Take 100 people.
Put them on 2 buses.
Put 50 on one bus and 50 on another.
Now, presume each person is great at counting, and perfectly honest.
Ask them how many people were on their bus.
Average those numbers to find out the average on the buses.
Now put 70 on one bus and 30 on another.
But wait, how? You have 70 answers with 70, and 30 with 30. 70*70=4900 and 30*30=900. 4900+900=5800.
When you concentrate people in an area and ask them how many people they are, you end up over the real number, even if the people are perfectly honest.
Toronto, for example, has 2.5 million people, and is 6% muslim, and also 15% south asian. These will skew any numbers you ask for immigrants, etc. Consider as well Toronto is 7.2% Black, and african-canadians are, in general, not as widespread.
Beyond that you have class differences.
I am working class. Polls say 1/3rd of Canadians are.
Working class people like me end up in working class jobs, and in working class areas, and take working class transportation, like the bus.
I can assure you, from personal experience, that far more than 6% of people I see in these places are muslim, though this likely is more due to my location than my class.
All these reasons explain why you'd end up with a higher than average number of muslims when you ask "How many muslims are there" polls. In addition, urban people, I find, are more likely to report what they see (I've spoken to many people about these polls) whereas rural people almost always say "well I know in the city..." and so they too over-estimate the numbers for the same reason.
Polls like this are nonsense for that reason. They heavily imply the only reason to over-estimate muslims is...
1 - Fear, racist fear
2 - Immigration, anti immigration
3 - Hate, muslim hate
4 - Stupidity, from not knowing what is going on
The reality is, in almost every country, you can account for half or more than half of the "over-estimation" with just the math alone, and when you add in class differences, and rural estimations of urban life, you get very close to reality.
In addition to all of this, there are more issues
If these polls were done in urban areas that could explain everything with just one step.
If they ignored cell phone users they could have had an older base, and thus, skewed the results.
And lastly, in the poll I linked to above, you can see some of the answer itself in the list of 9 things people mis-estimate.
This does not come from hate, or racist attitude. This comes from what the media reports on. Everything that respondents were wrong about, every last one, was something that has been the focus of media stories.
With all that it's no wonder the numbers were not even higher.