#58 - May 2022 - Garden Stuff - Sandy Lang - slang@xtra.co.nz

SCIENTIFIC PROOFMay/June: Late autumn/early winter. Tidy, prune, fix fences, gates, paths. Mulch under trees, shrubs.

“It works for me…” Leaving aside wilful misinformation, there’s much confusion over the nature of scientific proof - homeopathy, fluoridation, ginseng, vaccination. Science is a heavy user of experiment and of statistics. Here’s a short ‘stats for science’ course…

Question: You have a headache, you take an aspirin, the headache goes away. Did the aspirin cure the headache or did the headache just go away? Science tries to answer this sort of question by experiment and then statistics to give a probabilistic answer. These are the steps…

Hypothesis: First, a hypothesis - ‘Aspirin cures headaches’.

Experiment: Then an experiment to see if your hypothesis survives scrutiny. (1) Repetitions – get a group of people with headaches (the more, the better). (2) Randomly split these sad people into two about-equal groups - G1 and G2. Give the G1 people an aspirin (‘treatment’), don’t do anything for the G2 people (‘control’). (3) After two hours check who still has a headache. Write down the results.

Statistics: A statistical analysis will give you the probability (P) your results, arose by mere chance – i.e., that aspirin did nothing - some people just got better, some just didn’t.

Probability: The smaller the P value, the less likely the results were mere chance and the more likely your hypothesis is right. If P=0.05 (5%) you can be 95% sure aspirin had a real effect, but there remains that 5% of doubt. It’s better if P=0.01, then you’re 99% sure aspirin cures headaches, but there’s still a 1% chance you’re wrong. Most scientists, for most purposes, accept it’s just OK to be wrong 5% of the time (P=0.05) but it’s not OK to be wrong 10% of the time (P=0.1). So, by tradition, P=0.05 is a magic number.

Bombs N’ Washing: But the critical P value really depends on how bad it is if you’re wrong. So, if the probability of rain is P=0.20, you still hang out the washing because it doesn’t much matter if it gets wet. You’ll cope with the 20% chance it gets wet but an 80% chance it gets dry. But if you found an old grenade in the bush, and you knew the chances it was live were only P=0.001, I doubt you’d take it home to show your kids…___________________________________

SCIENTIFIC PROOFMay/June: Late autumn/early winter. Tidy, prune, fix fences, gates, paths. Mulch under trees, shrubs.

“It works for me…” Leaving aside wilful misinformation, there’s much confusion over the nature of scientific proof - homeopathy, fluoridation, ginseng, vaccination. Science is a heavy user of experiment and of statistics. Here’s a short ‘stats for science’ course…

Question: You have a headache, you take an aspirin, the headache goes away. Did the aspirin cure the headache or did the headache just go away? Science tries to answer this sort of question by experiment and then statistics to give a probabilistic answer. These are the steps…

Hypothesis: First, a hypothesis - ‘Aspirin cures headaches’.

Experiment: Then an experiment to see if your hypothesis survives scrutiny. (1) Repetitions – get a group of people with headaches (the more, the better). (2) Randomly split these sad people into two about-equal groups - G1 and G2. Give the G1 people an aspirin (‘treatment’), don’t do anything for the G2 people (‘control’). (3) After two hours check who still has a headache. Write down the results.

Statistics: A statistical analysis will give you the probability (P) your results, arose by mere chance – i.e., that aspirin did nothing - some people just got better, some just didn’t.

Probability: The smaller the P value, the less likely the results were mere chance and the more likely your hypothesis is right. If P=0.05 (5%) you can be 95% sure aspirin had a real effect, but there remains that 5% of doubt. It’s better if P=0.01, then you’re 99% sure aspirin cures headaches, but there’s still a 1% chance you’re wrong. Most scientists, for most purposes, accept it’s just OK to be wrong 5% of the time (P=0.05) but it’s not OK to be wrong 10% of the time (P=0.1). So, by tradition, P=0.05 is a magic number.

Bombs N’ Washing: But the critical P value really depends on how bad it is if you’re wrong. So, if the probability of rain is P=0.20, you still hang out the washing because it doesn’t much matter if it gets wet. You’ll cope with the 20% chance it gets wet but an 80% chance it gets dry. But if you found an old grenade in the bush, and you knew the chances it was live were only P=0.001, I doubt you’d take it home to show your kids…___________________________________