WE ARE WINNING, Part 2
As exercise in program/project evaluation (and to calm my nerves as a citizen who feels under siege with this lockdown), I have analyzed the daily incidence of infected cases. And as a netizen, I have posted on my FB page a truncated table showing the declining infected rates, and crowed, WE ARE WINNING!
The short analysis focused on the declining infection rates from the inflection on March 24. two days straight. I know, two days are too short a period to draw conclusions from, but as the lockdown and extended lockdown were already implemented, I thought should see some real trends.
I immediately received comments from a high school classmate who was (is?) a VP for promotions in one of the biggest services group in the country. I also got comments, also almost instantly, from former college students who are now professors, and government consultants on their own.
The most incisive of all their comments was that the data set from which the analysis proceeded is too small to draw any valid (objective) conclusion. Knowing they were right on this, I nevertheless defended my post by saying I was biased for hope and good.
No one knows the extent my soul loathes confinement, and though an incorrigible Statist, still want to be free roaming the malls, and moving about. OF COURSE, the biggest come on for risking my reputation as a positivist was that this putang inang virus would be defeated, and destroyed forever, starved of hosts and dead from hunger like the wretched of the earth.
Then the heroism of our medical frontliners would be rewarded by shouting and celebrating on the streets, that WE WON ! We would pack the airconditioned malls, eat of this and that, and drinking merry soon.
THE ANALYSIS THAT WAS
The basis for the conclusion that we were winning, because the lockdown was effective is the geometric rate on the last column, declining from the high point of March 24. I thought that the social distancing and the isolation techniques were slowing the spread of the pathogen, and that the downward trend would continue.To add a hypothesis, the ambient heat of the approaching summer in the Philippines, would weaken the virus, since there are studies which suggest that it disintegrates at 56 -60 degrees Celsius. To that effect, the infected person’s immune system would be enabled to mount an effective defense against the weakened, if not dying pathogen.
THE DEBACLE
On March 27, the DOH released the number of new infections, and it was heartbreaking. There was an increase of 96 new cases, stopping the decline in its tracks, and totally reversing the trend. It even exceeded the March 24 saddlepoint, or the apex of the early curve. It was sad, I was saddened, and so were my former students. The weeping icon of FB was everywhere. And hope was devastated. I was devastated.
The death blow came the next day, March 28, when the new infection datum came in: 272 new cases! The total number of infections breached the thousandth level. I withdrew from social communication and moped, became morose, and showed general childishness. How can not the ever
increasing clampdown on social contact, and the merciless heat of the Day Star not be having a positive effect?
NOW THE COMEBACK
I would have you know, that I was trained in Demography, first in the UP Diliman’s Population Institute, under the positivistic tiger of a Dean, Dr. Mercedes Barredo Conception: statistics to the right, statistics to the left, up and down, censuses and surveys. We were taught data treatments, when they were few and misbehaved: transformations, non parametric techniques, extrapolations and interpolations, the stuff that makes humans transform into forbidding walking computers. Then I have
the honor of being advised by that prominent Harvard sociologist Dr. David Heer, and also did my postdoctoral in Demographie in Berlin’s Humboldt University.
So you understand I can’t just accept that I let my hope overrule my mind, don’t you?
The comeback is this: the sharp rise in new infections are artificially created. They increased more than they should because TESTING KITS INCREASED. The DOH has now more capability to test more suspected patients, and can do them faster, due to the arrival and deployment of medical testing supplies and processor from abroad.
I’ll try to prove it. I mean I’ll try my damned best to convince you. One does not prove in statistics, you know. You only appeal to the improbable. Everybody who took hypothesis testing in elementary statistics should know this.
STANDARD SCORES and OUTLIERS. SIGNIFICANT DIFFERENCES
The problem can be stated like this: how would you know if the increase in cases is really due to faster spread of the virus? In other words I have to show that the increase I new cases were unusual given the trajectory or the curve, or given that the daily increase in cases were of the usual amount, and that they were bound to an average number. Any unusual increase, any startling inflection in the curve, would be caused by a new factor, such as the increase in testing kits and the faster processing of results.
This is exactly what happened. Dr. Rosette Vergeire said as much in her press conference last night. It is common knowledge so I will leave the verification of this new parameter to you. Here, I would only be a statistician, and provide mathematical, or logical evidence. Call it speculative, if you want.
In the above table, are the standardized values of the daily increase from March 21 to March 29.
Whether the reference population is the limited set used initially (on the left), or the second set including the much increased cases (on the right), the results are informative, and heartwarming. When the initial cases are used as reference (M=6), the increased cases for the succeeding 2 days were unusual, suggesting that they were artifacts of increased testing capability, and NOT because the virus is more virulent.
The results are even more encouraging when we use the second reference set: the last number given last night (March 28) when Vergeire stated on national television that they had now increased the capability of the RITM and the collaborating laboratories. The ultra sharp increase to 272, representing 183 percent increase, is an outlier, something that can virtually NEVER HAPPEN if the usual conditions of testing hold.
Because the two sets of scores are too small (6 vs. 2), a non parametric, but very powerful test of significance was used, the randomization test. The difference really cannot be attributed to the probability that testing conditions for both sets were the same. Therefore, I conclude that the very very sharp increase in cases infected were due to increased capacity to test by the health ministry.
The coefficient was .03 that this was due to usual conditions, a very unusual, very improbable situation.
The short analysis focused on the declining infection rates from the inflection on March 24. two days straight. I know, two days are too short a period to draw conclusions from, but as the lockdown and extended lockdown were already implemented, I thought should see some real trends.
I immediately received comments from a high school classmate who was (is?) a VP for promotions in one of the biggest services group in the country. I also got comments, also almost instantly, from former college students who are now professors, and government consultants on their own.
The most incisive of all their comments was that the data set from which the analysis proceeded is too small to draw any valid (objective) conclusion. Knowing they were right on this, I nevertheless defended my post by saying I was biased for hope and good.
No one knows the extent my soul loathes confinement, and though an incorrigible Statist, still want to be free roaming the malls, and moving about. OF COURSE, the biggest come on for risking my reputation as a positivist was that this putang inang virus would be defeated, and destroyed forever, starved of hosts and dead from hunger like the wretched of the earth.
Then the heroism of our medical frontliners would be rewarded by shouting and celebrating on the streets, that WE WON ! We would pack the airconditioned malls, eat of this and that, and drinking merry soon.
THE ANALYSIS THAT WAS
| March Date | Total Cases | Delta | Compound r |
| 20 | 230 | ||
| 21 | 307 | 77 | 0.33 |
| 22 | 380 | 73 | 0.24 |
| 23 | 462 | 82 | 0.22 |
| 24 | 552 | 90 | 0.19 |
| 25 | 636 | 84 | 0.15 |
| 26 | 707 | 71 | 0.11 |
The basis for the conclusion that we were winning, because the lockdown was effective is the geometric rate on the last column, declining from the high point of March 24. I thought that the social distancing and the isolation techniques were slowing the spread of the pathogen, and that the downward trend would continue.To add a hypothesis, the ambient heat of the approaching summer in the Philippines, would weaken the virus, since there are studies which suggest that it disintegrates at 56 -60 degrees Celsius. To that effect, the infected person’s immune system would be enabled to mount an effective defense against the weakened, if not dying pathogen.
THE DEBACLE
On March 27, the DOH released the number of new infections, and it was heartbreaking. There was an increase of 96 new cases, stopping the decline in its tracks, and totally reversing the trend. It even exceeded the March 24 saddlepoint, or the apex of the early curve. It was sad, I was saddened, and so were my former students. The weeping icon of FB was everywhere. And hope was devastated. I was devastated.
The death blow came the next day, March 28, when the new infection datum came in: 272 new cases! The total number of infections breached the thousandth level. I withdrew from social communication and moped, became morose, and showed general childishness. How can not the ever
increasing clampdown on social contact, and the merciless heat of the Day Star not be having a positive effect?
NOW THE COMEBACK
I would have you know, that I was trained in Demography, first in the UP Diliman’s Population Institute, under the positivistic tiger of a Dean, Dr. Mercedes Barredo Conception: statistics to the right, statistics to the left, up and down, censuses and surveys. We were taught data treatments, when they were few and misbehaved: transformations, non parametric techniques, extrapolations and interpolations, the stuff that makes humans transform into forbidding walking computers. Then I have
the honor of being advised by that prominent Harvard sociologist Dr. David Heer, and also did my postdoctoral in Demographie in Berlin’s Humboldt University.
So you understand I can’t just accept that I let my hope overrule my mind, don’t you?
The comeback is this: the sharp rise in new infections are artificially created. They increased more than they should because TESTING KITS INCREASED. The DOH has now more capability to test more suspected patients, and can do them faster, due to the arrival and deployment of medical testing supplies and processor from abroad.
I’ll try to prove it. I mean I’ll try my damned best to convince you. One does not prove in statistics, you know. You only appeal to the improbable. Everybody who took hypothesis testing in elementary statistics should know this.
STANDARD SCORES and OUTLIERS. SIGNIFICANT DIFFERENCES
The problem can be stated like this: how would you know if the increase in cases is really due to faster spread of the virus? In other words I have to show that the increase I new cases were unusual given the trajectory or the curve, or given that the daily increase in cases were of the usual amount, and that they were bound to an average number. Any unusual increase, any startling inflection in the curve, would be caused by a new factor, such as the increase in testing kits and the faster processing of results.
This is exactly what happened. Dr. Rosette Vergeire said as much in her press conference last night. It is common knowledge so I will leave the verification of this new parameter to you. Here, I would only be a statistician, and provide mathematical, or logical evidence. Call it speculative, if you want.
In the above table, are the standardized values of the daily increase from March 21 to March 29.
Whether the reference population is the limited set used initially (on the left), or the second set including the much increased cases (on the right), the results are informative, and heartwarming. When the initial cases are used as reference (M=6), the increased cases for the succeeding 2 days were unusual, suggesting that they were artifacts of increased testing capability, and NOT because the virus is more virulent.
The results are even more encouraging when we use the second reference set: the last number given last night (March 28) when Vergeire stated on national television that they had now increased the capability of the RITM and the collaborating laboratories. The ultra sharp increase to 272, representing 183 percent increase, is an outlier, something that can virtually NEVER HAPPEN if the usual conditions of testing hold.
Because the two sets of scores are too small (6 vs. 2), a non parametric, but very powerful test of significance was used, the randomization test. The difference really cannot be attributed to the probability that testing conditions for both sets were the same. Therefore, I conclude that the very very sharp increase in cases infected were due to increased capacity to test by the health ministry.
The coefficient was .03 that this was due to usual conditions, a very unusual, very improbable situation.
Comments