<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Dan Quintana on Medium]]></title>
        <description><![CDATA[Stories by Dan Quintana on Medium]]></description>
        <link>https://medium.com/@dsquintana?source=rss-91d5a7b2343a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*-fo7kJU8CzV-dcF_.jpeg</url>
            <title>Stories by Dan Quintana on Medium</title>
            <link>https://medium.com/@dsquintana?source=rss-91d5a7b2343a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 07:36:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dsquintana/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How to start your own podcast]]></title>
            <link>https://medium.com/@dsquintana/how-to-start-your-own-podcast-8a7d0d6b79a5?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/8a7d0d6b79a5</guid>
            <category><![CDATA[podcast]]></category>
            <category><![CDATA[guides-and-tutorials]]></category>
            <category><![CDATA[audio-engineering]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Mon, 02 Mar 2020 20:58:08 GMT</pubDate>
            <atom:updated>2020-03-02T21:36:10.912Z</atom:updated>
            <content:encoded><![CDATA[<p>I love making podcasts. They work on my behalf around the clock-anyone with an internet connection can listen to my ideas and to topics that I care about at their convenience. They also take comparatively less time to make than blog posts, at least in my experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gwshwkm_b7rnpI2CjDPcjw.jpeg" /></figure><p>I’ve been podcasting for over three years as co-host and producer of <a href="https://everythinghertz.com/"><em>Everything Hertz</em></a>, so I’ve learnt a thing or two along the way. Given the increasing interest in podcasting, and the growing number of requests I get for a podcast guide, I’ve put together this blog post. It will walk you through all you need to know to start your own podcast, including hardware, software, recording, editing, hosting, fundraising, and promotion.</p><h3>Microphone</h3><p>If there’s one thing you should spend money on it’s a microphone. Sure, you could record using your desktop or smartphone’s built-in microphone, but the sound quality won’t be great. Check out this <a href="https://marco.org/podcasting-microphones">massive guide</a> on podcast microphones, which suits a range of budgets. If you’re just getting started, it’s best to begin with a USB microphone, as there’s much less hassle with these, compared to a XLR microphone.</p><p>A great starting microphone is the <a href="https://www.amazon.com/Samson-Handheld-Microphone-Recording-Podcasting/dp/B001R747SG/ref=sr_1_2?crid=5ANZU62MI3MY&amp;keywords=samson+q2u&amp;qid=1583147598&amp;sprefix=samson%2Caps%2C236&amp;sr=8-2">Samson Q2u</a>, which includes a USB output and is great value for money. Importantly, there’s also a headphone output built into this microphone (more on this later).</p><p>When it comes to microphones, one area in which your audience will give you a little grace is if your guest doesn’t have a dedicated microphone. I think this is because audiences are accustomed to TV and radio guests occasionally having poorer quality audio-but they’re not used to <em>hosts</em> having poor audio. Of course, it’s preferable if your guest has a dedicated microphone, but this shouldn’t stop you from interviewing them if they don’t have one.</p><h3>Headphones</h3><p>Your podcast headphones perform three primary functions:</p><ol><li>Hearing your co-host/guest</li><li>Hearing yourself</li><li>Preventing your microphone from picking up your guest’s audio</li></ol><p>If you’re starting out, they type of headphones you use doesn’t matter too much, but an important consideration if you’re working with guests or co-hosts is that you don’t want their voices to bleed out of your headphones into your microphone. This can be a pain to edit out. You either want to use “in ear” headphones, or “over the ear” headphones. Here’s <a href="http://www.audiogurus.com/learn/headphones/in-ear-on-ear-over-ear-headphones/208">an explainer</a> on what these terms mean if they’re new to you.</p><p>If you have a budget, I recommend <a href="https://www.amazon.com/Sony-MDR7506-Professional-Diaphragm-Headphone/dp/B000AJIF4E/ref=sr_1_3?keywords=Sony+MDR-7506&amp;qid=1581202715&amp;sr=8-3">Sony MDR-7506</a> headphones. These are a good option because they provide a very ‘neutral’ sound, compared to other headphones that add a lot of unnecessary bass. They’re also comfortable to wear over long periods.</p><h3>Software for recording audio</h3><p>If you’re on a Mac, you can use <a href="https://support.apple.com/quicktime">QuickTime</a> to record audio. This app comes with all Macs and it’s easy to use. If you’re on a PC, you can use the <a href="https://www.microsoft.com/en-us/p/windows-voice-recorder/9wzdncrfhwkn?irgwc=1&amp;OCID=AID2000142_aff_7593_159229&amp;tduid=(ir__3rhrklckl9kftmaukk0sohz3x22xlcqcxplbopcx00)(7593)(159229)()(UUwpUdUnU62766YYwYg)&amp;irclickid=_3rhrklckl9kftmaukk0sohz3x22xlcqcxplbopcx00&amp;activetab=pivot%3Aoverviewtab">Voice Recorder</a> app. I haven’t used this myself, but I understand it’s also straightforward to use.</p><h3>Working with guests and co-hosts</h3><p>You can use video conferencing software, such as Skype or Google Hangouts to chat to your guests/co-host and ask them to record the audio on their end. After the recording, they can send you their file which you can use for editing. This approach is often referred to as a ‘ <a href="https://podcastengineeringschool.com/what-is-a-double-ender/">double-ender</a>’, because the audio is recorded at both ends. <a href="https://wetransfer.com/">WeTransfer</a> is a good option for sending large audio files, but there are plenty of alternative available.</p><p>It’s also possible to record your guest’s audio, but this is risky if the video conferencing audio cuts out. So unless your guest is not comfortable recording their own audio, I would advise against this.</p><h3>Podcast hosting</h3><p>You’ll need to host your audio files somewhere in order to distribute your episodes to podcast directories. Along with your audio file, you will also need to include some standard information, like podcast title, episode titles, and episode descriptions. With these files and information, you can produce an RSS feed, which are sent to podcast directories, like Apple Podcasts or Google Podcasts.</p><p>This may seem a bit daunting, but there are many services that will do the technical legwork for you for relatively little money (or for free). All you have to do is supply the files and episode descriptions. Most podcast hosting services provide clear instructions on how get your podcasts on the main podcast directories.</p><p>There are several options available for hosting, and you generally get what you pay for. If you want to get started for free, you should check out <a href="https://www.acast.com/en/">Acast’s ‘starter’ tier</a>. If you have a budget, I recommend <a href="https://fireside.fm/">Fireside</a> ($19 per month), as you get a nice website for your podcast, it’s a pleasure to use, and you get great analytics regarding where your listeners are located and how many downloads each episode is getting.</p><h3>Podcast editing</h3><p>Editing is an important consideration for your podcast, but you don’t need expensive software to make your podcast sound good. Editing typically takes anywhere from 1.5 to 2 times the length of your episode, so factor this time in. Some people prefer to heavily edit the content of their podcasts, which is fine, but I prefer to leave the recording relatively untouched as this feels more like a natural conversation. It’s up to you.</p><p>There are six key audio elements that you should edit in your audio files. The following demonstrations primarily use iZotope RX elements ($129). However, some of these elements can addressed in Auphonic (free for up to two hours of processing a month), which is covered below. These elements are ordered from most important to least important, in my opinion.</p><h3>Editing raw audio files</h3><p><strong>1. Background hums and hisses</strong></p><p>Hums that can interfere with your recording can have various sources, including laptop/desktop fans, air conditioners, or other any other appliance that emits a hum. I could try and describe the impact of hums on your recording, but as we’re dealing with sounds this is much easier to demonstrate with a video clip.</p><p>Part 1 of the next video begins with an audio clip with no talking, so that you can hear the background hums and hisses, followed by some talking. Notice that you can still notice the hums and hisses during the talking segment. I’ve visualised the audio using a spectrogram via iZotope RX. In part 2, I remove the background noise by providing a sample of dead noise so that the software can get a ‘fingerprint’ of the background noise in this recording. I do this using the ‘Spectral De-noise’ module. With this fingerprint, I can then strip the background noise from the entire audio file. Notice how the spectrogram changes from the first sample. In part 3, I play the cleaned clip.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FoHSf2WFm0mo%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DoHSf2WFm0mo&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FoHSf2WFm0mo%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/c85f271289030c0afa0bd2e7831911d6/href">https://medium.com/media/c85f271289030c0afa0bd2e7831911d6/href</a></iframe><p><strong>2. Mouth clicks</strong></p><p>These mouth clicks (also known as mouth or lip smacking) are caused by saliva, and can annoy your listeners. In the next clip you can hear a clip with mouth clicks (Part 1), followed by an audio sample demonstrating how to clean these mouth clicks using the ‘Mouth De-click’ module in iZotope RX (Part 2) and what the clean version sounds like (Part 3). There’s one stubborn section that I need to go over again, by increasing the repair area (Part 4).</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FNOrtOt6PE3s%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DNOrtOt6PE3s&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FNOrtOt6PE3s%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/62125a740606bf1a677dc43af619e962/href">https://medium.com/media/62125a740606bf1a677dc43af619e962/href</a></iframe><p><strong>3. Consistent volume</strong></p><p>This is both in terms of within your recording and also in terms of volume compared to <em>other </em>podcasts. There’s actually a loudness standard for podcasts, so that they volumes between podcasts are roughly the same. You can fix this in Auphonic, using the Target loudness setting (more on this below).</p><p><strong>4. Plosives</strong></p><p>These are hard consonant sounds (especially the letters ‘p’ and ‘b’) that can produce a wind-like noise, which can irritate listeners. Using a pop/wind shield can help reduce these plosive sounds. In the next clip, you can hear an audio sample with a plosive (Part 1), followed by a demonstration of how to clean this plosive sound using the ‘De-plosive’ module in iZotope RX (Part 2), and what the clean version sounds like (Part 3).</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F_hEKlkaLGA8%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D_hEKlkaLGA8&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F_hEKlkaLGA8%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/dbff350a4acd1a506c7e5e136c72e7be/href">https://medium.com/media/dbff350a4acd1a506c7e5e136c72e7be/href</a></iframe><p><strong>5. Audio clipping</strong></p><p>This is when your audio level is <a href="https://thepodcastcreative.com/podcast-audio-breakdown-clipping-headroom-noise-floor/">too loud</a> for the capability and/or settings of your recording device. Clipping can be reduced in production, but you’re better off preventing this in the first place This can occur if you’re speaking too close to your mic, for instance. Most recorders will flash red when this occurs, to give you a visual warning. Make sure you test your levels and mic position before you recording, and keep an eye out for clipping during your recording.</p><p><strong>6. Esses</strong></p><p>Harsh ‘s’ noises can sound grating for many people. Good mic positioning can help reduce this, so try a few different positions. In the next clip, you can hear an audio sample with an ess sound (Part 1), followed by a clip demonstrating the isolation of the ess sound (Part 2). Here, you can adjust the detection sensitivity. Next, I remove the ess sound (Part 3) and then play the cleaned clip (Part 4).</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FpsnAscvrUw0%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DpsnAscvrUw0&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FpsnAscvrUw0%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/4dc3d0896f076aaad60ccebd29a53c65/href">https://medium.com/media/4dc3d0896f076aaad60ccebd29a53c65/href</a></iframe><h3>An overview of Auphonic for audio editing</h3><p>As mentioned above, two hours a month of audio processing with Auphonic is free, and up to 9 hours of processing costs $11 per month. You can purchase an unlimited personal licence for $89. Here are the main features of Auphonic:</p><ul><li>Target loudness: This adjusts the overall loudness of your podcast to the industry standard. Select the ‘Podcasts and Mobile’ option, obviously. Here’s more on <a href="https://auphonic.com/blog/2013/01/07/loudness-targets-mobile-audio-podcasts-radio-tv/">loudness standards</a>, if you’re interested</li><li>Adaptive leveler: Makes your podcast the same loudness throughout. This will help fix if your guest recorded their audio using a lower or higher volume</li><li>High pass filter: Removes low frequency interference</li><li>Noise and hum reduction: Removes annoying background hisses and hums</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L7kvIkbZvzpY0KzuqZceGA.jpeg" /></figure><p>Despite the availability of these tools, you’ll save yourself time by adopting good mic technique via monitoring your audio <em>during</em> your recording via your headphones. If you’re too close to your mic, you increase the risk of clipping or mouth clicks. Having the mic directly facing your mouth can also increase the risk of plosive sounds.</p><p>Here is a clip demonstrating how Auphonic performs when cleaning audio clips. It does a really good job considering Auphonic is automatic and free (Part 1 is the raw audio and Part 2 is after cleaning the audio using Auphonic). The background noises and mouth clicks are taken care of nicely The only thing that Auphonic misses is some of the harsh ess sounds, but I’m being a bit picky here.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FndbI3r5oFkY%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DndbI3r5oFkY&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FndbI3r5oFkY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/d458a0f5b032a19409cf27e86c3a8be8/href">https://medium.com/media/d458a0f5b032a19409cf27e86c3a8be8/href</a></iframe><h3>Putting your clean clips together</h3><p>Now that your clips have been cleaned, the second main part to editing is putting everything together and cutting sections you want to remove. You can use either GarageBand or Audacity for this. If you have the budget, <a href="https://hindenburg.com/">Hindenburg</a> is a great option, as it’s primarily designed for editing podcasts and radio.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mz7cO11no1y-KI0dmzlPHA.jpeg" /></figure><p>If you want to add royalty-free music, perhaps for an intro or outro, GarageBand has a large library. You can also use music from <a href="https://leerosevere.bandcamp.com/album/music-for-podcasts-6">Lee Rosevere</a> for free, with acknowledgment.</p><p>When you’re done with editing, make sure you export your file as an MP3. This is the preferred format for podcast hosting services, as these files aren’t too big. Exporting your file in mono format will also help reduce the file size.</p><p>So here’s a summary of a editing workflow that will cost you nothing:</p><ol><li>Record audio tracks as WAV files, using good mic habits to reduce clipping, mouth clicks, and plosives.</li><li>Put individual audio tracks through Auphonic, using the standard settings.</li><li>Put together audio in GarageBand/Audacity/Hindenberg, exporting as a WAV file.</li><li>Run this file through Auphonic, but only select target loudness and export as an MP3.</li></ol><h3>Interviewing guests</h3><p>You can use video conferencing software, such as Skype or Google Hangouts to chat to your guests and ask them to record the audio on their end. Of course, it’s possible to also record their audio, but when using Skype the recorded signal will include both your co-host and your guest. This can be harder to edit given the different recording conditions between you and your guest/co-host.</p><p>You can also record the video of your conversation using <a href="https://obsproject.com/">OBS</a> (free) or <a href="https://www.ecamm.com/mac/ecammlive/">eCamm Live</a> ($12 per month, Mac only). Some people prefer watching videos, so you can post your episode as a video on YouTube.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FTzMnlm_IcU4%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DTzMnlm_IcU4&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FTzMnlm_IcU4%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/745578a2319df46ae14eeab08cf227b0/href">https://medium.com/media/745578a2319df46ae14eeab08cf227b0/href</a></iframe><p>Even if you don’t post your entire episode, you can use short video clips to promote episodes on social media (more on this below).</p><p>When it comes to preparing for guests, this is up to you and the style of your podcast. Of course, you should respect the time of your guests and come prepared, but also consider a more freewheeling interviewing style for your podcast, if that suits you. It’s also good to give your guests a heads up on the sorts of questions you’ll be asking, so that they can prepare.</p><h3>Financial support</h3><p>Before I get into financial support, I want to remind you that it’s possible to create and produce a good podcast for $40.</p><p>Here’s a breakdown of hardware and software you’ll need:</p><ul><li>Microphone: Samson Q2u ($40)</li><li>Headphones: The ones you already own (free)</li><li>Recording: Quicktime/Voice Recorder App (free)</li><li>Editing: Garage band/Audacity (free) + Auphonic (free for 2 hours of audio processing per month)</li><li>Hosting: Acast (free)</li></ul><p>You should start considering financial support if you would like to improve your recording gear, software, and services that you use, but you don’t need this, especially for the first bunch of podcasts that you record.</p><p>If you would like to get additional financial support, there are two options. Direct support from your listeners (e.g., Patreon, merchandise) or advertising. Don’t even think about this until you’re up to episode 20, just focus on building momentum and on the podcast itself. For Everything Hertz, we didn’t start a Patreon account until around episode 75 and advertising and merchandise until episode 100. You need to demonstrate to yourself AND your potential supporters that your podcast has longevity.</p><p>Of course, I understand that people have different financial situations. I’m on an average full-time salary, so was fortunate enough to have <em>some</em> money left over to spend on podcast gear and services before we got any financial support. But when after we started a Patreon account, we were able to get a new logo designed, move our podcast hosting to Fireside, and upgrade our equipment.</p><p><strong><em>Patreon:</em></strong><em> </em>This is a great way to help ensure recurring financial support, but make sure you offer a valuable bonus to your patrons. An easy bonus to make is an extra episode. For our Everything Hertz <a href="https://www.patreon.com/hertzpodcast">$5 per month patrons</a>, we release a 15–20 minute bonus episode every month. This is on top of our regular 2 episodes per month.</p><p><strong><em>Advertising:</em></strong> I was hesitant to talk about advertising, because the chances of getting someone on board to sponsor your new podcast is very slim-how will sponsors know people will be listening? However, I decided to address this anyway because this is a question I get a lot.</p><p>If you’re going to include advertising, make sure it’s a good match for your audience and that you actually believe in the product. It’s easy to tell when podcasts hosts are just going through the motions with the products and services they’re advertising. You’ll want <em>at least</em> 1000 listeners per episode before you can even begin attracting advertisers, in general.</p><p>Don’t feel that you need ads just because the podcasts you listen to have ads. I recommend holding off for as long as possible. If you’re starting to gain attention in your niche, companies who also reside in your niche may contact you. Otherwise, you can reach out to potential advertisers and offer an advertising spot. Again, don’t rush into this and think long term.</p><p><strong><em>Merchandise:</em></strong> It’s <em>really</em> easy to set up an online store on a platform like <a href="https://teespring.com/">Teespring</a>, which sells t-shirts, hoodies, stickers, sweaters, and much more. Each product has a minimum price, so the profit that you get, over and above the minimum price, is up to you.</p><p>On the <a href="https://teespring.com/stores/everything-hertz-podcast">Everything Hertz online store</a>, we make a a very modest profit, as we’d rather our listeners have merchandise. Specifically, each product has a 20% profit margin, so that we can offer discounts to our Patreon supporters, who get our merchandise at (basically) cost price, and so that we can have seasonal promotions (e.g., 20% off sale for Christmas).</p><p>All you need to start your own merch store is a high quality image of your podcast logo and you’re good to go. The great thing about Teespring is that there’s no upfront cost. If no one buys merch, it won’t cost you anything. At the very least you can buy merch for yourself.</p><h3>What you should spend your money on</h3><p>Here’s my recommended order of what you should spend money on, whether this is from your own budget, from listener support, or from advertising.</p><ol><li>Starter USB microphone with headphone output ($4o-$100)</li><li>Good headphones if you don’t already own some ($80)</li><li>Host your podcast on Fireside ($19 per month) + domain name ($10-$20 per year)</li><li>iZotope RX elements ($129)</li><li>Hindenburg ($100)</li><li>Podcast logo ($300 on 99designs.com)</li><li>Record your episodes using eCamm live ($12 per month)</li><li><a href="https://www.amazon.com/dp/B0002BACBO?tag=marcoorg-20">Upgrade</a> your microphone ($250, plus a $100+ <a href="https://www.amazon.com/dp/B003VZG550/?tag=marcoorg-20">XLR interface</a>)</li></ol><p>Remember: The only thing you need to get started is a microphone! Don’t worry if you don’t have the budget for these other items, this is just a wishlist. Feel free to use a different order according to your needs.</p><p>Don’t badger your audience to give your podcast five starts on Apple podcasts. There’s no clear evidence, that I know of, that this actually increases your audience. This is one of things that people do because they hear other people doing it. A more effective way for people to discover your podcast is for you to share it on social media and to encourage your listeners to share it to their own networks, either via word-of-mouth or social media.</p><p>They key with getting someone to listen to a podcast that’s mentioned on social media is to reduce friction as much as possible. Here’s how you can share your podcast on Twitter and Facebook, ordered from the approach with the <em>most</em> friction to the approach with the <em>least</em> friction.</p><ol><li>Announcing you have a new episode</li><li>Announcing you have new episode with a link to the episode (many podcast hosts provide a link that will create a mini-player in Twitter, also known as an audio card)</li><li>Announcing you have a new episode using an audiogram or video, along with a link to the episode.</li></ol><p>Here is an example of a preview video, recorded using eCamm Live:</p><h3>Everything Hertz on Twitter</h3><p>102: Master of none https://t.co/pdy5IkL5ru Should research scientists build their knowledge and skillset broadly at the risk of being a master of none? In this episode, we discuss this thorny question, along with a recent editorial on the use of Twitter in academia https://t.co/ZhhEamp3WA</p><p>And here’s an example of an audiogram, made with <a href="https://make.headliner.app/">Headliner</a>, which is useful for when you don’t have video:</p><h3>Dan Quintana on Twitter</h3><p>I often have poor self-control when it comes to avoiding distractions from my writing and coding, so I put tools in place *ahead of time* to help me get stuff done. In the latest @OOSP_ORIONPod, I share how I do this practically https://t.co/YkxiCszslo https://t.co/2EUsB1dwGb</p><p>Notice that transcription was added to these videos. Many people can’t use audio when they’re browsing their feed, so adding a transcription <em>really</em> helps.</p><p>To add transcription to your videos, just upload them to YouTube, which performs an automatic transcription (even on non-public videos). Your results will vary, but it’s much faster than doing this manually. This happens in the background, and usually takes about four times the length of your episode to create these automatic captions. Like YouTube, Headliner’s automatic transcription is fairly accurate but you will need to do some edits.</p><h3>Common concerns</h3><p>Here are a few concerns I often hear when I’ve spoken to people about podcasts, and my responses:</p><p><strong>I hate hearing recordings of my voice: </strong>We all do, you’ll just need to get over this. It gets <em>slightly</em> easier over time.</p><p><strong>My topic area is too niche:</strong> If people are talking about your topic online, or there’s a <a href="https://www.reddit.com/subreddits/">subreddit</a>, then it’s not too niche. There’s an audience for pretty much everything out there.</p><p><strong>I won’t have any listeners:</strong> Every podcast started with no listeners. Building an audience takes consistency and patience.</p><p><strong>I’ll have nothing to talk about: </strong>Follow discussion of your topic on social media and you’ll never run out of episode ideas.</p><p><strong>I don’t have a budget: </strong>All you need is a mic. The tools to make a great sounding podcast are free.</p><h3>Some final points that weren’t covered in the sections above</h3><ul><li>It’s hard (but not impossible) for a single person to carry a podcast from episode-to-episode. Either get a co-host or regularly invite guests. Most people love talking about their work and ideas, so you should’t have trouble finding guests to join you.</li><li>Record your first five episodes ahead of time and let your listeners know that they can expect more to come as they’re already recorded.</li><li>To build an audience, you need to be consistent with your episode releases-even to the hour. Some people like knowing that they’ll have an episode ready for their commute, for example. Your release schedule is up to you, but I think the longest time between episodes should be a month if you’re doing a regular podcast. Release your episodes using a day of the week schedule (‘every Thursday’) or a day of the month schedule (‘first and third Thursday of the month’).</li></ul><p><em>Did I miss something in this podcasting guide that you wanted covered? Just </em><a href="https://www.dsquintana.com/#contact"><em>contact me</em></a><em>, and I’ll do my best to help you out. You can follow me on Twitter and Facebook, and subscribe to my regular bi-monthly podcast, ‘Everything Hertz’ (methodology in the biobehavioral sciences) and my semi-regular podcast, ‘Physiology and Behavior’ (psychoneuroendocrinology and psychophysiology). I also have an old podcast series, ‘The Startup Scientist’ (building your scientific career).</em></p><p><em>Originally published at </em><a href="https://www.dsquintana.blog/podcast-guide/"><em>https://www.dsquintana.blog</em></a><em> on March 2, 2020.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8a7d0d6b79a5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Oxytocin isn’t the “love hormone”, but the reality is far more intriguing]]></title>
            <link>https://medium.com/swlh/oxytocin-isnt-the-love-hormone-but-the-reality-is-far-more-intriguing-c79029b25262?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/c79029b25262</guid>
            <category><![CDATA[science-communication]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[relationships]]></category>
            <category><![CDATA[valentines-day]]></category>
            <category><![CDATA[love]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Fri, 14 Feb 2020 14:09:22 GMT</pubDate>
            <atom:updated>2020-02-18T14:38:36.395Z</atom:updated>
            <content:encoded><![CDATA[<p>It’s Valentines Day, which means there’s a lot of talk about oxytocin-the so-called “love hormone”. But despite the appeal of this label it’s actually a little misleading. So let me walk you through this hormone’s history and why the <em>real</em> role of oxytocin is much more interesting.</p><p>Oxytocin-like hormones are about 600 million years old, which is a long time before there was any evidence of “love”, or couple bonding. The sea sponge, considered one of the most ancient animal species, has no oxytocin-like signalling, suggesting it emerged in more modern species.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/622/1*WI97tP7NXxFRFlYA25cQgw.jpeg" /></figure><p>The sea squirt, which is a little less ancient as the sea sponge, uses oxytocin-like signalling, but what is its purpose if there’s no sea squirt bonding? Well, it turns out that oxytocin-like signalling in the sea squirt <a href="https://ncbi.nlm.nih.gov/pubmed/18583415">regulates tissue contraction</a> for salt regulation. Oxytocin-like signalling’s effect on tissue contraction is also seen in other species, such as <a href="https://ncbi.nlm.nih.gov/pubmed/8900077">earthworms</a> and <a href="https://ncbi.nlm.nih.gov/pubmed/10451417">leeches</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/471/1*XHv5RdJoGhaq19xWtatGRA.jpeg" /></figure><p>Another role of oxytocin that’s been observed in primitive animals is learning. Roundworms without oxytocin signalling <a href="https://t.co/nm9oKQc2tc?amp=1">fail to demonstrate</a> typical food-related learning behaviours.</p><p>Fast-forward a couple of hundred million years and the effects of oxytocin-like signalling on tissue contraction remains in humans. It’s involved in birth, breastfeeding, sex, digestion, and cardiovascular regulation, among other functions.</p><p>But let’s return to this “love hormone” idea in the context of tissue contraction. It plays a role in childbirth via uterine tissue contraction but mammals <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC38121/">can still give birth to healthy young</a> without oxytocin signalling. While oxytocin is involved, but not necessary, for childbirth, it’s crucial to nurse offspring. Mothers without oxytocin signalling produce milk, but they <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC38121/">can’t eject it</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*srpfMtYkXth6xXG7_Azvvw.jpeg" /></figure><p>Oxytocin’s role in birth and nursing has been known for about a century, so where did this “hug hormone” idea originate? Probably <a href="https://t.co/rBIuti5TRd?amp=1">this 1988 rat study</a> that reported increased blood concentrations of oxytocin after “gentle stroking”. Do a twitter search on “oxytocin” and you’ll see lots of tweets like, “Hugging your partner for [insert random number] seconds increases oxytocin”.</p><p>This idea started with a rat study (n=8) that found the same results with PINCHING.</p><p>Another origin of this “love hormone” ideas comes from animal research in prairie voles, who are socially monogamous. Oxytocin administration directly into the brain accelerates pair bonding while blocking oxytocin receptors restrict bonding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VktSNpE8_ovK7OLopCd1xg.jpeg" /></figure><p>But what about humans? Well, in 2005 a study was <a href="https://www.nature.com/articles/nature03701">published in Nature</a> that concluded intranasal oxytocin increases trust in humans. We’ll return to this idea soon, but this study was the catalyst for several other studies that reported similar results that oxytocin increases prosocial behaviours and the ability to understand what others are thinking or feeling, otherwise known as “theory of mind”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GXQxsANiZh1tf1D1TwTYXQ.jpeg" /></figure><p>All these findings contributed to a ton of hype, with <a href="https://io9.gizmodo.com/10-reasons-why-oxytocin-is-the-most-amazing-molecule-in-5925206">hyperbolic news stories</a> and even more hyperbolic claims (based on VERY preliminary work) of what oxytocin can do. I doubt what’s sold on Amazon contains <em>any</em> oxytocin, so don’t buy this, obviously.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fvvosKgcrMb7PLva1V7M_Q.jpeg" /></figure><p>Then came the studies that revealed the dark side of oxytocin, like that it <a href="https://www.ncbi.nlm.nih.gov/pubmed/19640508/">increases schadenfreude</a> and <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3029708/?wptouch_preview_theme=enabled">reduces cooperation</a> with outgroup members. Not what you would call “positive” behaviors.</p><p>And remember that ‘trust’ study from 2005? This finding <a href="https://www.ncbi.nlm.nih.gov/pubmed/26581735/">hasn’t replicated</a>, with 7 studies cumulatively suggesting that oxytocin doesn’t increase trust. Other studies have also failed to replicate oxytocin’s effects in different social domains.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CcZD-OHT2YMNU7T39NpV3g.jpeg" /></figure><p>So what does the oxytocin system actually do? The best that we can say right now is that it can influence our behavior and mental states (both positive and negative), but only under some circumstances. Based on this research and work by other teams, it seems that the oxytocin system is involved in several physiological <em>and</em> psychological functions, which happens to include social behavior.</p><p>But how can one hormone do so many things, often simultaneously? Regarding physiology: Oxytocin receptors are sensitive to oxytocin’s release schedule. For instance, milk let-down occurs in response to strong pulses whereas salt levels respond to steady low-level release.</p><p>When it comes to behavior, oxytocin is released two ways in the brain: direct transport from oxytocin-producing cells to specific regions and broad diffusion throughout the brain for action on its receptors.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2RP1b4-0jRSgORgq3FMLDg.jpeg" /></figure><p>And the location of these receptors are associated with mental states (as measured by a meta-analysis of thousands of fMRI studies) can be broadly categorized as anticipatory, appetitive, and aversive. In fact, there seems to be more evidence oxytocin system supports metabolism than social behavior.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bFaXFH1ENl74jMC-.jpeg" /></figure><p>Rather than evolving new sets of signalling systems, evolution has figured out a way to repurpose one system for several purposes via the location of oxytocin receptors in the brain and by making peripheral receptors sensitive to specific release schedules.</p><p>If biological signalling systems were like radio transmission waves, this would be like adopting morse code and adjusting the locations of receiver antennas sensitive to this frequency instead of creating a new transmission frequency.</p><p>This post was adapted from a <a href="https://twitter.com/dsquintana/status/1095927629050531840?s=20">Twitter thread</a>. I have also written a new manuscript which discusses lots of the ideas shared here. The paper is currently under review, but you can read a <a href="https://osf.io/j7tnf/">preprint version</a>.</p><p><em>Originally published at </em><a href="https://www.dsquintana.blog/oxytocin-isnt-the-love-hormone-but-the-reality-is-much-more-fascinating/"><em>https://www.dsquintana.blog</em></a><em> on February 14, 2020.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c79029b25262" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/oxytocin-isnt-the-love-hormone-but-the-reality-is-far-more-intriguing-c79029b25262">Oxytocin isn’t the “love hormone”, but the reality is far more intriguing</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The rabbit/raven illusion, explained]]></title>
            <link>https://medium.com/@dsquintana/the-rabbit-raven-illusion-explained-aa3815de8389?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/aa3815de8389</guid>
            <category><![CDATA[twitter]]></category>
            <category><![CDATA[ravens]]></category>
            <category><![CDATA[rabbits]]></category>
            <category><![CDATA[illusion]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Thu, 22 Aug 2019 08:16:33 GMT</pubDate>
            <atom:updated>2019-08-22T19:54:41.511Z</atom:updated>
            <content:encoded><![CDATA[<p>Last Sunday, I came across a video posted on the Imgur twitter account.</p><h3>Imgur on Twitter</h3><p>I think there&#39;s something wrong with your rabbit. https://t.co/Llr6QQ6Nd1 https://t.co/V4LHapsZD3</p><p>As a psychological scientist, I knew that this video and comment alluded to the famous bird/rabbit illusion, first popularised by psychologist <a href="https://www.independent.co.uk/news/science/duck-or-rabbit-the-100-year-old-optical-illusion-that-tells-you-how-creative-you-are-a6873106.html">Joseph Jastrow</a>, but <a href="https://en.wikipedia.org/wiki/Rabbit%E2%80%93duck_illusion">originally published</a> in 1982.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bidBVtokAKUEOIA8Jcfbkg.png" /><figcaption>Source: <a href="https://en.wikipedia.org/wiki/Rabbit%E2%80%93duck_illusion">Wikipedia</a> (Public domain)</figcaption></figure><p>Even in light of the comment attached to Imgur’s tweet, I still thought it was fairly obvious that the animal was a bird. So using Twitter’s “share video” function, I reposted the video (which attributes Imgur as the original poster underneath the video on Twitter’s website and Twitter smartphone apps) and added my comment, to spice up the illusion.</p><h3>Dan Quintana on Twitter</h3><p>Rabbits love getting stroked on their nose https://t.co/aYOZGAY6kP</p><p>By first directing the viewer’s attention to the nose, I was trying to distract viewers from the ears/beak, one of the clear giveaways that this was a video of a raven-a <a href="https://en.wikipedia.org/wiki/White-necked_raven">white-necked raven</a> to be precise.</p><p>But to me, the most obvious giveaway was the eye. Here’s a snippet of <a href="https://www.cnet.com/news/bird-or-bunny-mystifying-animal-video-has-the-internet-squawking/">my response to CNET</a>, who reached out to me over a Twitter direct message.</p><blockquote><em>“I thought it was fairly clear that the video was of a bird …as you can see the translucent </em><a href="https://en.wikipedia.org/wiki/Nictitating_membrane"><em>nictitating membrane</em></a><em> sweep across the eye horizontally (rabbits don’t have membranes like this)”</em></blockquote><p>I thought a few of my followers, many of which are also psychological scientists, would have had a chuckle at this famous static image coming to life.</p><p>Turns out I was wrong.</p><p>As I’m writing this, the video in the tweet has been viewed almost 8 million times. To my surprise, the tweet has been featured on <a href="https://news.google.com/stories/CAAqSQgKIkNDQklTTERvSmMzUnZjbmt0TXpZd1NoOGFIV1JCWDBaSFJscE1MVkpsWlZObVRVVnNNRmcxZDB4MGIxODJhMHROS0FBUAE?q=rabbit+bird&amp;lr=English&amp;hl=en-US&amp;gl=US&amp;ceid=US:en">several websites</a> and his was even shared on <a href="https://twitter.com/Chris_1791/status/1164142521578115074?s=20">TV news</a> segments.</p><p>As a scientist, I understand the importance of acknowledging people’s work. I shared the Imgur video in good faith that the <a href="https://imgur.com/gallery/x38lMZe">original poster</a> owned the video. Only later did I discover that this video was originally posted over two years ago by <a href="https://www.instagram.com/pythonpaige/?hl=en">Paige Davis</a>, who is the curator of bird training at the <a href="https://www.worldbirdsanctuary.org/">World Bird Sanctuary</a>. When I discovered this, I added <a href="https://twitter.com/dsquintana/status/1164418984382554112?s=20">a clarification</a> to the original tweet thread, acknowledging the original source.</p><p>To be honest, the “success” of this tweet is a little bittersweet. Compared to the 30-seconds it took me to write this tweet, I spend <em>years</em> working on <a href="https://www.dsquintana.com/">my research </a>on the hormone oxytocin, only for these outcomes to be shared about a hundred times on Twitter, <a href="https://twitter.com/dsquintana/status/1093814606441648128?s=20">at best</a>.</p><p>Well, at least <em>one</em> person discovered my research as a result of the tweet, so maybe this was 30-seconds well spent.</p><h3>Eff Luls on Twitter</h3><p>@dsquintana I found you through the &quot;rabbit&quot;, but rest assured I&#39;m following for your content. This research is real fuckin&#39; interesting</p><p><em>Originally published at </em><a href="https://www.dsquintana.blog/the-raven-rabbit-illusion-explanation/"><em>https://www.dsquintana.blog</em></a><em> on August 22, 2019.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aa3815de8389" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to respond to reviewer’s comments on your manuscript]]></title>
            <link>https://medium.com/@dsquintana/how-to-respond-to-reviewers-comments-on-your-manuscript-11befc1ad27f?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/11befc1ad27f</guid>
            <category><![CDATA[science-communication]]></category>
            <category><![CDATA[academia]]></category>
            <category><![CDATA[phd]]></category>
            <category><![CDATA[peer-review]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Tue, 23 Apr 2019 12:39:28 GMT</pubDate>
            <atom:updated>2019-04-23T12:39:28.066Z</atom:updated>
            <content:encoded><![CDATA[<p>You might be frustrating the people who peer-review you manuscripts without even realising it. Here are some brief tips on reducing peer review friction from my (limited) experience as an author, reviewer, and associate editor.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wKqXbOLqISoFSnYNkrwiOA.jpeg" /></figure><p>Copy and paste each reviewer comment (number them yourself if you have to) and respond to each one in turn. This seems like common sense, but I’ve come across many responses to my peer-review queries in which I have to spend time searching and figuring out which comment the authors are actually responding to.</p><p>Don’t just write something like, “We updated the introduction”. Paste in the chunk of text you updated into the response document, also noting its location in the manuscript. Some journal systems add line numbers to PDFs. If they don’t, add your own so you can refer to them in your response.</p><p>Err on adding more info rather than less. When reviewing a paper, I’ve never thought, “That author provided too much info in their response”.</p><p>It’s fine to disagree with a reviewer, but actually explain why, rather than deferring to your authority in the field. I’ve gotten a response like, “We disagree because we’ve been doing this for 20 years” a few times recently. This is not a classy move.</p><p>Unless the journal explicitly tells you not to, paste figures and tables in relevant sections of the manuscript. Don’t make the reviewer work harder than they need to by making them move back and forth to the end of the document. I’ve heard of a few people that just do this anyway, even if the journal advises against this. The worst that can happen if you take this approach is that the editorial office returns your submission for correction.</p><p>If you can choose your citation style upon first submission (lucky you!) choose an “author, year” in-text citation system (unless this is very uncommon in your field) so reviewers don’t have to flip to the back to check which paper you cited.</p><p>Don’t ignore hard-to-address comments thinking that reviewers will miss these. I’ve seen authors try this (it’s always the hard comments they “miss”). Reviewers notice these things.</p><p>All these things come down to giving your reviewers LESS work to do. We like to think that peer review is objective, but it’s not because a human is doing it. All things equal, you’re more likely to get a favourable evaluation from a happy reviewer than an annoyed reviewer.</p><p><em>This blog post was </em><a href="https://www.dsquintana.com/post/23_apr_2019_peer-review/"><em>originally posted on my website</em></a><em>. If you found this post useful, you might also like </em><a href="https://shows.pippa.io/dsquintana/episodes/twitter-manuscript-revisions-and-worklife-balance-a-chat-wit"><em>my podcast episode with Stu Murray</em></a><em> on dealing with manuscript revisions, about getting the most out of twitter, and work-life balance.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=11befc1ad27f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to avoid three common mistakes in meta-analysis]]></title>
            <link>https://medium.com/data-science/how-to-avoid-three-common-mistakes-in-meta-analysis-eb64427e8569?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/eb64427e8569</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[rstats]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Wed, 19 Sep 2018 18:35:33 GMT</pubDate>
            <atom:updated>2018-09-19T20:55:33.447Z</atom:updated>
            <content:encoded><![CDATA[<h3>Three common mistakes in meta-analysis</h3><p>Meta-analysis has become a popular tool to synthesise data from a body of work investigating a common research question. As this provides a putatively objective means of summarising a body of data, it has become the <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1120936/">gold standard for evidence</a> when developing health guidelines.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3b4d2MQCJ94CtzDaKI3VvA.jpeg" /></figure><p>However, there are several issues in meta-analysis that can contribute to inaccuracies. Due to these limitations, meta-analysis — or “mega-silliness”, to borrow a term from <a href="http://psycnet.apa.org/doiLanding?doi=10.1037%2F0003-066X.33.5.517.a">an early detractor</a> — has been criticised for almost as long as the method has been available.</p><p>There are several tools to help ensure meta-analysis quality, which will be outlined later in this post. But first, let’s consider this consider this (fictitious) extract from the results section of a meta-analysis that I will use to briefly illustrate three common issues in meta-analysis:</p><blockquote>“As hypothesised, a meta-analysis synthesising eight effect sizes from seven studies suggests that social media usage is positively associated with self-reported levels of anxiety [r = .073, 95% CI ( .004, 0.142); p = .038; Fig. 1A]. Egger’s regression test suggested no evidence of publication bias (p = .19; Fig. 1B).”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Bo2qtILa02eQwInT3V9TsQ.jpeg" /><figcaption><strong>Figure 1. The relationship between social media usage and self-reported anxiety. </strong>The forest plot (a) visualises point estimates for a set of hypothetical studies with filled squares. The diamond reflects the estimated summary effect size. The funnel plot (b) illustrates each effect size against its standard error</figcaption></figure><p>When this imaginary author — who also happens to be releasing a book on the dangers of social media — initially searched for studies, they actually found nine studies fulfilling their inclusion criteria. However, the summary effect size of this meta-analysis of nine studies [<em>r</em> = .062, 95% CI (- .004, 0.127)] was associated with a<em> p</em>-value of .061 (Figure 2).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aTaqu_9Wmhqft63YJWNecw.jpeg" /><figcaption><strong>Figure 2</strong>. <strong>The original set of studies for the assessment of the relationship between social media usage and self-reported anxiety. </strong>Upon realising that the removal of study six would result in a statistically significant result, our fictitious researcher removed this study using a post hoc justification. Meta-analysis pre-registration would have prevented a post hoc adjustment of exclusion criteria dressed up as being <em>a priori</em></figcaption></figure><p>As this outcome would hinder the chances of publication (and book sales), they took a closer look at study six, which reported a negative association between social media usage and anxiety levels. This study happened to be the earliest study included in the meta-analysis.</p><p>Study six did not include Instagram usage (which became popular during 2012), so they adjusted the eligibility criteria to only include studies published from 2013 onwards, justified by a (post hoc) hypothesis that Instagram usage is most associated with anxiety.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GtEwHQfjNsjMnG3Xran1Vw.jpeg" /><figcaption>Is Instagram usage associated with anxiety? Probably not, but our fictitious author used this idea as a post hoc justification for removing a study from their meta-analysis</figcaption></figure><p>With the removal of the study, the meta-analysis became statistically significant. While this fabricated scenario might seem extreme at first glance, <a href="https://www.ncbi.nlm.nih.gov/pubmed/22508865">almost 40% of over 2,000 surveyed psychologists admitted</a> that they had excluded data after looking at the impact of doing so.</p><h4>Pre-registering your analysis plans</h4><p>Meta-analysis pre-registration provides a remedy to analytic flexibility as it provides an <em>a priori</em> record of the researcher’s analysis intentions. In the above example, exclusion criteria would have been specified before the analysis was conducted.</p><p>A common critique of pre-registered analyses is that they offer no flexibility for analysis. However, this is incorrect as pre-registration does not lock the researcher into a specific plan but instead helps clearly delineate pre-planned and exploratory analysis.</p><p>In the above case, the researcher would be free to report their analysis with the removal of the study, but they would have to state their reasoning, which is flimsy at best. For guidance, the Preferred Reporting Items for Systematic Reviews and Meta-Analysis protocols (<a href="http://www.prisma-statement.org/Extensions/Protocols.aspx">PRISMA-P</a>) document provides a list of items that should be addressed in a meta-analysis protocol.</p><p>There are several options available for meta-analysis pre-registration. First, a time-stamped meta-analysis protocol following PRISMA-P guidelines can be uploaded to <a href="http://osf.io/">Open Science Framework</a>.</p><p>Second, if the meta-analysis investigates a health-related outcome, it can be pre-registered at <a href="https://www.crd.york.ac.uk/prospero/">PROSPERO</a>. The PROSPERO website guides the researcher through a step-by-step registration form, which is used to generate a pre-registration record. These records can be amending, but any changes are timestamped.</p><p>Third, a meta-analysis protocol can be submitted for peer-review at a number of journals, such as <a href="https://systematicreviewsjournal.biomedcentral.com/">Systematic Reviews</a>. While the peer-review process obviously takes more time than uploading a protocol, it can provide important feedback that can improve the meta-analysis.</p><p>Fourth, a meta-analysis can be submitted using the <a href="https://cos.io/rr/">Registered Report</a> format, in which the meta-analysis is reviewed in two stages. In the first stage, the meta-analysis introduction and methods sections are peer-reviewed before data collection. If the study rationale and methodological approach is deemed appropriate, the journal will provide provisional acceptance of the final paper.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*SYRdhMi0-XLz_GP86d95xg.png" /><figcaption>An overview of the Registered Reports format (Source: <a href="https://cos.io/rr/">Center for Open Science</a>)</figcaption></figure><p>Once the data is collected and analysed, the paper is submitted again for the second stage review which includes with the same introduction and methods as stage one, but with the addition of results and conclusions. In the second stage of review, the focus is on whether the results and interpretation are in line with the pre-approved method.</p><p>Importantly, paper acceptance is not contingent on the statistical significance of results. As of August 2018, there are <a href="https://cos.io/rr/">42 journals</a> that offer Registered Report meta-analysis, with the majority of these journals publishing research from the biobehavioral sciences.</p><h4>Publication bias</h4><p>Publication bias is a <a href="https://www.ncbi.nlm.nih.gov/pubmed/21784880">well-recognised issue</a> in meta-analysis. As statistically significant studies are more likely to be published, it is likely that there are several non-significant studies left languishing in researcher’s file-drawers for any given research area, which cannot contribute to meta-analyses.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YVRj1KL-SQW5SZELqHn0xQ.jpeg" /><figcaption>How much data is left to be forgotten in file drawers or lab archives?</figcaption></figure><p>Thus, a meta-analysis summary effect size may not best represent the “true” effect size. A common approach to assess the risk of publication is to construct a funnel plot, which plots effect sizes against a measure of variance (e.g., standard errors). A symmetrical funnel plot is indicative of low risk of publication bias. Despite the inherent subjectivity of visually inspecting a funnel plot to assess evidence of asymmetry, many researchers still rely on this tool.</p><p>Now let’s return to our imaginary study. As our fictitious researcher was aware of issues with subjective funnel plot assessments, they objectively assessed asymmetry by constructing a funnel plot. As <a href="https://www.ncbi.nlm.nih.gov/pubmed/16085192">researchers are poor at identifying asymmetry</a> from funnel plots, they followed recommendations and performed Egger’s regression test to complement the funnel plot (Fig. 1A).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FuUGVTsLFZ-eKARHbOb66Q.jpeg" /><figcaption>It can be hard to subjectively decide whether a funnel plot looks asymmetrical, especially if there’s a conflict of interest</figcaption></figure><p>A statistically significant test is indicative of funnel plot asymmetry. In this case, the test was not statistically significant (<em>p</em> = .19). Thus, our researcher concluded that there was no risk of publication bias. A closer look at the data, however, reveals that this conclusion was most likely incorrect.</p><p>A common misunderstanding of funnel plots and their associated tests is that they measure publication bias, when in fact they measure <a href="https://www.ncbi.nlm.nih.gov/pubmed/21784880"><em>small study</em> bias</a>. This type of bias might include publication bias, but it also encompasses other types of bias commonly associated with small studies, such as study quality.</p><p>A relatively simple funnel plot modification, called a contour-enhanced funnel plot, can help better understand the risk of publication bias compared to small study bias. For a given range of effect sizes and variances, it is possible to calculate levels of statistical significance at every point. As a result, it is possible to visualise ranges of statistical significance (e.g., <em>p</em> = .05 to .01) on a funnel plot.</p><p>Comparing our original funnel plot (Fig. 3A) with a contour-enhanced funnel plot (Fig. 3B) reveals that despite no apparent evidence of funnel plot asymmetry, half of the included effect sizes had <em>p</em>-values were very close to .05. It is highly improbable that half of the available <em>p</em>-values would lie so close to .05. Either these significant studies were also the result of selective analyses or there are several unpublished studies that exist.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H37ktfhqjEkMQJiHfeHCPg.jpeg" /><figcaption><strong>Figure 3. A comparison of conventional meta-analysis funnel plots vs. contour-enhanced funnel plots. </strong>Compared to the original funnel plot (a), which is also presented in Fig 1B, a contour enhanced funnel plot (b) visualises ranges of statistical significance. The gray-shaded region corresponds to p-values between .10 and .05, whereas the dark gray-shaded region corresponds to p-values between .05 and .01</figcaption></figure><p>Several measures have developed to correct for bias in meta-analysis, such as <a href="https://www.ncbi.nlm.nih.gov/pubmed/26186117">p-curve</a>, <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5117126/">p-uniform</a>, and the <a href="https://www.ncbi.nlm.nih.gov/pubmed/27694467">three-parameter selection model</a>, with the latter demonstrating <a href="https://osf.io/jn6x5/">superior performance</a>. But without access unpublished research these methods will always be estimates, at best. One plausible solution to this scenario is to conduct a meta-analysis that only includes pre-registered empirical studies, such as <a href="https://www.tandfonline.com/doi/abs/10.1080/23743603.2017.1326760">the recent meta-analysis</a> on “power posing” from Gronau and colleagues.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VYV-hRPuhNPrJ59BwDHZqQ.jpeg" /><figcaption>The idea that adopting an ‘expansive’ body posture makes people feel more powerful makes for a nice story (and TED talk) but the evidence supporting this effect is shaky</figcaption></figure><p>Now to our final common issue in meta-analysis: effect size dependence. Notice that in the earlier meta-analysis example there were eight effect sizes but only seven studies? This was due to two effect sizes being extracted from study four (Fig. 1A). This is a common occurrence as papers sometimes report several outcome measures.</p><p>However, as these effect sizes were derived from the same study population they are statistically dependent. As typical meta-analysis procedures assume that effect sizes are statistically independent, combining dependent meta-analyses can lead to <a href="https://www.amazon.com/Methods-Meta-Analysis-Correcting-Research-Findings/dp/141290479X">biased results</a>.</p><p>There are several approaches to for addressing statistical dependency in meta-analysis. First, you can simply choose one effect size per study. While straightforward, a pre-specified system is required to help select the chosen effect sizes, otherwise it can be tempting to simply choose the largest effect. What’s more, this approach excludes potentially valuable effect sizes from your analysis. To include more than one effect from a study, it is possible to statistically aggregate effects. But to do this, you need to include a measure of within-study correlation, which is almost never reported in papers.</p><p>If the paper contains raw data, an occurrence which is almost as rare, within-study correlations can be calculated. Therefore, researchers almost always need to estimate within-study correlations if this approach is to be taken. Alternatively, <a href="https://www.ncbi.nlm.nih.gov/pubmed/26056092">robust variance estimation</a> can be used to account for effect size dependency without a need to know within-study correlations.</p><p>Meta-analysis is an important tool for better understanding often disparate research areas. This method can provide a more objective summary of a research area than simply “vote counting” the number of significant and non-significant results.</p><p>But like almost any statistical tool, meta-analysis can facilitate inaccurate inferences if used inappropriately. With this post, I hope to have provided a brief overview highlighting some common issues in meta-analysis and practical solutions to these problems.</p><p>This is an edited version of an article to be published in “<a href="http://psykologisktidsskrift.no/">Psykologisk Tidsskrift NTNU</a>”, a psychology magazine run by students at the Norwegian University of Science and Technology. <a href="https://osf.io/5skuv/">Here</a> is a citable preprint of the article and the R script and data to perform the analyses described in this post are available <a href="https://osf.io/3vjm5/">here</a>. My thanks to the Editor of Psykologisk Tidsskrift for kindly providing permission to repost this article as a blog post.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eb64427e8569" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/how-to-avoid-three-common-mistakes-in-meta-analysis-eb64427e8569">How to avoid three common mistakes in meta-analysis</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to determine whether a non-significant result supports the absence of an effect]]></title>
            <link>https://medium.com/@dsquintana/using-summary-statistics-to-determine-whether-a-non-significant-result-supports-the-absence-of-an-1ff61e97f7cf?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ff61e97f7cf</guid>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[bayesian-statistics]]></category>
            <category><![CDATA[research-methods]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Sat, 21 Jul 2018 06:08:55 GMT</pubDate>
            <atom:updated>2018-07-25T01:05:30.360Z</atom:updated>
            <content:encoded><![CDATA[<h3>Using summary statistics to determine whether a non-significant result supports the absence of an effect</h3><p>I just came across <a href="https://www.sciencedirect.com/science/article/pii/S0941950017300520">a new study</a> that investigated the relationship between generalised anxiety symptoms and automonic nervous system function. This is an interesting research question given that generalised anxiety symptoms are often associated with autonomic arousal symptoms, like sweating and palpitations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*laybCkTn1J_Il-Rslb46pg.jpeg" /><figcaption>Generalised anxiety symptoms are often associated with autonomic arousal symptoms</figcaption></figure><p>In this study, 32 participants with social anxiety disorder and 23 neurotypical controls were recruited. Heart rate variability (HRV) was used to approximate autonomic control of the heart rate. The authors concluded that there was no relationship between generalised anxiety symptoms and HRV.</p><p>Now the thing with conventional null hypothesis test <em>p</em>-values is that you can only <em>reject</em> the null hypothesis. A non-significant test is not support of the null hypothesis, now matter how large the <em>p</em>-value.</p><p>I was curious as to whether the data actually supported the absence of an effect. In this post, I’m going to walk you through what I did to better understand this result.</p><p>Before we get into the analysis, let’s first look at the effect size that the study design had 80% statistical power to detect using the “pwr” R package. To calculate this, we only need to enter the group numbers, a specified significance level, and the desired power level.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8940813f29ec1ad31c793b503b13dddc/href">https://medium.com/media/8940813f29ec1ad31c793b503b13dddc/href</a></iframe><p>For an independent samples <em>t</em>-test, this study can effect size of <em>d</em> = 0.78 with 80% power. While this would conventionally just fall short of a large effect size (<em>d</em> = 0.8) according to Cohen’s guidelines, Cohen only intended for his effect size guidelines to be used as a last resort when the researcher did not have a good grasp of expected effect sizes from the published literature.</p><p>Fortunately, <a href="https://onlinelibrary.wiley.com/doi/abs/10.1111/psyp.12798">I already went to the trouble of calculating the distribution of effect sizes from HRV case-control studies</a>, because I’m a nerd like that.</p><p>I found that for anxiety studies, a large effect size should be classified as <em>d </em>= 0.64 (instead of <em>d</em> = 0.8, which is Cohen’s suggested threshold for large effects), as this was the 75th percentile of effect sizes from 55 studies. In other words, only 25% of anxiety case control HRV studies have effect sizes that are larger than <em>d </em>= 0.64.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2LYqhtrCDDJgs41g4FjM3w.png" /><figcaption>The distribution of effect sizes for HRV case-control studies categorised by different disorder groups.</figcaption></figure><p>So it looks like this study is only statistically powered to detect large effect sizes. But let’s plow ahead anyway.</p><p>For our first approach to better understand this non-significant result, we’re going to use <a href="http://journals.sagepub.com/doi/abs/10.1177/1948550617697177">equivalence testing</a>. Equivalence testing uses a frequentist framework to determine whether you can reject the presence of a smallest effect size of interest (SESOI).</p><p>The first thing we need to do is to specify our SESOI. There are <a href="https://psyarxiv.com/v3zkt">several ways to specify this</a>, but here I will specify our SESOI as <em>d</em> = 0.26. I’ve chosen this SESOI as this would <a href="https://onlinelibrary.wiley.com/doi/abs/10.1111/psyp.12798">reject the upper 75% of effect sizes</a> in the anxiety case-control HRV literature, which seems pretty reasonable to me.</p><p>We’ll have a look at the RMSSD measure of HRV, as this is one of the most commonly used measures in the literature. From the paper, we can extract group numbers, means, and standard deviations and then run a short script in R using the <a href="https://cran.r-project.org/web/packages/TOSTER/index.html">TOSTER</a> package.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/f544c45958040c8d2ac6e07b4e1f5b7c/href">https://medium.com/media/f544c45958040c8d2ac6e07b4e1f5b7c/href</a></iframe><p>As the equivalence test was non-significant (<em>p</em> = 0.35) we cannot reject effects that are as large or larger than <em>d</em> = 0.26. The plot below visualises this result. Notice that the thick line, which is the 90% two one-sided tests confidence interval, falls outside the dashed equivalence bounds of -0.26 and 0.26.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/483/1*ger0WDnWMZ8SUsW8Qlq8dA.png" /><figcaption>The plot of our equivalence test</figcaption></figure><p>My second approach was a Bayesian hypothesis test. With this test you can quantify how much more likely the data are under a null hypothesis model compared to an alternative model, given a prior probability (here’s a <a href="https://link.springer.com/article/10.1186/s12888-018-1761-4">primer</a> if you’re new to Bayesian hypothesis testing).</p><p>To perform a Bayesian hypothesis test from summary statistics, we just need to enter a <em>t</em>-statistic and group numbers. The authors didn’t report <em>t</em>-statistics in their results, but it’s possible to calculate this from the means, standard deviations, and group sizes.</p><p>The method section in this paper is very light on detail, so it’s unclear whether the <em>p</em>-value was generated from a Student’s <em>t</em>-test (assuming equal variances), Welch’s <em>t</em>-test (equal variances not assumed), or a Mann-Whitney test. RMSSD values are <em>usually</em> normally distributed, so I’ve gone with Welch’s <em>t</em>-test to be safe.</p><p>To calculate the <em>t</em>-statistic and corresponding <em>p</em>-value, I used the function below from <a href="https://stats.stackexchange.com/questions/30394/how-to-perform-two-sample-t-tests-in-r-by-inputting-sample-statistics-rather-tha?noredirect=1&amp;lq=1">this post</a>.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/b25b87a39834bf1b704d2b6f2b7d0ae0/href">https://medium.com/media/b25b87a39834bf1b704d2b6f2b7d0ae0/href</a></iframe><p>The <em>t</em>-statistic was -0.59 and the <em>p</em>-value was 0.56. This wasn’t exactly what was reported in the paper (<em>p</em> = 0.594), which could be due to a variety of reasons, but it’s still quite close so we’ll use it anyway for pedagogic purposes.</p><p>Before we run our Bayesian hypothesis test, we also need to specify our prior distribution. Here, I’ll use a default Cauchy prior with a scale of 0.44, which presumes we are 50% confident that the true effect size will lie between −.44 and .44. I chose this figure because 0.44 <a href="https://onlinelibrary.wiley.com/doi/abs/10.1111/psyp.12798">is the median effect size</a> for anxiety case-control HRV studies. Note that there are a <a href="https://link.springer.com/article/10.1186/s12888-018-1761-4">few other options</a> for specifying your prior for a Bayesian <em>t</em>-test in JASP, which might be better suited for your research question.</p><p>Now, we just plug our <em>t</em>-statistic, group numbers, and prior distribution into the <a href="https://jasp-stats.org/2018/04/11/teaching-bayesian-estimation-with-the-summary-stats-module/">Summary Stats module of JASP</a>. This gives us a BF_01 value of 2.26, which suggests that the null model was preferred to the alternative model by a factor of 2.26. Conventionally speaking, this is only considered anecdotal evidence for the null model relative to the alternative model but you can make up your own mind as to the degree of evidence.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1016/1*ep8obqHD0SfS7SJkbdOvoQ.png" /><figcaption>The prior and posterior plot for our analysis</figcaption></figure><p>If we use a more generous Cauchy width the relative evidence for the null model increases (see below). However, I would argue that these wide and ultrawide widths are <em>too</em> wide given what we know about the literature.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*exuvm_ZZ5IgV5Du33yzdyg.jpeg" /><figcaption>A robustness check can assess the impact of various Cauchy widths on the Bayes factor</figcaption></figure><p>By using both a frequentist and Bayesian approach, we can conclude that this non-significant result was probably not indicative of the absence of an effect. What’s more likely is that this non-significant result was due to the study being statistically underpowered.</p><p><em>If you liked this post, you’d probably enjoy our podcast episode with </em><a href="https://twitter.com/lakens/"><em>Daniel Lakens</em></a><em> on p-values. Follow </em><a href="https://soundcloud.com/everything-hertz/52-give-ps-a-chance-with-daniel-lakens"><em>this link</em></a><em> or listen using the player below. After 65 episodes and over 125,000 downloads, this is one of our most popular episodes.</em></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fw.soundcloud.com%2Fplayer%2F%3Furl%3Dhttp%253A%252F%252Fapi.soundcloud.com%252Ftracks%252F347738049%26show_artwork%3Dtrue&amp;url=https%3A%2F%2Fsoundcloud.com%2Feverything-hertz%2F52-give-ps-a-chance-with-daniel-lakens&amp;image=http%3A%2F%2Fi1.sndcdn.com%2Fartworks-000248018394-queqnf-t500x500.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=soundcloud" width="800" height="166" frameborder="0" scrolling="no"><a href="https://medium.com/media/4740948306cb24a1c8454f47ef96a080/href">https://medium.com/media/4740948306cb24a1c8454f47ef96a080/href</a></iframe><h4>More where this came from</h4><p>This story is published in <a href="http://blog.usejournal.com/">Noteworthy</a>, where thousands come every day to learn about the people &amp; ideas shaping the products we love.</p><p>Follow our publication to see more product &amp; design stories featured by the <a href="https://usejournal.com/?utm_source=usejournal.com&amp;utm_medium=blog&amp;utm_campaign=guest_post">Journal</a> team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ff61e97f7cf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Kisspeptin: the serendipitously named reproduction hormone]]></title>
            <link>https://medium.com/@dsquintana/kisspeptin-the-serendipitously-named-reproduction-hormone-35c89ac14602?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/35c89ac14602</guid>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[hormones]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[podcast]]></category>
            <category><![CDATA[psychology]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Thu, 19 Jul 2018 06:37:58 GMT</pubDate>
            <atom:updated>2018-07-20T10:05:28.630Z</atom:updated>
            <content:encoded><![CDATA[<p>I spoke with Matt Wall about a recent study he co-authored investigating the role of kisspeptin on emotional brain processing for my <a href="https://shows.pippa.io/dsquintana">Physiology and Behavior show</a>. Here’s the transcript of the episode.</p><p>To listen to this episode, use the player below or follow <a href="https://shows.pippa.io/dsquintana/kisspeptin-and-emotional-brain-processing">this link</a> to the show page.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.pippa.io%2F5b33637a7b9cf5a0177f18b5%2Fepisodes%2Fkisspeptin-and-emotional-brain-processing&amp;url=https%3A%2F%2Fplayer.pippa.io%2F5b33637a7b9cf5a0177f18b5%2Fepisodes%2Fkisspeptin-and-emotional-brain-processing%3Ftheme%3Ddefault%26cover%3D1%26latest%3D1&amp;image=https%3A%2F%2Fres.cloudinary.com%2Fpippa%2Fimage%2Ffetch%2Fc_pad%2Cw_600%2Ch_315%2Fhttp%253A%252F%252Fs3.amazonaws.com%252Fassets.pippa.io%252Fshows%252F5b33637a7b9cf5a0177f18b5%252F1531915490998-fafca19bfd0d7c6f179e0b0403be84f6.jpeg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=pippa" width="640" height="110" frameborder="0" scrolling="no"><a href="https://medium.com/media/c755ac8c8c5112972d6b4457eeb12cf7/href">https://medium.com/media/c755ac8c8c5112972d6b4457eeb12cf7/href</a></iframe><p><strong>Dan:</strong> A paper recently came out on how the <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5272173/">hormone kisspeptin modulates sexual and emotional brain processing</a>. I asked Matt Wall, one of the study co-authors, about the origins of kisspeptin and it’s unusual name.</p><p><strong>Matt</strong>: Kisspeptin is a peptide hormone, it was discovered in the late 90s, and it was first thought to play a role in cancer. It was discovered in a town called Hershey, Pennsylvania, where there’s also a massive factory where they make Hershey’s chocolates, in particular the little chocolates they call Hershey’s kisses, hence the name, kisspeptin.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zIDxcUDfLBmb4ZznHpBR3w.jpeg" /></figure><p><strong>Matt (continued)</strong>:It was only later it was found that it had a big role to play as a sex hormone, particularly in puberty, so it’s the hormone that starts off the whole cascade of other hormones that you get that kick off puberty. Rats that don’t have the kisspeptin gene don’t go through puberty and they stay sexually immature.</p><p>So the people that named it originally, even though they didn’t know it at the time, they gave it an incredibly appropriate name.</p><p><strong>Dan: </strong>In this fMRI study 29 healthy heterosexual men were administered either kisspeptin or placebo intravenously in separate visits to better understand kisspeptin’s effects on neural activity.</p><blockquote>“So what we think is that that kisspeptin serves as a link between the brain systems that are processing emotional, sexual, and love-related information and the rest of the bodily reproductive system.” — Matt Wall</blockquote><p><strong>Matt: </strong>What we did was we showed people a lot of different emotional pictures, some were sexual, romantic, positive or negative, different facial expressions, and we found that kisspeptin enhanced activity in certain regions of the brain that are known to be related to emotional processing, like the amygdala, other bits of the limbic system, the anterior cingulate, but specifically in relation to the sexual and romantic images — nothing else.</p><p>So what we think is that kisspeptin serves as a link between the brain systems that are processing emotional, sexual, and love-related information and the rest of the bodily reproductive system.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FeRyUvBUQ85Y%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DeRyUvBUQ85Y&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FeRyUvBUQ85Y%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/efc9a2b7a61134f96ba850a97706d370/href">https://medium.com/media/efc9a2b7a61134f96ba850a97706d370/href</a></iframe><p><strong>Dan: </strong>An important question with peripherally administered peptides is whether they actually reach the brain, as the blood brain barrier often prevents molecules from entering the brain compartment.</p><p>It’s possible to administer a medication and see changes in brain activity, but this doesn’t necessarily mean that it’s directly acting in the brain. Perhaps the peptide has an indirect effect on the brain via the activation of peripheral receptors.</p><p>To test this, the study authors administered radiolabelled kisspeptin peripherally to male mice. They found increased levels of radioactivity, in various areas areas of the brain, suggesting that these brain imaging effects were probably due to direct action in the brain, which is known to contain kisspeptin receptors. Of course, these results to be replicated in humans, but observing this in mice is a great start.</p><p>This study helps open the door for kisspeptin’s development as a treatment for reproductive dysfunction. Although researchers have already demonstrated its role in non-human mammalian reproduction, this is the first study to demonstrate that kisspeptin can influence the neural processing of reproduction related images in humans.</p><p>Subscribe to my podcast in your favourite podcast app or <a href="https://www.instagram.com/dsquintana/channel/">watch the video episodes on Instagram</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=35c89ac14602" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why I launched a new podcast]]></title>
            <link>https://medium.com/@dsquintana/why-i-launched-a-new-podcast-23f945c7bd70?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/23f945c7bd70</guid>
            <category><![CDATA[podcast]]></category>
            <category><![CDATA[psychology]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Wed, 18 Jul 2018 13:53:32 GMT</pubDate>
            <atom:updated>2018-07-18T18:11:45.764Z</atom:updated>
            <content:encoded><![CDATA[<p>I just launched a <a href="https://shows.pippa.io/dsquintana">new show</a>, where you’ll find a mix of episodes on research that’s catching my eye in psychiatry and psychology, interviews with other scientists, me sharing my daily research process, and Q&amp;A sessions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NMzhN3iG7iQgWvu3NQfMzw.jpeg" /></figure><p>There are two reasons I’ve started this show. First, a few months ago I noticed I was spending almost all my time writing papers and almost no time reading papers. My reading of papers was reactionary, in that I was reading papers only so I could cite them to prove a point or address a comment from a manuscript reviewer.</p><blockquote>“…in science there’s a disproportionate focus on the end product, like publications, compared to the day-to-day grind of getting things done”</blockquote><p>Let’s be honest. If you look hard enough you can find a paper that will support almost any point you want to make. I was beginning to get the sense that my overall knowledge of my field had ossified sometime between the end of PhD and start of my postdoc, when I had more time to read, which was occasionally supplemented by the papers I came across on Twitter.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*88PdaI7jPTfA-9IvKZ1tBQ.jpeg" /></figure><p>To pull away from this blinkered view of the literature, I started making a conscious effort to set aside time to read more papers in my field.</p><p>As part of the show, I’m going to share some highlights of what I’ve been reading and invite study authors on the show for their perspectives. It’s one thing reading a paper or seeing an scientist’s comment in print, but I think it’s whole other experience <em>hearing</em> the scientist’s voice.</p><p>The second reason I wanted to start this show is that in science there’s a disproportionate focus on the end product, like publications, compared to the day-to-day grind of getting things done. With social media it’s easy to think that people never get paper or grant rejections. But that’s not reality of science. It can frustrating, messy, and full of both ups and downs. Sometimes its not even particularly exciting. So that’s why I’m going to be sharing my day-to-day research process, which will include what I’m working on, what’s inspiring me at the moment, and tools I’m currently learning.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fplayer.pippa.io%2F5b33637a7b9cf5a0177f18b5%2Fepisodes%2Fshow-introduction&amp;url=https%3A%2F%2Fplayer.pippa.io%2F5b33637a7b9cf5a0177f18b5%2Fepisodes%2Fshow-introduction%3Ftheme%3Ddefault%26cover%3D1%26latest%3D1&amp;image=https%3A%2F%2Fres.cloudinary.com%2Fpippa%2Fimage%2Ffetch%2Fc_pad%2Cw_600%2Ch_315%2Fhttp%253A%252F%252Fs3.amazonaws.com%252Fassets.pippa.io%252Fshows%252F5b33637a7b9cf5a0177f18b5%252F1531915490998-fafca19bfd0d7c6f179e0b0403be84f6.jpeg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=pippa" width="640" height="110" frameborder="0" scrolling="no"><a href="https://medium.com/media/875683af15199b9607573d53d4158d56/href">https://medium.com/media/875683af15199b9607573d53d4158d56/href</a></iframe><p>This will primarily be an audio show, <a href="https://shows.pippa.io/dsquintana">via podcast</a>, but some episodes will also be released with video. Add me on <a href="http://instagram.com/dsquintana">Instagram</a> to see these episodes via Instagram TV and to follow my day-to-day life as a scientist. Video episodes will also be posted on <a href="https://youtu.be/eRyUvBUQ85Y">YouTube</a>. If you want to keep up-to-date with episodes, make sure you subscribe to the show on your favourite podcast platform, just search “Physiology &amp; Behavior”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ee8fQLWlUBPJEX2S9QR4sg.jpeg" /></figure><p>Hope you enjoy what’s to come. The best way to find me online is via <a href="http://twitter.com/dsquintana">Twitter</a>, my DMs are open if you want to get in touch.</p><p>Here are the <a href="https://shows.pippa.io/dsquintana/show-introduction">audio</a> and <a href="https://www.instagram.com/tv/BlX7z6tHCe5/">video</a> versions of this post.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=23f945c7bd70" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Trickle-down metabolomics: Saliva oxytocin measures do not reflect peripheral plasma concentrations]]></title>
            <link>https://medium.com/@dsquintana/trickle-down-metabolomics-saliva-oxytocin-measures-do-not-reflect-peripheral-plasma-concentrations-ab74a91cb9d7?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab74a91cb9d7</guid>
            <category><![CDATA[research-methods]]></category>
            <category><![CDATA[psychology]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[oxytocin]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Thu, 17 May 2018 19:54:12 GMT</pubDate>
            <atom:updated>2018-05-18T04:14:20.368Z</atom:updated>
            <content:encoded><![CDATA[<p>The neuropeptide oxytocin is thought to play an important role in social behavior. To better understand how the oxytocin system influences our thoughts and behaviors, researchers often collect blood samples to calculate levels of circulating oxytocin for comparison to various measures of psychological functioning.</p><p>As lower oxytocin levels have been <a href="https://www.ncbi.nlm.nih.gov/pubmed/25349162">associated with psychological dysfunction</a>, oxytocin levels might provide a useful psychological biomarker. If it’s established that lower oxytocin levels are associated with dysfunction, then pharmacologically increasing oxytocin levels may help alleviate psychological and behavioral impairments.</p><p>In parallel to this, researchers also need to demonstrate that a dose of synthetic oxytocin increases peripheral concentrations, which have been shown to be <a href="https://www.ncbi.nlm.nih.gov/pubmed/28442403">representative of oxytocin concentrations in the brain</a>.</p><p>Blood measures can be a useful tool, but blood collection isn’t a straightforward task for many laboratories. The collection and preparation of blood samples requires a specific set of expertise and infrastructure that many labs don’t possess, particularly those in psychology departments.</p><p>Also, some people don’t like needles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ykRYDoIxwc4PGmiIfzozhA.jpeg" /><figcaption>If this image is making you a little woozy, you might have a needle phobia.</figcaption></figure><p>Saliva measures have emerged as a popular alternative to blood measures. Based on the premise that circulating molecules in blood can transfer to salivary glands via surrounding capillaries, it’s assumed that saliva oxytocin concentrations can be used to index blood oxytocin concentrations. Saliva is much easier to collect than blood, as participants just need to spit into a tube or suck on a cotton swab.</p><p>Nevertheless, there are a few issues with using saliva to measure peripheral oxytocin concentrations. First, hormone concentrations in saliva are <a href="https://www.ncbi.nlm.nih.gov/pubmed/12097361">much less than what you find in blood</a>. This isn’t a huge problem if you have a senstive measure and if your saliva measure is highly correlated with its corresponding blood measure. However, the relationship between blood and saliva oxytocin <a href="https://www.ncbi.nlm.nih.gov/pubmed/23665533">is relatively modest</a> (<em>r</em> ∼ 0.5). For comparison, more established saliva measures like <a href="https://www.ncbi.nlm.nih.gov/pubmed/17953627">testosterone</a> and <a href="https://www.ncbi.nlm.nih.gov/pubmed/6762264">cortisol</a> have demonstrated much stronger correlations between blood and saliva (<em>r</em> ∼ 0.9).</p><p>Second, we don’t really understand the origin of saliva oxytocin concentration increases after intranasal administration. Clearance of intranasally delivered oxytocin from the nasal cavity down the throat (also known as “trickle-down” oxytocin) is a considerable limitation, as it’s unclear whether you’re measuring circulating oxytocin, or the synthetic oxytocin that was just administered.</p><p>I’ve run a few intranasal oxytocin trials in my short-ish career. One of the most common responses to intranasal oxytocin administration from research participants is something like, “Oh, that tastes a little bitter”.</p><p>That’s the spray being cleared from their nasal cavity and trickling past their taste buds.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/214/1*8zBiQMFt-davPw2c-XGINA.jpeg" /><figcaption>Liquids that drain from the nasal cavity are likely to pass by the base of the tongue, which happens to be the location of taste buds that are specialised to detect bitterness.<a href="https://medical-dictionary.thefreedictionary.com/a+bitter+taste+in+mouth"> Image source</a></figcaption></figure><p>Another related question is how long does trickle-down occur after intranasal administration? The use of radiolabelled oxytocin coupled with PET imaging is an obvious solution to this problem, but the development of an radiolabelled oxytocin has proven to be difficult.</p><p>An alternative to understanding the origin of salivary oxytocin is to compare salivary levels after intranasal oxytocin administration with salivary levels after intravenous (IV) oxytocin. If saliva oxytocin reflects peripheral concentrations, then IV oxytocin administration should also increase salivary concentrations.</p><p>Here’s what happens to plasma when you administer 1 international unit (IU) of IV oxytocin — which is <a href="https://www.nature.com/articles/tp201593">roughly equivalent to a standard intranasal dose</a> — compared to intranasal oxytocin and placebo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sjfr-YMf3AnGdovtct9HvQ.png" /><figcaption>Figure from <a href="https://www.nature.com/articles/tp201593">Quintana and colleagues (2015)</a></figcaption></figure><p>There’s a five-fold jump in blood plasma oxytocin straight after IV administration (green line), and then these levels closely mirror intranasal administration (red and blue lines) from about 10 minutes after administration.</p><p>Research has yet to compare saliva and blood oxytocin concentrations after intranasal and intravenous administration. With this in mind, we collected both blood and saliva oxytocin concentration data as a part of <a href="https://www.nature.com/articles/tp201593">a clinical trial</a>. I eventually got the chance to analyse and write up this data, which is <a href="https://authors.elsevier.com/a/1X3PE,QxXSaja">now published</a> in the journal <em>Hormones and Behavior </em>(<a href="https://osf.io/s7ngd">preprint here</a>).</p><h4>Our experiment</h4><p>We used a crossover design, where we gave four treatments to 16 healthy males: low dose intranasal OT (8IU), typical dose intranasal OT (24IU), an equivalent IV dose (1IU), and placebo. This means that every participant visited the lab four times, and received one of treatments at each visit in a randomized sequence.</p><p>Because these treatments used different administration methods (i.e., intranasal vs. IV), we had to mask the treatment condition by using what’s called a double-dummy design. In each treatment condition, we administered both an intranasal and intravenous solution. For example, in the low-dose intranasal oxytocin condition, we also administered 8 IU of oxytocin and IV placebo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*47t20W-lUgs4PvTaUWZmPQ.jpeg" /><figcaption>During each treatment condition, participants received both intranasal and intravenous solutions. The contents of these solutions depended on the randomized expoerimental condition the visit.</figcaption></figure><p>Both saliva and blood plasma samples were collected for oxytocin concentration analysis at 6 timepoints: baseline, 0 mins (immediately after the completion of spray administration), 10 mins, 30 mins, 60 mins, and 120 mins.</p><h4>What we found</h4><p>Overall, we discovered that intranasal oxytocin (8IU and 24IU) administration increased saliva oxytocin concentrations in comparison to saliva oxytocin concentration levels after intravenous and placebo administration.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-t303cgEQBpWJiY2aH5wHg.jpeg" /><figcaption>Saliva oxytocin concentrations after four types of treatment: low dose intranasal oxytocin, typical dose intranasl oxytocin, intravenous oxytocin, and placebo. * = <em>p </em>&lt; 0.05 compared to both placebo and 1IU IV oxytocin; ‡ = <em>p </em>&lt; 0.1 compared to placebo.</figcaption></figure><p>Immediately after intranasal oxytocin administration with a typical 24IU dose, there was almost 400 pg/mL of oxytocin detected in saliva. For reference, around 15 pg/mL is found in blood at this same time point after the same treatment. Normal levels in blood without any intervention are between 1–10 pg/mL.</p><p>We observed unusually high levels of salivary oxytocin for up to two hours after intranasal administration. It’s certainly possible that intranasal administration kickstarted the production of natural oxytocin in brain, but then this should probably also be have been reflected in blood plasma too — but it wasn’t.</p><p>We also found that saliva oxytocin concentrations were not related with plasma oxytocin concentrations after either intranasal or intravenous oxytocin administration at any of the these six timepoints. In other words, high levels of saliva oxytocin generally didn’t correspond with high levels of blood oxytocin.</p><p>If saliva oxytocin were representative of peripherally circulating oxytocin, then one would expect saliva levels to be similar after intranasal and IV oxytocin administration. However, as saliva oxytocin concentrations after intranasal oxytocin were larger compared to IV administration — even after two hours — it looks like that cleared exogenous oxytocin from the nasal cavity into the oropharynx largely contributes to these increases.</p><p>Bringing everything together, we concluded that saliva concentrations should not be used as a proxy measure of peripherally circulating oxytocin after intranasal exogenous oxytocin administration.</p><p>Method validation isn’t especially sexy, but it’s important to understand which measures can be used to accurately index measures of peripherally circulating oxytocin.</p><p>Paper citation: Quintana DS, Westlye LT, Smerud KT, Mahmoud RA, Andreassen OA, Djupesland PG (2018) Saliva oxytocin measures do not reflect peripheral plasma concentrations after intranasal oxytocin administration in men. <em>Hormones and Behavior, </em>DOI: <a href="https://doi.org/10.1016/j.yhbeh.2018.05.004">10.1016/j.yhbeh.2018.05.004</a></p><p>If you don’t have access to this journal, you can <a href="https://osf.io/s7ngd">read the paper preprint</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab74a91cb9d7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Constructing contour-enhanced funnel plots for meta-analysis]]></title>
            <link>https://medium.com/data-science/constructing-contour-enhanced-funnel-plots-for-meta-analysis-6434cc8e51d0?source=rss-91d5a7b2343a------2</link>
            <guid isPermaLink="false">https://medium.com/p/6434cc8e51d0</guid>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[meta-analysis]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[rstats]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Dan Quintana]]></dc:creator>
            <pubDate>Thu, 03 May 2018 05:51:57 GMT</pubDate>
            <atom:updated>2018-05-04T13:39:48.915Z</atom:updated>
            <content:encoded><![CDATA[<p>Meta-analyses are used to synthesize bodies of research and can carry considerable weight when it comes to directing policy. However, a big limitation with this method is that the studies included in a meta-analysis can be biased.</p><p>Publication bias is a well-known source of bias. For instance, researchers might shelve studies that aren’t statistically significant, as journals are less likely to publish these kind of results. Researchers might also use questionable research practices — also known as <em>p</em>-hacking — to nudge an effect across the line to statistical significance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zwmIOnBQblfYUhW47evGNw.gif" /><figcaption>P-hacking in action</figcaption></figure><p>Consequently, the risk of publication bias needs to be considered when performing a meta-analysis.</p><p>Two interrelated approaches are typically used to assess the risk of publication bias in meta-analysis. First, the risk of publication bias is visualised by constructing a funnel plot which visualizes a measure of effect size against a measure variance (e.g., standard error), with the funnel centred on the summary effect size.</p><p>Let’s consider the two funnel plots below in Figure 1.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m9cKB_9dMi6RIOUrsPmOmw.png" /><figcaption>Figure 1</figcaption></figure><p>Figure 1A seems to be fairly symmetrical, with roughly the same number of studies either size of the summary effect size (i.e., the vertical line in the middle of the funnel). Studies with greater variance have a larger spread around the summary effect size, near the bottom of the plot. In contrast, Figure 1B is not symmetrical. It seems that the studies with more variance (i.e., studies with fewer participants) only fall to the right of the summary effect size. Many researchers would conclude that Figure 1B is indicative of publication bias, due to the clear asymmetry.</p><p>Egger’s regression test is often used as an objective measure of funnel plot asymmetry, as it assesses the relationship between the effect size a measure of sampling variance. A statistically significant effect is indicative of funnel plot asymmetry.</p><p>Now let’s consider the two funnel plots in figure 2, each with ten effect sizes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d7GGs7a59Oyqy0uWCu-F0Q.png" /><figcaption>Figure 2</figcaption></figure><p>Both plots demonstrate asymmetry, but note the differences in observed outcomes (i.e., effect sizes). The effect sizes in figure 2A range from 0.2 to 0.65, whereas the effect sizes in figure 2B range from -0.1 to 0.35.</p><p>For a given range of effect sizes and variances, it’s possible to calculate statistical significance at for any combination of these two variables. Consequently, you can visualise a set of significance thresholds on a traditional funnel plot by constructing what’s called a <a href="https://www.ncbi.nlm.nih.gov/pubmed/18538991">contour-enhanced funnel plot.</a></p><p>The funnel plots in Figure 3 are a duplicate of the plots in Figure 2, except key areas of statistical significance have been superimposed on the funnel, and the plot is now centred at zero. The red zones show effects between <em>p</em> = .1 and <em>p = </em>.05, and the orange zones show effects between <em>p</em> = .05 and <em>p = </em>.01. Effects in the white zone are greater than <em>p</em> = .1 and effects in the grey zones are smaller than <em>p</em> = .01.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LaoWPregiv6rmW66FPiu6A.png" /><figcaption>Figure 3</figcaption></figure><p>Running Egger’s regression test on both of these plots yield identical outcomes (<em>z</em> = 2.78, <em>p</em> = 0.01), which are suggestive of asymmetry. However, there’s clear pattern in Figure 3A of studies falling into the statistical significance channel on the right of the funnel. Figure 3B displays clear asymmetry, but none of the studies are statistically significant.</p><p>Figure 3 illustrates that funnel plots and Egger’s regression tests only assess the risk of <em>small study</em> bias, insofar that smaller studies tend to have greater variance. Sure, an asymmetrical plot <em>might </em>be due to publication bias, but there could be other reasons that can better explain funnel plot asymmetry. For instance, smaller studies with greater variance are associated with poor study design.</p><p>Here’s how to make these plots for your own meta-analyses, using R.</p><h4>Contour-enhanced funnel plots using the Metafor package in R</h4><p>The <a href="http://www.metafor-project.org/">metafor R package</a> provides a straightforward means of constructing highly customisable contour-enhanced funnel plots.</p><p>The following R script will generate a standard funnel plot and three different contour-enhanced funnel plots.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/295bb10931390c07ec2fae8ee53e5618/href">https://medium.com/media/295bb10931390c07ec2fae8ee53e5618/href</a></iframe><p>Figure 4A is a standard funnel plot. Figure 4B is a contour enhanced funnel plot — note that the vertical reference line is now at zero (instead of the summary effect size, like Figure 4A). Figure 4C has added a little colour, and Figure 4D has adjusted the contour lines to a single contour between <em>p</em> = .05 and <em>p</em> = .01.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VI6lchLCAcUo3a-5OyeYqA.png" /><figcaption>Figure 4</figcaption></figure><p>One of the many benefits of using R for your analysis is that you can easily share the analysis script with your manuscript. If you’re new to R, you should check out my <a href="https://www.frontiersin.org/articles/10.3389/fpsyg.2015.01549/full">step-by-step tutorial paper</a> to performing your own correlational meta-analysis using R. I also put together a companion <a href="https://www.youtube.com/watch?v=d1pYHfCKhyA">video</a>, if that’s more your thing.</p><p>Open materials are usually promoted for the purpose of other<em> </em>scientists to reproduce your analysis. However, you’ll quickly learn that one of biggest beneficiaries of open materials is your future-self, for when you need to revisit analyses.</p><h4>Final thoughts</h4><p>Contour-enhanced funnel plots provide a handy visualisation of the pattern of statistical significance reported in a body of research. While they’re a useful addition to traditional funnel plots, which only visualise small-study bias, they cannot be used to objectively assess the risk of p-hacking, or the degree of effect size inflation in a given field. Despite these limitations, funnel plots can still provide a valuable addition to meta-analyses, given the ease at which they can be put together.</p><p>If you like this article, you’d probably also enjoy <a href="https://soundcloud.com/stream">Everything Hertz</a>, a podcast on meta-science that I co-host. Here’s our <a href="https://soundcloud.com/everything-hertz/4-meta-analysis-or-mega-silliness">episode on meta-analysis</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6434cc8e51d0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/constructing-contour-enhanced-funnel-plots-for-meta-analysis-6434cc8e51d0">Constructing contour-enhanced funnel plots for meta-analysis</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>