Science on Youtube: what an audience likes…(with updates)

Posted by on Nov 21, 2014 in Articles


Most of my film work has been related to Science and Technology. But when I talk to communicators about what kind of videos are best, I get all sorts of different answers and ideas. Mostly, my impression is that nobody really knows (I am not claiming that I do!) and that many communicators are limited by policies of the Institutions they work for. But who defines these policies and based on what? Of course there are many studies on (science) communication, and many professionals working in the area have a good feel for the preferences of their audiences. But I like to look at the data from the real world if possible and see for myself. And the, or at least some, data is out there: YouTube statistics!

In preparation for a discussion with the NASA Goddard Space Flight Center (GSFC) communication team In April 2013, I did a simple study into what kind of videos on the NASAexplorer YouTube channel are most appreciated by its audience. Here is what I found.



I selected and watched a total of 62 videos found on this channel. In this pdf file I listed all the videos, their titles in column A are links to each of them. It occurred to me that I could separate the videos in five categories, based on different concepts and contents. I defined:

(1) Animations and Data (19 videos): videos that mostly CGI generated, based on scientific data and insights;


(2) Science Stories (8 videos): short videos on specific science findings, including interviews with scientists, animations and B-roll material;


(3) Documentary / Informative (10 videos): more general science stories about projects, including interviews with scientists, animations and B-roll material;


(4) Educational (17 videos): short videos specifically tailored to be used as educational material and in an educational setting;


(5) People Profiles (8 videos): short videos focusing on the scientist behind the science and the work they do;



For each video I retrieved and derived the following statistics:

(a) number of views: Nv;

(b) number of likes: Nl;

(c) number of dislikes: Nd;

(d) months passed since publication on YouTube: Mp;

(e) the average number of view per month: Av = Nv / Mp;

(f) the percentage of likes (in ‰) compared to the number of views: 1000 * Nl / Nv;

(g) the percentage of likes of the total of number of votes (likes + dislikes): 100 * Nl / (Nl + Nd).

Within each category, I ordered the rows as a function of increasing Av. The data can be found in the Tables in the pdf here.

As can be seen in columns MIN and MAX of Table 1 below, which presents a summary of the data, there are significant variations of Av in these rather small samples.


Table 1: the Median, Minimum and Maximum of the Average Number of View per Month (Av)


I next calculated the median value for Av in each category, the median being better suited for small samples with data that has shows variation as compared to the mean value. The result of this exercise is shown in column MEDIAN of Table 1 above and in graphical form in Figure 1 below.


Figure 1: Median of the Average Number of Views per Month for each category



Before discussing the results of this simple analysis, I want to stress that I know nothing about the Youtube audience in terms of its  constitution (age, gender distribution, background, etc.). I do consider it to be very likely that a large number of this audience members have a (strong?) interest in science in general and space exploration in particular.

The picture that emerges is quite simple: the Youtube audience of the NASAexplorer channel by large prefers to watch videos in the Animations & Data category.

Is this surprising? Perhaps not that much. The most popular video of the sample (and in the Animations & Data category) is entitled Magnificent Eruption in Full HD (see the first example video above, it has an Av of 449,221 views/month in April 2013). It is a 2-minute video of animated images from eruptions from the Sun, with no specific science explanation and a funky beat below it. It is short enough to keep the attention span of the internet viewer, it looks cool and sounds great: it is simple (at least at this first level).

Other popular videos in Animations & Data share the same characteristics. It makes sense I guess: “I want to ‘chill out’, space is cool, but please don’t try to explain too much.

In this category there is also a huge variation in Av: MIN Av = 212 views/month and MAX Av = 449,221 views/month (Table 1). The least popular video of the sample in this category (see here) is based on animation and real images from the Moon, but no music (at some point in the video there are voices from Apollo astronauts). That also makes sense: music (and more generally, sound) carries the emotion of a video: no music (sound) = no emotion = no interest.

Science Stories are next in the list, with an median Av of almost 2.5 times less than Animation & Data. In this category, Av varies much less in range. Next are Documentary / Informative and Educational with another factor of 3.5 lower Av, but quite a large variation in MIN Av and MAX Av. And at the bottom is People Profiles with an Av about 9 times lower that Documentary / Informative and 72 time lower than Animation & Data!

Another interesting feature is that the percentage of likes among people who vote (so people who click the “like” or “dislike” button) is almost always well into the 90%. And, the percentage of voters among the number of viewers is almost always well below 1% (see the last two columns in the pdf). It means that this statistic is not very useful, as it measures only people who have a strong reaction (mostly positive) to what they see.



I do not think anything deep can be drawn from this simple study, a more in-depth study is necessary to reach more detailed conclusions. I am most interested in performing such a study, in collaboration with other science communicators and people who understand about Audience Studies. But I do think the general trend is clear and will not change in a more detailed study: people love to chill out, be in awe, entertained and experience an emotion. After all, this is why (good) film is so engaging: it is an emotional experience. So it looks like that publishing videos along these lines can help draw attention and might help generate more interest and traffic to other videos, with more science content, and to the Institution’s activities. It also helps to fix the name of the Institution in the audience’s mind.

In this simple study I did not get any information about the actual audience. As far as I know, there is no easy way to know what it is composed of, at least as an outsider to the Institution (in this case NASA), as more information might be available to the administrators of the YouTube channel, such as geographic distribution. My guess is that the audience in this case is not representative of the general population in the sense that one has to have a pre-interest in science to go to a channel such as the NASAexplorer channel. And of course, one has to have access to a decent internet connection (nothing so annoying as trying to watch a video on a slow connection). But other than that, the audience members can be from anywhere on the planet, from any section of the society and of any age.

Finally, a thought on production value. Communication budgets within Science Institutions are generally comparatively small. The videos presented on the NASAexplorer and similar channels are therefore fairly simple in scope. This does not mean that they are of bad quality or content, but that they are made with limited means, both in terms of people and equipment. That said, I do not count in the cost of the space missions themselves, that actually provide with most of the data and images! One can wonder, does the limited budgets have an impact on the communication reach and goals?

Here is the latest adventure in science communication by the European Space Agency called Ambition:


with an Av = 1,123,614 views / month (as on 21 November 2014, film published on 24 October 2014), which at the date of publication of this post is about 2.5 times higher than the most popular video in the presented NASAexplorer sample “Magnificent Eruption in Full HD” (and that in April 2013 had been online for 7 months). For sure, Ambition is serving its goal of getting the word about the exciting Rosetta mission out there. The budget needed to produce this film is of a scale rarely seen in science communication, I think. However, the €/view ratio is probably still very reasonable. Could this be a new format, a new way to draw attention to science? It will be very interesting to see how this concept develops in the future!

Update on 18 January 2015: the new Av for Ambition is 399,152 views / month (2.8 months after posting): it dropped by a factor of 2.8 over the 2 months after posting, meaning there have been very little additional views. Magnificent Eruption in Full HD, the best scoring  video on the NASAexplorer YouTube channel in the selection of April 2013, analysed in this article, now has an Av of 143,338 views / month (28.4 months after posting): it dropped by a factor of 3.1 over the 21 months that past since the first measurement in April 2013. For some reason this last video has managed to stay very popular over the 28 months since it posting, where the Ambition seems to be rapidly declining in popularity in terms of the number of views. How exactly this must be read and what are the underlying mechanisms are not instantly clear to me.

Update on 05 April 2015: the new Av for Ambition is 213,645 views / month (5.5 months after posting): it dropped by another factor of 1.9 over the 2.5 months since the last update, again meaning there have been very little additional views. Magnificent Eruption in Full HD, the best scoring  video on the NASAexplorer YouTube channel in the selection of April 2013, analysed in this article, now has an Av of 134,739 views / month (31.1 months after posting): it dropped by a factor of 1.06 since the last update 2.7 months ago (factor of 3.3  since the first measurement 23.7 months ago in April 2013).

It is interesting to compare the viewing statistics on a daily basis, given by YouTube, even though one has to be careful as the time axis have a different spread (5.5 months for Ambition an 31 months for Magnificent Eruption in Full HD).  Access to the actual numbers would be necessary to do a true comparison. Clearly both videos show a similar behaviour, with a peak immediately after the date of posting (two peaks for Rosetta, the second peak corresponding to Philae’s landing on 12 November 2014). The actual height of that peak is lower by more than a factor of 2 for Ambition as compared to Magnificent Eruption in Full HD. In addition, that peak seems to be more spread out in time (but it is hard to judge from this graph without access to the full data). Could this reflect a difference in efficiency in spreading the word? Is it related to the fact that the NASAexplorer YouTube channel simply has more subscribers (250,000 versus 96,000 for ESA)?

Ambition Stats 20150405

Magnificent Eruption stats 20150405


Update on 24 January 2018: I presented a talk at the International Symposium for Education in Astronomy and Astrobiology (ISE2A) that was held in Utrecht, the Netherlands, in July 2017. I could not attend myself, but presented a video talk:


Together with social scientist Jan van den Bulck from the University of Michigan, I am writing a short conference paper of this talk that will soon be published in the ISE2A proceedings. Here is the ResearchGate project describing the project in more details.


Other links:
Serene Universe: an inner journey into outer space