In August 2020, I conducted an audience survey in collaboration with nine other authors who write fiction stories involving technology themes. We offered readers ten FREE eBooks in exchange for responses to twenty questions about entertainment habits, genre, and tone preferences. We received 1,300 responses in two weeks. In this post, I provide an in-depth analysis of the overall survey results and specific breakdowns by age, gender, and genre.

Ever since the start of the COVID-19 pandemic, I’ve been especially fascinated with how people are consuming TV, films, and novels to escape our current reality. As a fiction author, I’ve found myself wondering if entertainment habits have shifted significantly with the ongoing pandemic. For example, are audiences gravitating to more hopeful entertainment and avoiding apocalyptic content? Have they switched genres? I started asking people what they’re watching on TV and what they’re reading and found that some are trending away from darker genres (e.g., like my mom) whilst others (e.g., like myself) are leaning into entertainment with dismal end-of-the-world storylines. I remember watching the first two episodes of Man in the High Castle back when it first came out. I dropped it in favor of lighter fare. Apparently, I wasn’t in the mood for a dystopian story at the time. However, during the COVID-19 pandemic, I’m in the mood for some really dark stuff. I binge-watched all four seasons and ended up experiencing some palpable relief about my own reality. But not everyone is weird like me.

Rather than continue to speculate wildly on the matter, I decided to survey people on a larger scale and came up with the idea to run an audience survey combined with a free book bundle. Lucky for me, I lined up nine other authors eager to collaborate and learn the answers to these questions, and we got some interesting results. Note: since the survey does not offer a before-and-after snapshot for the COVID-19 pandemic, it only reveals what the surveyed audience is currently interested in consuming.

I’m going to skip right to the good stuff, but you might want to read the survey backstory (at the end of my post). Understanding the purpose of the survey from my perspective might help you see what I’m trying to ascertain with the different questions. Also, I’ve included a section on methodology at the very end.

Audience Demographics

The first section of my survey collected information on audience demographics. Basic factors such as age, gender, education, and nationality may significantly affect genre preferences and entertainment habits. Understanding such variances offers an important starting point for writing to market and selling your books.

The graph above shows the survey results by age group:

• 20 or younger – 1%
• 21 – 29 – 4%
• 30 – 39 – 8%
• 40 – 49 – 12%
• 50 – 59 – 21%
• 60 or older – 55%

More than 76% of the survey respondents were over 50 years old, which suggests that the survey audience was skewed significantly in favor of older readers.

Since most of the survey respondents came from the United States, the breakdown of the U.S. population by age, gender, and education serves as an important baseline for reviewing the results. The percentages on age above do not reflect the proportional breakdown of the adult population in the United States. According to the U.S. Census Bureau, there were roughly 54,058,263 adults over the age of 65 years old in 2019, representing 21% of the total U.S. adult population. In comparison, there were 30,219,206 adults from ages 18 to 24, representing 12% of the total U.S. population.

The make-up of the survey audience compared to the U.S. adult population suggests that the survey only reached readers of certain ages and not others. Since the survey was promoted by ten authors to their newsletters and social media platforms, the results of the survey are likely biased toward that collective audience rather than reflecting the preferences of a general audience..

Interestingly, my sci-fi mystery series attracts a much younger audience than represented by the survey. My eBook ranked the lowest among those chosen by survey respondents. We can still learn interesting things from this survey even if the audience is skewed toward a particular gender and age group. But do keep in mind that the audience surveyed may not reflect your ideal audience for your fiction.

The graph above shows the survey results by gender.

• Female – 74%
• Male – 25%
• Decline to State – 1%
• Prefer to self-describe – 0%

Roughly 74% of the survey respondents were female, which suggests that the survey audience is skewed in favor of female readers.

Again, these gender percentages do not reflect the proportional breakdown of the adult population in the United States. According to the U.S. Census Bureau, roughly 51% of the U.S. adult population is female, and 49% is male.

There may be several reasons for the discrepancy between the survey audience and the U.S. adult population. According to the PEW Research Center, more women read books then men in 2016. The results of the reader survey match up with this expectation.

As already mentioned, the results of the survey are likely biased toward the collective audience of the authors participating in the book bundle. The author with a large and active email list (more than 7,000 people) has an audience, consisting of majority of female readers over 60 who enjoy action/adventure and romantic suspense novels. Another factor may also be in play. Of the ten authors participating in promoting the survey, five of them were female, which could explain the strong leaning toward more female responses.

 

The graph above shows the survey results by education.

• Less than high school – 1%
• High school diploma – 9%
• Some collage, but no degree – 20%
• Associate degree – 9%
• Bachelor degree – 25%
• Graduate degree or equivalent – 31%
• Other – 3%

More than 50% of the respondents have a college education or higher. These results appear to match up with the survey conducted by the PEW Research Center above. Generally speaking, adults with higher levels of education read more than those with less education.

The graph above shows the survey results by country.

• United States – 72%
• Canada – 8%
• United Kingdom – 7%
• Australia – 5%
• South Africa – 1%
• New Zealand – 1%
• Germany – 1%

I was pleasantly surprised by the number of countries represented in the survey audience. The remaining countries listed in the graph above had less than 1%, which ranged from about 1-6 respondents per country.

Given the strong leaning toward respondents from the United States, we can be most confident of the survey results as representing U.S. preferences and habits.

Entertainment Habits

The second section of my survey explored entertainment habits, focusing primarily on reading, TV, and streaming services. As you review the following results, please keep in mind that the majority of the survey audience is female, college-educated, 50 years or older, and from the United States. My confidence in these results for other populations is lower given the smaller sample sizes for other groups.

For this question, respondents were allowed to select as many as were relevant. The top three modes of entertainment were books, television, and cinema and film.

About 97% of respondents preferred books, which makes sense given the incentives for completing the survey.  However, I was surprised about the low figures for social media and podcasts, but the results do match the preferences of the dominant age group of the survey. Generally speaking, I would expect younger people to consume more of their entertainment over social media, especially You Tube. I would expect the podcast results to increase for people in their mid-career since they are most likely to have a commute and need to multi-task.

To answer such questions, I decided to break down the answers by age group:

Now keep in mind that there were far fewer responses for the younger age groups versus the older age groups. That said, there are some interesting findings in this graph:

• Preference for books as a mode of entertainment appears to increase with age.
• Preference for social media and podcasts appear to decrease with age.
• Most age groups enjoy cinema and film and television around the same levels.

About 12% of respondents selected “other (please specify).” After reviewing those answers, I realized that many individuals have a broader understanding of the word “entertainment.” I tend to think of entertainment as something produced by creatives that can be consumed by an audience, but that reflects my own bias. However, there are many more activities that individuals find entertaining to do. This is where testing my survey first might have been helpful in honing some of the wording and generating more relevant results. Respondents listed other forms of entertainment to include music, gardening, cats, word and board games, crafts, travel, exercise, handiwork, nature, social gatherings, concerts, dancing, and art.

This question, along with Q7, revealed another source of confusion with word selections in my survey: what is the difference between television and streaming? There’s some overlap between the two words, but also areas where there may not be any. Since I stream most of my online content on my television, I think of it as TV along with my cable channels. Many individuals, however, stream online content from many other types of devices. Maybe someday, we’ll have a better term for shows that are not viewed on a television. Next time, I should test the survey in order to clarify some of the questions and get better results.

For this question, respondents were allowed to select as many answers as were relevant. I wanted to dig a bit deeper into the most used formats across the audience. The top four formats included eBooks, TV shows, print books, and movies. The high percentage for eBooks reflects the age/generation of the surveyed audience. I suspect, the number would be smaller for the younger generations. I was particularly surprised by the results for audiobooks, but then again, this figure likely represents an older audience. And as if we Indies don’t know this already, print books are definitely not dead.

To dig a bit deeper, I decided to break down the answers by age group:

Now keep in mind that there were far fewer responses for the younger age groups versus the older age groups. That said, there are some interesting findings in this graph:

• Preference for ebooks as a mode of consumption appears to increase with age.
• Preference for YouTube, podcasts, and graphic novels appear to decrease with age.
• Most age groups enjoy print books and TV shows around the same levels.

For this question, respondents were required to select one answer to indicate their reading frequency. I was rather blown away by the number of people who get a chance to read every day (70%), but I suspect that this figure has to do with the older composition of the audience.

For this question, respondents were required to select one answer to indicate their frequency of watching TV. About 51% of the audience watches TV more than one hour per day. Again, I suspect that the survey results were shaped by the older composition of the audience.

For this question, respondents were required to select one answer to indicate the impact of the COVID-19 pandemic on their choice of entertainment. While 71% of the respondents have not changed their entertainment habits, about 29% have made various changes.

About 23% of respondents added comments to their answers. In general, respondents expressed more time to read, watch TV, and stream online content during the global pandemic since they are unable to go to movie theaters, attend live events, travel, or watch professional sports. Several respondents mentioned, however, that they have much less time to read than before due to having kids at home or suffering from depression as a result of the social isolation. Many respondents have shifted to e-books since libraries have been closed and it seems less attractive to browse for print books at the book store. Several respondents noted a decrease in listening to podcasts and audiobooks since they no longer have a commute to get to work. Many respondents expressed a need to escape reality in their entertainment and therefore demonstrated a greater preference for lighter topics, familiar characters, and known settings. Several respondents described some innovative solutions as a result of the pandemic, e.g., organizing viewing parties over Zoom and watching content together on streaming services.

Genre and Tone

In the third and final section of the survey, I explored reader preferences on genre and tone.

The graph above shows the survey results by genre. A majority of the respondents preferred reading suspense genres or romance. Given the reader preferences for one of the participating authors, these results were not surprising. Remember the majority of that audience was female and over 50 years old.

To dig a bit deeper on genre, I decided to break down the answers by age group:

Now keep in mind that there were far fewer responses for the younger-age groups versus the older-age groups. That said, there are some interesting findings in this graph:

• All age groups appear to enjoy action/adventure and crime novels.
• Preference for comedy/satire, fantasy, and science fiction appears to decrease by age.
• Preference for thrillers, westerns, and mystery novels appears to increase by age.

To dig a bit deeper genre, I decided to break down the answers by gender:

The survey revealed some expected differences across gender:

• Whereas more female readers preferred mystery novels, more male readers preferred crime novels.
• While more female readers preferred romance novels, more male readers preferred science fiction, military/war, or thriller novels.
• There were only 14 non-binary responses, and most of those respondents preferred mystery or crime novels.

With this question, I wanted to explore interest in technology across different genre preferences. Given the audience make-up, I was pleasantly surprised that 62% of respondents (yes + maybes) were interested in reading stories about technology, especially since only 40% of the audience read science fiction and 48% of the audience read thriller. This result suggests a fairly broad tolerance for technology themes in a wide range of stories.

To dig a bit deeper, I decided to break down the answers by several genres of interest:

This graph illustrates a broad interest in technology-related topics across genre preference. As expected, there is slight higher tolerance for technology themes among science fiction and thriller readers.

With this question, I wanted to test the tolerance for learning about science and technology across all genres. Again, I was pleasantly surprised that a large majority (88%) of the audience expressed enthusiasm for learning technical details whilst reading fiction.

To dig a bit deeper, I decided to break down the answers by several genres of interest:

Again, I was struck by the large numbers of people who enjoy learning whilst reading fiction. I wasn’t surprised that science fiction readers were the most enthusiastic consumers of such information.

For this question, respondents were allowed to select as many answers as were relevant. I wanted to explore audience preferences about tone. To me, it’s interesting that so many readers selected complex as a word describing their story preferences. I was somewhat surprised by some these results and wonder about the extent to which the global pandemic is playing a role in shaping current preferences. I expected more respondents to enjoy dark and fearful stories, but it is possible that the tone preferences are skewed to the majority age group and genre preferences of the survey audience. This result could also be shaped by the global pandemic.

To dig a bit deeper on tone, I decided to break down the answers by several genres of interest:

The results by genre generally match the overall findings, but there are some interesting things in the graph above:

• Whilst most readers prefer complex storylines, thriller readers expressed the most enthusiasm for complexity. They also tend to like more serious storylines than other readers.
• It shouldn’t be surprising that readers who prefer comedy/satire like humorous stories. Duh.
• Fantasy readers expressed the most enthusiasm for hopeful and optimistic stories and came in second for liking humorous stories, but they also scored the highest for dark stories, which I found interesting.
• Fantasy, science fiction, and comedy/satire readers expressed the most (albeit low) enthusiasm for absurd storylines.

For this question, respondents were allowed to select as many answers as were relevant. I wanted to dive a bit deeper into tone to probe reader expectations for story endings. Given the audience make-up and majority preferences, I wasn’t all that surprised by the results with one exception. I was under the general impression that readers hate cliffhanger endings (as I do), but about 15% of readers enjoy such stories.

To dig a bit deeper on ending, I decided to break down the answers by several genres of interest:

Again, nothing surprising here. Most readers, regardless of genre, prefer hopeful, happy, or at least bittersweet endings at the moment. And it’s striking how evenly distributed those preferences are.

The graph above shows the survey results by gender preference. Given this audience, I expected more readers to express a preference for female leads. However, 78% expressed no gender preference for the main character.

For this question, respondents were allowed to select as many answers as were relevant. Given the majority genre preferences, these results were not surprising with one exception. About 48% of respondents expressed a preference for a brilliant scientist to play the main character.

The Survey Backstory

As with many things in life, my decision to conduct an audience survey begins with a personal anecdote. If you want to skip over this part, simply jump to the next headline. But you’ll miss the “why” of the survey from my perspective.

In August 2019, I came up with the idea to write a technothriller about nuclear weapons and artificial intelligence. For my day job, I’m a national security expert and have spent the past twenty plus years studying, writing, and analyzing existential threats to the United States including nuclear weapons, biological weapons, terrorism, the impact of new technologies such as artificial intelligence (AI), and yes… also, global pandemics.

At the time, I was researching and writing a comprehensive study on the impact of AI on weapons of mass destruction. To hook my non-fiction audience into a rather abstract and quite nerdy paper, I wrote a fictional scenario that takes place in 2033. I imagined how the U.S. Department of Defense might someday decide to develop an automated command and control system to launch U.S. nuclear weapons in response to an attack. In the future, the decision-making window for launching nuclear weapons will shrink to just a few minutes. This will happen as our adversaries increase their reliance on AI-enabled systems–including lethal autonomous weapons for use on the battlefield (aka killer robots). These systems increase the speed of warfare and will place incredible pressure on U.S. policymakers faced with the decision of launching a nuclear strike in retaliation to an attack. For this reason, I could envision the Defense Department seeking to automate aspects of nuclear decision-making in order to save precious time.

A week after including this gripping scenario in the introduction of my paper (not yet published), two nuclear deterrence experts actually proposed the Defense Department develop such a system. That’s when I decided to write the scenario as a technothriller novel, Rescind Order, which I recently published on June 30, 2020. By writing this story, I hoped to help raise awareness among the American public about the growing risk of nuclear weapons and the potentially devastating combination of nuclear weapons with AI–before it’s too late. I left my long-time job at the Department of Defense because I wanted to explore the power of storytelling for national security impact. But there was a problem with this plan as it pertains to my latest novel.

You see… not everyone likes to read technothriller novels.

If I want to reach the broadest possible audience with a story about nuclear weapons and artificial intelligence, I will have to tell the story in a variety of genres, from different points of view, and with different tones. In September 2019, I launched a project called American Doomsday with the aim of reaching a wide range of ages and preferences and educating them about this important national security issue. In addition to writing a technothriller novel, which my friend is currently adapting to screenplay, I decided to write a dark comedy stage play called American Doomsday. With a theater show, I hope to target actual decision-makers at the United Nations and across the U.S. government in New York and D.C. and raise awareness about this future risk.

By early 2020, I’d developed a pretty cool Kickstarter campaign to raise funds to produce the American Doomsday play as an audiodrama podcast, whilst the start of global pandemic was just getting started in China. The Kickstarter campaign was supposed to launch on April 1, 2020. By early March, I’d lined up an impressive media campaign to get the word out. And of course, as you would expect, the campaign was also called American Doomsday. This was intended to be a tongue-in-cheek play on the doomsday device featured in Dr. Strangelove, the classic film by Stanley Kubrick about nuclear deterrence, but the timing of my project added an unwelcome real-world twist.

By mid-March 2019, the COVID-19 global pandemic was already in full swing in the United States. National sports events were cancelled. Universities across the entire U.S. were closed, sending students home with less than a few days to adapt to their new realities. Even Disney World shut down, a clear sign of the apocalyptic times we’re living in.

The timing for my Kickstarter campaign was terrible. I just couldn’t stomach asking people to support a campaign warning them about a doomsday device that might lead to the end of the world with nuclear weapons. Not when we were facing another kind of terrible and deadly doomsday. This would also not be the right time to try to get a new theater show off the ground. As of the current moment, Broadway shows are closed until June 2021. So I decided to cancel the Kickstarter campaign indefinitely and regroup on my plan to reach other types of audiences. That’s how I came up with yet another idea to get my story in front of a broader audience.

Not everyone likes to watch stage plays anyway.

In May 2020, I decided to adapt my technothriller novel, Rescind Order, as a graphic novel (aka a long comic book) that would appeal to younger audiences. I started to ask myself some important questions. How should I alter the tone and content of the story to target younger people? What types of characters would appeal to them? What should the title be? I knew that in order to target a specific audience, I would need to better understand audience demographics and learn what ages, genders, and types of readers are attracted to different kinds of stories.

By this point, you’ve figured out that I’m planning a novel, a film, a stage play, an audiodrama, and a graphic novel using the same technical plot points, but with different characters, in different genres and with different tones. To reach a broader audience for a story about nuclear weapons and AI, this approach sounded reasonable to me in principle. But effective implementation would require some accurate data about audience preferences. I’d have to know what genres and characters are preferred across age groups and gender types. I’d have to understand the subtle differences in reader expectations for tone and technical content across a variety of genres. And most importantly, I’d have to establish if there is a broader audience for consuming stories about nuclear weapons and AI in the first place.

When I started to research audience demographics online, I found that there wasn’t a great deal of useful data available without spending a personal fortune to gain access to it. And the data I found wasn’t specific enough to answer my questions. For this reason, I decided to run an audience survey and generate my own data.

The Survey Methodology

As an academic, I have to offer all the usual boring caveats that come with any data generated by a survey instrument, describe the specific methods I used, and raise any concerns that I have with my own work that may have any bearing on the accuracy of the results.

To conduct an online reader survey, I chose an annual subscription (Advantage Annual at $409.35 per year) to Survey Monkey.  There are definitely less expensive options for running a survey, but I wanted readers to respond to a set of uploaded images in three of the survey questions. Survey Monkey supports this feature rather easily, and that was the main factor in my decision. Ironically, two of those three questions did not generate the results I was hoping for, so perhaps, I could have saved myself the money.

To get a higher rate of response, I asked nine other authors to join me in promoting the survey. Each of us offered a free eBook. That meant respondents could select up to ten free eBooks (from their covers only) in exchange for completing the survey. By collaborating with other authors, I hoped to increase the incentives (more free eBooks) for completion but also to expand its reach. The more eyeballs, the more responses, the better the results. And I was right because we got 1,300 responses in two weeks.

As you review the survey results, it is important to keep three caveats in mind for the design of any survey instrument and the data generated from it: 1) reliability; 2) validity; and 3) confidence level. Data from a survey instrument that does not have high levels of reliability, validity, and confidence should be scrutinized carefully for bias, error, and many other inference problems.

Reliability refers to whether the results of a survey would be reproducible–i.e., whether the survey would generate the same results under similar conditions with a similar audience. A survey instrument with high reliability would be able to identify changes in reader preferences and entertainment consumption over time, assuming the survey was conducted more than once with an appropriate sample size each time.

I think my survey has a medium-to-high level of reliability with one caveat.

I conducted the survey only one time during the COVID-19 pandemic for a brief period of two weeks–from August 15 to 29, 2020. While the survey includes a question about how entertainment habits have changed during the COVID-19 pandemic, most of the survey questions reveal current entertainment preferences. They do not address how preferences may have changed since the start of the pandemic or how they will change again once it’s over. If I were to run the survey multiple times during and after the COVID-19 pandemic, I believe we would see major shifts in entertainment preferences during and after a national crisis. Given that we’re living in extraordinary times, I don’t really know if the results in the survey are specific to our current crisis or if they would remain stable over time.

Validity examines the extent to which a survey instrument measures specific behaviors that we’re seeking to understand–i.e., that it produces data that accurately reflects peoples’ entertainment habits. For validity, the structure of the questions and their formulation matter a great deal. Survey instruments with low validity tend to ask leading questions, biased questions, or vaguely worded questions. To ensure the right questions are asked, it’s always a good idea to have other experts review a survey instrument and to test it out on a small audience.

I think my survey has a medium level of validity with a few caveats and notes.

The first is about testing. Though I had several people look over the layout and questions in the survey (you can see the questions here: Survey Monkey Genre Survey), I approached the survey design with insufficient rigor for producing new knowledge in an academic field. That said, I do have experience in conducting surveys and understand the challenges of survey methodologies from acquiring a Ph.D. in international relations. I am confident in the results for satisfying my curiosity about genre preferences and entertainment habits for people in the United States. In reviewing the survey results, I found a few problematic issues. There were two questions that did not seem to work out that well (#17 and 18). If I were to run the survey again, I would rethink those questions and reformulate them to elicit the right data. For now, I’ve decided not to analyze those results.

I also conducted the survey with a specific goal in mind (see the survey back story above). For this reason, the results may not be directly relevant to your fiction audience, genre, or specific themes in your writing.

Upon reviewing the results, I discovered a minor data bias problem. Since one of the authors involved had a particularly large and rather active email list (more than 7,000 people), the results appear to be skewed to reflect his reader audience rather than a more general one. A significant majority of the survey responses are from female readers over 60 who enjoy action/adventure and romantic suspense novels. This specific leaning will become somewhat apparent in the results. Due to this bias, the survey results would likely shift depending on the specific audience make-up (size and genre focus) of the ten authors involved in promoting the survey on social media and in newsletters. To correct this issue for my next audience survey, I may decide to focus on a specific genre and choose authors with similar-sized platforms.

Since many respondents enjoy a wide variety of genres, we can still learn a great deal from the results despite the bias, but it’s important to understand that the audience may not accurately represent the general reading population in the United States as a whole.

A major feature of validity is determined by the sample size (total number of responses), which produces a level of confidence in the survey results. The confidence level refers to the probability that the sample size accurately reflects the attitudes of a larger population. Since we’re fiction authors and focus on selling books written for adults in national markets, we’re primarily concerned with adults who read books in a particular country.

Since roughly 72% of the survey respondents came from the United States, I’ll use the U.S. as the guide for understanding the results. According to the PEW Research Center, about 73% of adults in the U.S. read at least one book per year in 2016. As of 2019, the number of adults (18 or older) in the U.S. was 255,200,373. Based on the PEW study, that means only about 186,300,000 of adults in the United States read at least one book per year.

This is the population we seek to understand for our purposes; this number will determine the sample size needed to produce a certain level of confidence with a specific margin of error–i.e., the range that a population’s response might deviate from answers of the sample. Using the calculation tool on Survey Monkey, the minimum sample size for our population at a confidence level of 95% with a 3.2% margin of error is 938. Since my reader survey generated 1,300 responses (and 936 are from the U.S.), we can have a pretty good level of confidence in the results for the U.S. market.

A survey instrument designed with high reliability and validity should produce tight clusters of data across similar groups of people. Wide variation in survey results could point to issues with reliability or validity.