Canadian Podcast Listener Survey results
Congratulations to Sequentia and Caprica for conducting the incredibly successful Canadian Podcast Listeners Survey. It was the first independent survey in Canada to attract significant participation - 928 respondents! Congratulations are also in order for the authors’ transparency about the process they followed in conducting this survey.
Facts worth noting:
- This is the first survey I have seen that reported a higher percentage of female listeners (52%).
- A surprising 69% of Canadians continue to get their news from traditional media sources such as radio, television, newspapers and magazines.
- 46% of respondents prefer a weekly episode.
- 35% of respondents prefer a Podcast which is ten minutes or shorter.
- Advertising/sponsorship opportunities may not be attractive in Canada. Only 28% of respondents earn $60K or more per year, 48% earn less than $40K.
- Only 25% of respondents would not tolerate an advertisement in a Podcast.
- 60% of respondents are likely to buy a product or service recommended by a Podcast host that they trust.
And now, my concerns:
- The survey missed an opportunity to identify how respondents viewed sponsorship messages versus advertisements.
- The authors made an assumption that respondents have “little knowledge of video podcasts”. The questions from which this claim is made do not support this assumption.
- The section labelled Content is King presents confusing results which are derived from a question in which respondents provided their top three choices. The results presented suggest that 58% of respondents prefer re-purposed content and 69% prefer original content meaning that the bar chart breakdown exceeds 100% - something that requires a better explanation. The section labelled Canada’s Favourite Podcasts presents a similarly confusing summary.
- For consistency, I encourage the authors to focus on two chart styles - pie charts with percentages, and vertical bar charts with specific figures.
Despite my concerns (name a survey which didn’t result in some controversy), the authors did a phenomenal job. The information in this report is incredibly valuable, and the core structure should serve as the foundation of a similar effort next year to allow for a worthwhile comparison. I hope that the authors consider timing their next survey so that the release of its report can occur at PAB2007.
December 31st, 1969 at 7:00 pm
Biggest problem here, methodologically, is the sample is self selected.
Now, that said, any survey that mentions both thunderbird six and Broca’s Area cannot be all that bad…
July 31st, 2006 at 11:12 am
Hey Dave!
Can you expand on “self selected”?
Thanks,
Mark
August 1st, 2006 at 6:27 am
Yeah sorry, I used some jargon there. Research design and all that. OK, it means that the respondents selected themselves into the study. So, what happens then is you get a biased sample. (A biased sample is simply one that is not representative). While the data look ok, (for example the top 10 podcasts are not that different than say the top iTunes ‘casts) we never really know if the sample actuallly represent the population. That is key in such stuff.
Now, it would clearly be super difficult and expensive to do this with a truly representative sample. Hell, you would need a random sample, or some variant thereof.
As I said, the data look ok, and the first real try at getting a snapshot of the Canadian audience is laudible.
DB
August 1st, 2006 at 6:44 am
Now where does this become a problem. OK, the claim is made, on page umm (looks up page) 8 that 67 percent of Canadians would be interested in listening to a podcast. The sample itself does not represent Canadians (note the sex ratio, just for starters, and the other demographic descriptive statistics).
As well, there are no margins of error around the estimates and no confidence intervals. (Now I am lucky enough that I can do the math for those myself, but most people out there do not teach statistics). We are given percentages of a total and then told that 69 percent ‘of Canadians’ want this or that or feels this way or that way. One cannot make such conclusions, and if one does the confidence intervals and margins of error must be supplied. Indeed such conclusions are impossible without a proper sample.
Check the line graph on page 14. You MUST mave standard error bars around the points. Those points look different, but they may not represent real world differences. Statistically you would have to use some inferrential procedure here. (In that case you find out what are called ’statistically significant’ differences)
You cannot make conclusions based on a biased sample (in this type of research) and you cannot make claimss about data points being different without tests.
I am not trying to be a wet blanket here. Though it is so hot and humid that it feels like I am one….
August 1st, 2006 at 10:16 am
I guess it’s good, that in my summary about the report, I stuck to the term “respondents”.
August 1st, 2006 at 2:49 pm
Yes Mark, exactly!
August 1st, 2006 at 6:55 pm
Interesting points Dave. No, you’re not being a wet blanket.After staring at that report for almost a month, I’m happy to see your feedback and Mark’s. both you and Mark. It gives a different perspective.
Actually, Mark & I spoke about it over the phone last week and we had a wonderful conversation. He gave me some pointers, I explained our challenges and it was great to get his opinion before we sent the report out to the world.
I’m certain someone is doing a survey looking at podcasting in Canada by randomly selecting respondents. I’d be interested to see how similiar (or different) the results turn out to be from our own.
At the end of the day, the results of this survey gives us insight into the state of podcasting in Canada. It’s certainly not meant to be definitive.
For many, the results will confirm (or challenge) beliefs they have about Canadians and podcasts. If this survey helps to get Canadians talking about podcasting (as is evident on this blog), then we’ve done our part.
August 1st, 2006 at 7:51 pm
Leesa, thanks.
The ball had to get rolling, and that has been done, so good one your for that. I imagine there is a bit of a umm, what is the word I am looking for, a skew, (that is not the word I am looking for but it will do) towards a somewhat different demo that ‘all canadians’ Of course, it is not like I know hat the skew is though.
I like the fact that the Top 10 list is close to what we would generally figure are the most popular. That is a good sign.
Sampling is a royal pain. This is why I do experiments…. (We don’t worry about samples so much).
There’s a grant application in here somewhere…
Dave