MetaVoice is a research tool, bridging the gap between research fields investigating voice patterns in different neuropsychiatric disorders. At the moment, MetaVoice contains studies on voice patterns within 1 domain(s): Voice Analysis. Within this domain, there are currently datasets related to 4 disorders: Autism Spectrum Disorder, Right Hemisphere Damage, Left Hemisphere Damage and Schizophrenia. These datasets include 399 effect sizes from 73 papers, including 3359 subjects.
These aggregated data help to acquire better estimates of effect sizes across disorders, languages of participants and experimental methods. Thereby, it also becomes possible to explore the impact of potential moderators. The power analysis tool allows researchers to plan future studies with adequate sample sizes.
Datasets tab.Statistical Approach tab.Field Specification tab.Contributing to MetaVoice tab.Export Data tab.Tutorials tab. Please note that data and visualisations are under development at the moment (Winter 2020) and should not be taken as definitive.
On this page you can find all the datasets currently available on the MetaVoice website. By clicking on the links, more information about the datasets can be found.
Citation: Parola, A., Simonsen, A., Bliksted, V., & Fusaroli, R. (2020). Voice patterns in schizophrenia: A systematic review and Bayesian meta-analysis. Schizophrenia Research, 216, 24-40.
N papers = 27, N effect sizes = 81
Curator: LNJ, ND
Comments: Dataset from the original MA can be found on OSF, as described in the original MA paper.
Voice Analysis: Autism Spectrum Disorder
Citation: Fusaroli, R., Lambrechts, A., Bang, D., Bowler, D. M., & Gaigg, S. B. (2017). Is voice a marker for Autism spectrum disorder? A systematic review and meta‐analysis. Autism Research, 10(3), 384-407.
N papers = 23, N effect sizes = 169
Curator: LNJ, ND
Comments: Dataset from the original MA can be found on GitHub, as described in the original MA paper.
Voice Analysis: Right Hemisphere Damage
Citation: Weed, E., & Fusaroli, R. (2020). Acoustic Measures of Prosody in Right-Hemisphere Damage: A Systematic Review and Meta-Analysis. Journal of Speech, Language, and Hearing Research, 1-14.
N papers = 16, N effect sizes = 107
Curator: LNJ, ND
Comments: Dataset from the original MA can be found on OSF, as described in the original MA paper. SDs of 0 were replaced through data imputation.
Voice Analysis: Left Hemisphere Damage
Citation: Weed, E., & Fusaroli, R. (2020). Acoustic Measures of Prosody in Right-Hemisphere Damage: A Systematic Review and Meta-Analysis. Journal of Speech, Language, and Hearing Research, 1-14.
N papers = 7, N effect sizes = 42
Curator: LNJ, ND
Comments: Dataset from the original MA can be found on OSF, as described in the original MA paper. SDs of 0 were replaced through data imputation.
First of all, thank you for your support and your wish to contribute to MetaVoice.
There are several ways in which you can contribute to the website and its contents. In general, you are always welcome to suggest new data and/or ideas for the website. Moreover, you are also more than welcome to join our team. Please see below, and contact us if you are interested in contributing.
Suggest and/or add data to an existing meta-analysis on the website:
If you find or have conducted a study related to the field of voice patterns in neuropsychiatric disorders, which is not yet available on the MetaVoice website, you can suggest it as an addition to an already existing meta-analysis. This applies to data from both published and unpublished papers. How: Email the curator of the existing meta-analysis with information on the study you found, and preferably include a link to or a pdf of the full text. This also applies if you are the author of the study. Ideally (but not necessarily), make the data fit the MA template, or make sure that the necessary data are provided in the email. From here, we will let you know how we/you can add the data to the existing meta-analysis.
Suggest and/or add a new meta-analysis:
If you find or have conducted a meta-analysis within the field of voice patterns in neuropsychiatric disorders, which is not yet available on the MetaVoice website, you can suggest this as a new meta-analysis. If your meta-analysis is related to a new domain see Suggest a new domain to be added below. How: Email one of the team members with information on the meta-analysis, and preferably include a link to or a pdf of the full text (if published), or make sure that information about the data and the data itself is provided in the email (if unpublished). Ideally (but not necessarily), make the data fit the MA template.
Create a new meta-analysis: If you want to contribute by creating your own meta-analysis, please see the tutorial on the MetaLab website. This page will guide you through the different steps of creating a meta-analysis. Be aware that the tutorial page was created based on the field of early language acquisition and cognitive development, rather than voice patterns in neuropsychiatric disorders. Therefore, the MA template for voice patterns in neuropsychiatric disorders, and the Field specifications should be used when creating a meta-analysis within this field.
Suggest a new domain to be added: If you have an idea of a domain to be added which is related to neuropsychiatric disorders, you can contribute by suggesting this domain to us. How: Email one of the team members with your suggestion, including the relevance of the domain and possible meta-analyses related to that domain. Please include a link to or a pdf of the full text(s), and we will get back to you.
As a curator, you will receive an email once a year, listing all of the meta-analyses which you are responsible for, and asking you whether you will continue to be a curator.
Suggest website features: If you want to contribute to the website by adding new features, or have other good ideas, you can let us know through GitHub or contact one of the team members. Please make it clear, what you would change, and how you would implement this.
Replicate the website within your own field of interest: Another way in which you can contribute, is by advocating the underlying aim of the website. That is, promoting a framework emphasising open science, open code, the importance of cumulative science, etc. You can do this by replicating the website within your own field of interest, and thereby make cumulative science more open and accessible in your field. If you have questions, you can contact us.
If you have contributed with a meta-analysis of which you are the author, you might have some questions regarding the ownership of your data. Even though your meta-analysis is added to the website, you remain the owner of your meta-analysis data. Therefore, researchers using data from MetaVoice have to cite the publications related to your meta-analysis.
All analyses on the site are conducted with the metafor package (Viechtbauer, 2010).
Effect size computation is handled by a function called mv_compute_es, which can be found in the script mv_functions.R. This function is based on a script compute_es.R, which has been developed by MetaLab. MetaLab cites Hedges & Olkin’s textbook (Hedges & Olkin, 2014) as providing most of the formulas in the script.
The meta-analytic visualisations and power analysis tools rely on a multi-level random effects model to conduct the meta-analyses ( rma.mv function of metafor). Random-effect models allow for variability between studies and unique data points. To be specific, in our models a random intercept was defined for each sample (group of participants), since data from the same sample can be expected to be more similar than from a different sample. Further, a random intercept for each study was added, as data from the same study can be expected to be more similar than from a different study. This, thereby also accounts for cases in which the same sample contributed to data in two different studies.
On the meta-analytic visualisation, we display model summary outputs for two different models. Both of them follow the random-effect structure described above. The Summary of meta-analytic model with intercept displays the output for the model which is used for all visualisations provided and for calculation of the overall effect size. The estimates in the model output express the difference to the intercept. More specifically, the model has the following structure:
Effect Size ~ 1 + (Moderator1) + ... + (1 | Sample / Study / Unique Row)
If a moderator is chosen, we also display a Summary of meta-analytic model without intercept. The estimates in the model output here correspond to the absolute values for the single levels/factors of the chosen moderator. More specifically, the model has the following structure:
Effect Size ~ 0 + Moderator1 + ... + (1 | Sample / Study / Unique Row)
The meta-analytic models are also accessible in the script shinyapps/visualization/server.R or shinyapps/power_analysis/server.R.
MetaVoice aims to follow open science practices. Therefore, all of the datasets used on this website can be downloaded. Working with unknown datasets can be challenging. The following information is supposed to help you understand the MetaVoice dataset.
This page gives the full specification for each field in the MetaVoice dataset, including: required fields (which must be included for every MA), optional fields (which are only used for some MAs), and derived fields (which are computed by the site).
You can download the data from both the home page, visualisation page and the documentation page.
From the home page
1.1. Click on the box corresponding to the domain you are interested in.
1.2. Once you are on the page of the domain, click on the box corresponding to the dataset/meta-analysis you are interested in.
1.3. In the data tab, click on the “Download” button.
1.4. Choose the data format that you want (EXCEL - recommended for manual calculations using a spreadsheet software - but you should make sure that your local spreadsheet software is set to use “.” as decimal separator to avoid errors, or CSV - recommended for reading into a statistical software such as R) by clicking on it in the conditional menu.
From the visualisation page
2.1. Open the “Domain” drop-down menu and click on the one you are interested in. Currently only one domain is available.
2.2. Open the “Dataset” drop-down menu and click on the dataset/meta-analysis you are interested in.
2.3. Open the “Feature” drop-down menu and click on the feature you are interested in.
2.4. Click on the “Download data” button above the “Dataset” menu. This will download the data in CSV format. You can also click on “View raw dataset” and follow steps 1.2. and 1.3.
From the documentation page
3.1. Open the “Datasets” tab and click on the dataset you are interested in.
3.2. Click on the “Download data” button above the “Dataset” menu. You can also click on “View raw dataset” and follow steps 1.2. and 1.3.
3.3. Choose the data format that you want (EXCEL - recommended for manual calculations using a spreadsheet software - but you should make sure that your local spreadsheet software is set to use “.” as decimal separator to avoid errors, or CSV - recommended for reading into a statistical software such as R) by clicking on it in the conditional menu.
For a more detailed tutorial on how to open and work with the data in excel, see the MetaLab Tutorial. In case you are encountering problems with opening the data in excel, you can also find troubleshooting of common issues on the MetaLab Tutorial. If you cannot find help there, you are also welcome to email one of the team members.
Please cite publications linked to the dataset, as described in Publications
If you would like to learn more about Meta-Analyses and how to conduct them, we can recommend educational content provided by MetaLab.