Wednesday, May 18, 2005

A better way to look at polls

Well, this post was Belinda-bumped yesterday. I thought about holding it until more post-rat polls come out but that blindered, biased boob named Jim Travers pushed me over the edge when he claimed on Don Newman's Politics that the Conservatives have a ceiling of thirty and the Liberals have the same as their floor.

I don't know if I've mentioned it before but Nodice.ca is a great resource for tracking polls. They seem to post poll results soon after they're released and present them in an easily comparable format. A quick look shows that two polls published within five days of each show a ridiculous swap of support. On May 11 Decima reported a Liberal lead of 37-28 while on the 16th Robbins SCE Research (whoever the hell that is) reported a Tory lead of 35-27. That's seventeen points of lead change in five days!! Obviously both of these polls can't be right. Or at least can't provide a sustainable picture of what will happen between now and election day.

Many partisans have tried to sort through this deluge of unclear polls and have pointed to reasons to discount polls they don't like. Such as: This is the first poll done by this company; A company is run by known partisans; The poll asked unfair lead-in questions; The poll was done over too many (or the wrong) days; The poll doesn't ask for Green support; Not enough people were surveyed. There is probably a measure of truth and relevance in all of these objections. The problem is that partisan devotion often causes one to over-emphasise the objections to polls that paint a bleak picture for one's own party. It is just too difficult to objectively balance the negative effect of all of these objections. I think the better way to consider polls and the momentum they supposedly represent is through a series of moving averages. The noise from both sides should cancel each other out and provide a more gradual, and accurate trend line.

Using the numbers available at Nodice.ca for the thirty-three polls published between February 4, 2005 and May 17, 2005 I created a spreadsheet and calculated both the moving seven-poll average and the moving fifteen-poll average for the four major parties. Between February and mid-April polls were taken about every three weeks. Since then the frequency has dramatically increased to about one every three days. On one hand this may seem inaccurate because the changes in opinion for February may be represented by one of two polls while the changes in April might have twelve or fifteen data points. On the other I think it is an accurate representation because of how closely the last month has mirrored the attention given to politics during an election campaign. Each data is more likely to represent a distinct change because more attention equates to a higher chance of changing affiliations.

DateLib 7MACPC 7MANDP 7MABQ 7MA
April 11, 200535.8630.0317.3611.23
April 12, 2005 (3rd)31.5731.618.0711.66
April 23, 2005 (2nd)29.2334.717.9912.28
May 3, 2005 (2nd)31.3631.517.4313.14
May 11, 2005 (2nd)32.5130.7118.2713.31
May 17, 2005 (2nd)30.7131.1419.1412.57
Note: The ordinals (e.g. 1st) listed above distinguish between two polls listed on Nodice.ca with the same date. Lower ordinals are further down the page.

These six polls from thirty-three represent the points of interest for the two major parties. In other words they show either where one party's advantage in terms of moving average reached a local maximum or where the moving averages crossed. I'll see about getting a graph of this posted here for those who are visual-minded.

The trend(s) represented in the table seem to be: 1. Liberal lead collapsed by April 12; 2. The Conservatives reached the height of their lead by April 23; 3. By May 3 it was back to a dead heat; 4. The Liberals regained a 2-point lead by the 11th; 5. Right now we are back to a dead-heat. I've mainly ignored the lesser parties but it seems that the NDP is trending upwards.

This comparison has the obvious failing that it concentrates on a short time frame--seven polls represents as little as seven days between April 23 and April 30. So let's look at the fifteen-poll moving averages:

DateLib 15MACPC 15MANDP 15MABQ 15MA
April 20, 200532.2432.1417.6911.69
April 29, 2005
29.31
33.4718.0312.43
May 17, 2005 (2nd)
31.37
31.218.2612.81
Note: The ordinals (e.g. 1st) listed above distinguish between two polls listed on Nodice.ca with the same date. Lower ordinals are further down the page.

As should be expected the trend here is much more gradual. 1. The two main parties start in a dead-heat on April 20; 2. The Tories take a four point lead by the 29th; and 3. The parties are back to a dead-heat.

The second poll published yesterday is really an interesting technical point. It gives the Liberals a scant lead (33-31) but because it pushes different old polls out of each of the averages it gives the Tories a slight advantage in the seven poll average and the Liberals a slight lead in fifteen poll average, both compared to the opposite situation after the one poll before it. I.e. this is a crossing point, on both graphs, going opposite directions.

Conclusions:
1. Jim Travers is totally useless when it comes to poll analysis. It looks like the CPC ceiling right now is more like 34.5 and the Liberal floor is about 29.
2. Short term momentum is even while medium term momentum is barely with the Liberals.
3. Obviously the Stronach switch is a major fundamental change that might make this technical analysis quite obsolete. I'll update this again in a week or two when things settle down (or get more excited).

19 Comments:

Anonymous robbins sce research said...

I think your analysis is pretty good. However some of the assumptions made about Quebec and the Liberals are part of the problem. Every 5% disagreement in Quebec means more than 1% for the country. If you consider Liberals vs. Conservatives a disagreement here could mean a swing of 2-3% per 5%. ROBBINS doesn't agree with the other pollster numbers in Quebec. Also, who is this? when you are speaking of ROBBINS. If you are aware of Nodice than you must be aware of us, and currently it looks like are penchant for accuracy is pretty good. At this juncture we have the seat call, the initial BC Liberal NDP spread and STV nailed. We are also the only public (private actually) polling firm that nailed the U.S. election. Certainly you could have provided a fairer depiction of our efforts to date.

Glen P. Robbins
CEO

8:38 PM  
Blogger Rhetoric said...

Mr. Robbins,

Thank you very much for your comment. Here are my responses:

1. I apologise for being so flippant in my description of your company. To tell you the truth, I had never heard of your company (outside the list on Nodice.ca) before. For everyone else here is their website. If you want a more detailed idea of who they are try googling "Robbins SCE Research".

2. Your Quebec explanation goes right to my point. I'm sure that every polling company thinks that they have built a better mousetrap when comes to at least some part of judging the national political opinion. Some claims may be more valid than others. These claims are usually impossible to evaluate because for understandable proprietary reasons polling companies are unwilling to show us just how this better mousetrap is made. The point, though, is that I don't care. By averaging the poll numbers I hope to weed out the noise and establish a more legible view of where trends are moving to.

3. The other question raised for me is: What are polls really trying to do? Do they try to "nail" the results of future elections? Do they try to gauge where public opinion is moving? Or are they just a snapshot of how people would vote right now? When put side-by-side I think it should be obvious that while all options have a grain of truth the last is most accurate. I hope that my (admittedly rudimentary) analysis that created the moving averages does a better job than just one poll of accomplishing all three goals.

9:09 AM  
Anonymous ROBBINS Sce Research said...

robbins sce research

We think the snapshot in time idea has outlived its usefullness, and often serves to be as much an apology as anything else. "Nailing" the numbers is important because we are measuring the relative values of art and science in a poll. The science is of course the statistics which can be as complex as one wants it to be. The art is understanding, or getting a sense of what is happening over a particular time in history. For instance, historically speaking, the Liberals had a very difficult time getting into Quebec. I believe it was Prime Minister Wilfred Laurier who eventually accomplished Liberal penetration in that province. Quebec has always been nationalistic and like 'France' has looked to symbols like religion and language to support this. The argument of federalism vs. national sovereignty has come up again and again in our history. Mr. Trudeau and Mr. Chretien were primary players in the development and implementation of the Charter of Rights and Freedoms which Quebec was not a signatory to. Many Quebecors have always held a bitterness towards Mr. Chretien and this is evident in the challenge to federalism which was nearly lost in the early 1990's (94?). Accordingly, when you apply a particular technique to assessing the veracity of one poll to another, sometimes there are extraordinary circumstances beneath any primae facie rationales which could be significant. If I am correct in saying therefore that support is rapdily declining for Liberals in Quebec and another pollster says this is not true, it does not necessarily mean one is wrong or right, but it could mean that one has misread or another has failed to see the significance of implications of other factors on the final results. This is particularly true in Canadian polling where regional differences are so acute.

1:46 PM  
Anonymous ROBBINS Sce Research said...

As an addendum to my previous comments, I want to say something about 'trends and weeding out the noise', I'm guessing as a way to comment on politics based in part on public opinion. One thing that has to be considered, and I present this as a critic of the current polling culture, is the relationship between the establishment, including government, and the media. Governments at the provincial and federal levels purchase through agencies or crown corporations advertising from media which (at best) is oligopolistic. This relationship tends to be more captive than arm's length. The media and lobby groups (in fact one lobby group owns a mainstream polling company) have a working relationship with these mainstream pollsters. This unusually proximate triangle of interaction is driven more on the mutual validation of cause and affect, rather than the desire to create an accurate depiction of the public sentiment (in our opinion at least). Media is dominated from central Canada. CanWest Global is owned my a Jewish family with very close ties to Israel and even closer ties to the federal Liberal Party. Because the federal Liberal party has been the dominant party in Canada for most of our history, the tendency with the mainstream pollsters is to have a closer historical relationship to the federal Liberal Party than with any other party. Accordingly, it is our contention that there are more pollsters who have had historical relationships with Liberal governments and if polls are inclined to lean towards one party over another, it would be the Liberals (the establishment party), than the Conservatives, or the N.D.P.

Only Compas and ROBBINS seem inclined towards better results for the Conservative Party (in a situation where all things are apparently relatively equal.

Therefore there are more pollsters who will lean towards the Liberals than the Conservatives in Canada (and probably the United States as well). There is more media which leans towards the Liberals than the Conservatives. The end result is if my hypothesis has any validity, is that your average will become skewed as well.

6:56 PM  
Blogger Rhetoric said...

Mr. Robbins,

Your comnents have inspired me to write a new post on polling that will update my moving averages and provide a graphical representation of where I see trends going. I will also respond there to arguments you raise that I think will interest others.

The Aspers: I don't mean to jump to conclusions or put words in your mouth but I hope that you aren't implying that the Asper family acts a certain (undesirable) way because of their religion. I have absolutely no patience for intolerance and will immediately delete any future comments that even come close to this line.

Polling culture in generaL: You may be right--polling companies may be too closely related to certain parties. But, while you're pointing fingers don't you think it would be fair to reveal your own biases? I note from a simple Google search that you were an abortive candidate for the leader of the BC Reform Party. Don't you think readers of your website would like to know this? You say: "Only Compas and ROBBINS seem inclined towards better results for the Conservative Party (in a situation where all things are apparently relatively equal." Are you saying that you have a bias?

In the end all your argument does is prove my point that moving averages are more indicative of the true state of affairs than individual polls. If "mainstream" polling companies are in fact colluding with "mainstream" media to produce false or biased polls companies like yours will step into the void and produce your own polls. There are three cases for what this will create so long as your polls disagree with the "mainstream" polls. Either you are both biased, they're biased and you're being accurate, or they're being accurate and you're being biased. The bottom line is that with very little effort a moving average is more likely to be closer to the unbiased truth than any other single option.

10:18 AM  
Anonymous robbinssceresearch said...

You are being far too trigger happy about my comments regarding the Asper family. Look, governments go to great lengths to provide a composite of the people. If population for Asians is 27% in Vancouver, government will attempt to reflect that in its hiring for that region. Read any textbook on Public Administration (I have 20-30) and it is all there in black and white.

When Conrad Black ran the newspapers that the Aspers bought out, it was widely reported that he had become Roman Catholic. This has implications.

Why is it we mention the "J" word and everybody is Hitler. That is naive and silly. The fact is the Asper's (now the sons) are very intelligent people. They produce high quality publications. They have very strong ties to Israel (that's okay I like Israel). Some of my most creative (brilliant actually) friends are strong Jewish. When I am around them sometimes I consider converting!! They are very strong in their faith and their beliefs, but they are 1% of the total Canadian population producing 70% of media. That simply isn't healthy, no matter what arguments you make about the independence of the publishers. This is an argument I am putting forward which underscore media monopolization generally. Also, a high percentage of Canada's total Jewish population is in Winnipeg where CanWest Head Offices are. We are slowly seeing some aspects of government agencies including Canada Customs and Revenue moving to Winnipeg from other places like Hull, Quebec. Are you suggesting that in any other context, Hindu, Catholic, with all other considerations this would not trouble you just a little, I mean as an undercurrent to the idea of monopolization. When you link this to CanWest's high debt load to Hollinger Inc., 1.25 billion with interest rates adjusted from 12% to 9%, and the fact that the other two networks are CTV (BEll) and CBC (government), and that the mainstream pollsters are linked or owned by these same media agencies, and that media makes millions from government advertising, one can surely see that particularly in a large country with a smaller population, control of the many by a few is easily established. Theoretically, it is a cause of concern, no matter what the religion is. The intolerance you might be more concerned with is that of control, and not of commentary.

11:21 AM  
Anonymous robbnssceresearch said...

Sorry, I missed your other comments. I was never a candidate for Reform because I was not a member of the party. I was campaigning as a type of 'ghost' candidate in conjunction with the polling. If you read my news releases you would see some of my rhetoric is being implemented, particularly the healthy food stuff.

On the Compas ROBBINS end of it, I am more of a swing voter. I helped Mobina Jaffer for the Liberals in 1997. I donated my publishing offices etc. I like Paul Martin when he first came on board, but felt more inclined towards Harper during the election. I thought the media interfered with that last federal election, so personally I felt some empathy towards the Conservatives. I now believe the bias toward the federal Liberals is so strong that I am resolute to poll and publish numbers that I believe reflect honest public opinion. To be honest, I think that all the main leaders are pretty good. Personally, I could change my mind at any time about who I would vote for, but I would not change my polling results for anything.

11:29 AM  
Anonymous robbins sce research said...

R.w.M-I thought you might want to review this new ROBBINS poll-and by the way with a margin of error that is standard of between 2.5-2.5% (for a total of 5%-7%) isn't it possible that the average formula you suggest (5)pro Liberal (2) pro conservative (just for arguments sake) will invariably end up 'correct'? anyhow. Isn't the real important thing to measure is the trend upwards or down i.e. the surge of the Conservatives or the collapse of the Liberals. My point at this juncture is this, never mind the numbers think about the politic science-do we agree the Liberals are not popular with the majority of Quebec? Do we agree that there is a reasonable liklihood that western provinces will support Conservatives- a la Reform with reasonable vigour? If both of these suggestions are reasonable than we need to identify what Ontario, which has one third of seats is going to do-I submit that all things being equal the Canadian penchant for 'change' would tend to put the breeze behind the Conservatives based simply on the number of years that Liberals have been in government. Add to this the never ending sponsorship scandal and a reasonable person (let alone a political scientist) would say the force is with the Conservatives, notwithstanding the tedium of distraction. Is the question really how many seats will Paul Martin lose? He needs to make inroads with Reform Liberals like Chretien who don't like him, and the recent budget pushed him in that direction. The NDP vote is firming up, and the bottom liners are casting a jaundiced eye at the PM. I went through this in the last U.S. Presidential election where the media I felt was polling to get advertising dollars from some fairly large budgets. I don't believe this last election no go was simply about another election because the only voters who wanted one were Conservatives and some in Quebec. All of these parties needed to get their financial ducks in a row, and 18 months is probably a more acceptable time for the electorate.
I believe my numbers are an accurate and fair depiction.


May 30, 2005

ROBBINS Sce Research (1998)


Vancouver, Canada
Seattle, U.S.A.

Highlights
q Federal Liberals (26%), Conservatives (34%), NDP (20%), Bloc (15%)
q Slight majority of Canadians against same sex legislation.


Question #1-If an election were held in Canada one month from today for which leader and party would you caste your ballot?

Stephen Harper and the Conservative Party of Canada-34%
Jim Harris and Green Party of Canada-06%
Paul Martin and federal Liberal Party of Canada-26%
Jack Layton and New Democratic Party of Canada-20%
Gilles Duceppe and Bloc Quebecois-15%

Question #2- With respect to the first question, how would your characterize the degree of support you have for the leader and party you selected?

(i) A lot of support-80%
(ii) Moderate support-20%

Paul Martin and federal Liberal Party of Canada- (i) 19%; (ii) 07%.
Stephen Harper and Conservative Party of Canada- (i) 28%; (ii) 6%
Jim Harris and Green Party of Canada- (i) 03%; (ii) 03%
Gilles Duceppe and Bloc Quebecois- (i) 14%; (ii) 01%
Jack Layton and New Democratic Party of Canada- (i) 16%; (ii) 4%

Question #3- Do you want your Member of Parliament to vote for or against changes to marriage laws in Canada which would provide same sex couples with the same legal status presently enjoyed by heterosexual couples?

For-45%
Against-55%

Question #4- In your opinion, if marriage laws in Canada are changed to provide the same legal rights to same sex couples as heterosexual couples will Canada’s ‘social fabric’ and family traditions likely remain about the same as they are now?

Yes-31%
No-67%

Commentary- The Conservative Party of Canada and the Bloc Quebecois enjoy the highest degree of support with (90%) and nearly (100%) of ‘a lot’ of support. The federal Liberals and federal NDP have (80%) of ‘a lot’ of support.

Across the country the Conservatives enjoy (50%) support from British Columbia to Manitoba. East of Manitoba, the Conservatives average (28.5%) with the highest support in Ontario (34%).

The federal Liberal Party is in decline in Canada’s three most western provinces and Quebec, where the four-province average is (19%). Liberal support is highest in Ontario and the Maritimes at (38%) and (42%). Paul Martin’s Liberals average (24%) in Canada’s three largest provinces. The Conservatives under Stephen Harper are (7%) higher across the same three provinces.

The main difference between the last (2) ROBBINS polls and the majority of the mainstream pollsters, is that ROBBINS sees significantly less support for the Liberals in Quebec (where federal Liberal support is pre-Laurier low), and slightly less support for the Liberals in Ontario. This combination impacts on global averages for the Liberals by approximately (4%).

The New Democrats are showing well in all western provinces with the exception of Alberta where support is under (20%). NDP support averages slightly under (30%) in B.C., Saskatchewan, Manitoba, and the Maritimes.

Canadians in rural provinces are solidly against changes in marriage laws allowing same sex couples to enjoy the same legal marriage status as heterosexual couples. Rural regions in the country average nearly (70%) against ‘marriage status’ for gay men and women. Vancouver, Toronto and Montreal ‘proper’ average nearly (60%) ‘for’. Only Quebec and Manitoba supported this ‘new marriage proposal’ over (50%). The three most western provinces of B.C., Alberta and Saskatchewan average (35%), with Alberta the least supportive at less than one in four respondents in support.

ROBBINS has conducted numerous polls on the subject of gay marriage and this poll reveals the highest support ‘for’ gay marriage that we have witnessed to date. (It should be noted that the proper scientific term “homosexual” was not used, the term ‘against’ is usually less desirable as a choice than the term ‘for’ and the question was constructed with a slight push generally in favour of same sex marriage). It also denotes a reversal of sorts in Quebec where the number of respondents ‘for’ legal marriage for same sex couples is barely over one-half with much of this support coming from Montreal, and the majority of respondents ‘against’ in rural Quebec, where many Bloc supporters now appear to see nationalism in terms of language and religion, superceding federalism, perceived legal Rights under the Charter, and secularism as a social objective. Average totals ‘for’ same-sex marriage rights from Ontario, Quebec and the Maritimes equal (49%).

More than (12%) of same sex marriage ‘supporters’ in this poll understand that there will be changes to Canada’s social fabric and family traditions if same sex marriage is implemented. Respondents in this group cited ‘the Charter’, ‘equity and fairness’, and even the fact that “gay people are very kind” as reasons why they should support gay marriage despite the fact that “there will be some noticeable change.” Liberal problems deepen with this question as many respondents who only moderately support the party are ‘against’ same sex marriage. As well, most New Democrat supporters in areas outside mains cities in the country are ‘against’ gay marriage as well.

Insight- Stephen Harper’s Conservative Party is in a good position according to this (and previous) ROBBINS polls despite the kind of negative talk coming from the press in central Canada. He has a solid base, and if the same sex legislation is adopted in the next month you will see a Majority Conservative government seven months from now if Prime Minister Paul Martin keeps his word to hold an election. Despite some comments of chagrin from Tory ranks about the increase in nominations of ‘religious candidates’, the voters have already heard the fear mongering surrounding Stephen Harper in the June 2004 general federal election and it is more up to Paul Martin to convince the population why the Liberals should be kept around well into a second decade, than it is for Stephen Harper to apologize essentially for having a high I.Q’ (see: sarcasm for rampant press rationalizations). The populace is not comfortable with the present state of affairs in Ottawa, and if the religious folks become galvanized behind Harper, watch out!


This is a total representative scientific sample of 18,018 Canadians by way of targeted sample of respondents from a previous poll and a digit dialing of an additional 525 respondents throughout all regions of the country between May 23-30, 2005. This poll is sponsored in part by Friends of Glen P. Robbins, and Glen P. Robbins personally as part of ongoing research and business development. This survey features a margin of error of 3.5%, 18 times out of 20, @ 98% competency.


Glen P. Robbins
(604) 942-3757
-30-

12:49 PM  
Blogger Rhetoric said...

Mr. Robbins,

On the issue of the Aspers this will be the last comment (I'm sorry but this is my blog and I get the last word) made here. I am suprised that you quote such soft statistics (70% of media production--surely you realise the difficulty you would have proving this) as a pollster. Even if we assume that this concentration of ownership does exist consider the following. Most people that I know get their printed media from either home delivery or convenience stores/news stands. At least in my area both of these means of media "distribution" are controlled by Canadians who happen to hail from south- or south-east Asia. I couldn't care less. Have you noticed how many aboriginals own hockey teams? Definitely not a representative number compared to the general population. I don't expect every industry in Canada to mirror the general population, that would be silly. People should be free to work whereever the want and to buy whatever they can afford.

Thank you for posting that poll. I was not sure what you meant in your pre-amble by: "by the way with a margin of error that is standard of between 2.5-2.5% (for a total of 5%-7%) isn't it possible that the average formula you suggest (5)pro Liberal (2) pro conservative (just for arguments sake) will invariably end up 'correct'?". First, I don't accept that five polling firms are actually "pro-Liberal". Look at the Ekos poll of April 11, or the Ipsos poll of April 16 or the Decima poll from the 20th. Also, I must admit that I really don't understand the statistics theory behind aggregating margins of error.

IMHO, the "winds of change" argument won't be enough to swing Ontario. The same argument was around last time and it didn't make enough of difference. Local candidates are more important to getting Ontarians to sign on with Team Harper. I guess, though that this long pre-writ campaign might work against this by forcing voters to make up their minds before they know who their non-incumbent candidates are.

I won't comment on your poll analysis--there is just too much in there that I only slightly disagree with--but perhaps you can help me clear up some of my ignorance of polling science. What does this line in the last paragraph mean: "This is a total representative scientific sample of 18,018 Canadians by way of targeted sample of respondents from a previous poll and a digit dialing of an additional 525 respondents throughout all regions of the country between May 23-30, 2005." When were these 18,000+ surveryed? Were they all asked the same questions? In the same order?

3:38 PM  
Blogger Rhetoric said...

Also, I have two more posts on polling here and here.

3:40 PM  
Anonymous robbins sce research said...

Generally, most pollsters only telephone in the city and take that data to make assumptions about other regions. Sometimes pollsters produce statistical polls. I think many of the same sex survey polls were of this type.
If you canvass Statistics Canada extensively, you will get the idea.
Pollster may make assumptions based on area code, earnings, patterns, ethnic composition etc. In Surrey, BC we have about 30% of homes where English isn't the first language. Well it certainly isn't easy to call a poll in Surrey without considering this fact a fair depiction.

Exit polls which are often used remain flawed owing to the feeling that voters feel voting time is confidential (see U.S. Presidential election 2004).

The media concentration in BC is held by four hands CBC, CTV, CanWest Global, and David Black. Our independent newspapers are owned by CanWest and DAvid Black predominantly. CanWest has at least 70% in this province. The main news as I stated earlier in Canada is CBC, CTV (BELL Globalmedia) and Canwest Global. I think the Toronto Star is owned by someone else, and maybe the Globe and Mail.

I call attention to the federal 2004 election where a tape of Abbotsford Conservative MP Randy White was used at the end of the election. The tape was weeks old, a lifetime by media and political standards. The tape was used and advanced by all of the media because Martin was in trouble. That tape should have stood alone within the context of the interview in which it was related. The press has since gone to great lengths to somehow make a relevant piece of that election process which is wasn't.

Normally a digit dialing of a number of respondents will extrapolate into a representative sample of upward of 14 times each respondent. So in the methodology, one can call it a survey a representative survey, etc. what I am attempting to do is open the doors and windows a little on this so it isn't quite so mystical. One can decrease that out of an abundance of caution to "a fair a accurate representation" which is the legal requirement as I understand it.

If for example you are in a region with many seniors, more than above the norm, you might be confronted with less of that demographic wanting to answer questions owing in part to their fears of telemarketers. A pollster has to take all of that into consideration

4:28 PM  
Blogger jon said...

book school year are so expensive. I agree, We have been looking for book school year all night for a new book school year class but havent been able to track down used book school year that I can afford. Anyway, I enjoyed looking at you book school year blog...

jon

11:50 PM  
Anonymous Anonymous said...

Hey nice info you posted.
I just browsing through some blogs and came across yours!

Excellent blog, good to see someone actually uses em for quality posts.

Your site kept me on for a few minutes unlike the rest :)

Keep up the good work!

Thanks!.

2:54 AM  
Blogger banner ads said...

There is alot of Blogs, I never guessed I'd find some usefull information.

Thanks.
I'll be back later to see if anymore good updates are available.

4:24 PM  
Blogger cheap car insurance said...

Hey nice blog, doing a good job,

10:32 AM  
Blogger Car Insurance said...

Nice blog very informative.

I have a blog too. It is about car insurance

Thanks for the information.

10:32 AM  
Blogger Car Insurance uk said...

Good blog, Have to go to work when I get back I'll read a little more.

10:33 AM  
Blogger 性爱 said...

This comment has been removed by a blog administrator.

7:25 AM  
Blogger Auto Insurance Center said...

Hi, Thanks for your interesting blog. Keep up the great work! I also have a site & blog about auto insurance quote, please feel free to visit.

3:20 AM  

Post a Comment

<< Home