Fifty-nine per cent of Canadians surveyed believe that federal prisoners should not be entitled to government pensions.Fifty-nine per cent of Canadians surveyed believe that federal prisoners should not be entitled to government pensions. (CBC)

A majority of Canadians believe that federal prisoners should not be entitled to government pensions, a new survey suggests.

The poll, conducted by research firm EKOS and released Thursday exclusively to CBC News, found 59 per cent of respondents think all federal prison inmates should lose their federal pension benefits while in prison.

The issue has come under scrutiny in recent weeks after it was reported that serial killer Clifford Olson, 70, has been receiving monthly Old Age Security and the Guaranteed Income Supplement. Under the law, federal prisoners who are 65 years or older are entitled to get federal pensions.

Prime Minister Stephen Harper has ordered a review of the old-age pension system after learning that several hundred convicts and Olson, who killed 11 boys and girls in British Columbia before being imprisoned for life in 1982, are each getting about $1,100 a month.

The poll also indicated that one in four Canadians thinks only federal inmates with life sentences should lose their benefits to pensions while in prison.

About 17 per cent of respondents say all federal prisoners who are entitled to receive federal pensions should continue to do so.

The poll indicated that those who say all federal prisoners should lose their pension benefits tend to be Conservative supporters, residents of Manitoba and Saskatchewan, women, aged 45 to 64, and those who are college/CEGEP-educated.

Those who support the existing entitlement tend to be Liberal supporters, residents of Ontario, and those who are university-educated.

The survey of 909 people was conducted by telephone between March 31 and April 6 and has a margin of error of plus or minus 3.25 percentage points, 19 times out of 20.

Because of the Easter holiday, the EKOS poll had an unusually low number of respondents, leading to the larger-than-usual margin of error.