite

Obtaining evidence that something does not exist requires knowing how big it would be were it to exist. Testing a theory that predicts an effect thus entails specifying the range of effect sizes consistent with the theory, in order to know when the evidence counts against the theory. Indeed, a theoretically relevant effect size must be specified for power calculations, equivalence testing, and Bayes factors in order that the inferential statistics test the theory. Specifying relevant effect sizes for power, or the equivalence region for equivalence testing, or the scale factor for Bayes factors, is necessary for many journal formats, such as registered reports, and should be necessary for all articles that use hypothesis testing. Yet there is little systematic advice on how to approach this problem. This article offers some principles and practical advice for specifying theoretically relevant effect sizes for hypothesis testing.

show abstract

Estimation for Better Inference in Neuroscience

Cited by 119 publications

References 83 publications

Drug-responsive autism phenotypes in the 16p11.2 deletion mouse model: a central role for gene-environment interactions

Drug-responsive autism phenotypes in the 16p11.2 deletion mouse model: a central role for gene-environment interactions

Changing the Way We Report, Interpret, and Discuss Our Results to Rebuild Trust in Our Research

Obtaining evidence for no effect

Contact Info

Product

Resources

About