http://hockeyschtick.blogspot.com/

I am surprised that the 5.2mm slope is “statistically significant” with a p-value = 0.00007. But that is with the large El Nino dip. The p-value for the trend since 1999 is 0.73 – meaning that this particular slope has a 73% probability of resulting from the “random” variability in the data.

The post is great because it accurately depicts the give and take between scientists.

Technical question. The slopes are so small for these lines. Surely the R2 << 1 for these lines indicating that the slope is unneccesary for modeling the data. Is this the case? Even though this is not the purpose of the post, can you show the p-values for all three slopes? I suspect that neither the 0.6 nor the 5.2 is statistically significant (p < 0.05). This would at least point out that two scientists arguing over the differences in these slopes would be having a significantly negative argument over an insignificantly different answer.

As a non-scientist I am totally comfortable with the idea that, depending on your scientific expertise, your focus, your assumptions, and your chosen mathematical and statistical approaches you may well get a different result (and reach different conclusions), than the scientist next door.

Which is why I’m totally into this post. It’s annoying in the extreme that “some folks will immediately describe differences in opinion or interpretation to the fact that someone is lying.”

Calling other people liars, cheats, crooks, crackpots and frauds is playground stuff. We really do need a grown-up discussion – in which it’s acknowledged that smart, sincere people will sometimes come to different, good-faith conclusions.

The effect is much the same. I got 5.1 mm/yr using the data with the el niño data intact, and 3.1 mm/yr removing it and leaving a gap. Similar to what you get with the fill-in data, but at least there’s a warm feeling of not having inserted other data to make it look better.

Wow. Do you really think you are doing “science?” You are taking data with breaks in it, from only one station, for only a 17 year range, cutting years of data out, and then, arbitrarily(is this word too big?), you insert data from the previous year to estimate the rate of sea level change around Tuvalu.

Also, I love how your idea of “sensitivity analysis” appears to involve no propagation of error and linear best-fit lines for such an obviously non-linear function.

Nonetheless, *I* would be remiss were I not to acknowledge the great amusement your blog provides me through your hilarious and birdbrained attempts to promote climate skepticism.

Give my regards to commenter Y.

The question is then can you make homogenity adjustments in a reasonable manner? My view is that you shouldn’t be making homogenity adjustments if you want to determine a trend. Just compute the trends for homogeneous data, and then aggregate the trends. No assumptions needed.

The current adjustment fiasco means that current temperatures are adjusted upwards, when the devices should be accurate. Feed this into the models where radiation effects are proportion to the absolute temperature to the power of 4!, and the errors are magnified.

Nick

1. Water extraction
2. It’s a coral atol atop a sea mount, that will be sinking.