A question of probability - help please!

Discussion in 'Taylor's Tittle-Tattle - General Banter' started by zztop, Feb 18, 2019.

  1. zztop

    zztop Eurovision Winner 2015

    Some of yopu may be aware that we have been looking after our little grandaughter for about a year now as our daughter is seriously ill and in hospital. The dad has been off the scene for some time, until last September when he went to get court to get regular contact. Contact is on Sundays. He is a total ******, but that is not the point.

    Over a period of 20 weeks he has missed contact with her 5 times. On those occasions, he says he has been ill. Fair enough.

    However, it has now come to my notice that the local football team he plays for usually plays on a Saturday, But on 5 occasions they played on a Sunday, exactly the same 5 days that he says he "was ill".

    The background to all this is actually quite interesting, if it wasn’t so serious, as he is now going for custody, but I need a little help from some of you with sharper, younger minds!

    Now I studied statistics at A Level, but it was a long time ago. Now, for the life of me, I just cannot be sure how to do what should be a relatively simple probability calculation. My mind has gone blank!

    The question is;

    What is the mathematical probability of him “falling ill” on the exact 5 days that his football team just happened to playing on the Sunday, over a period of 22 weeks? And please show your workings as I will probably use this at the next court hearing!
     
  2. Knight GT

    Knight GT Predictor extraordinaire 2013/14

    Not sure you the need probability working out. It is blatantly obvious what has happened and maybe the team this bloke plays for could help or even the league they are in would probably have team sheets from the games. How is he going to work this if he gets custody? I loved maths at school but it was a long time ago and the memory is not what it was. If you had a bet on a single day off sick it would 365/1 but if you had it as an accumulator it would be 6567580836576/1 for for somebody to pick 5 days at random for somebody to be sick on over a year!
    I'm sure some far greater minds will give a better answer than mine but I think it's fair to say that it's extremely unlikely to be sick on the same five days your team plays football!
    I wish you luck and of course your daughter as well
     
    zztop likes this.
  3. I would have thought that courts would deal with facts rather then probability - I'd suggest getting a copy of his team's fixture list and mark it accordingly.
     
    zztop likes this.
  4. Jumbolina

    Jumbolina First Team

    What a **** to prioritise playing football over seeing his kid.
     
    Robert Peel and zztop like this.
  5. Jumbolina

    Jumbolina First Team

    Odds are going to be tiny aren’t they? 4 days average sick days per year divided by working days in year is percentage chance of being sick on one given day for an average worker. Seems reasonable assumption to start with. Exact 5 day combo would be that to power of 5?
     
    zztop likes this.
  6. Otter

    Otter Gambling industry insider

    Next time his team plays on a Sunday get a mate to go along and record it. That evidence along with anything else you can find which correlates his no shows with his team's past fixtures should be a slam dunk.
     
    zztop likes this.
  7. zztop

    zztop Eurovision Winner 2015

    Thanks for the comments, and I agree wuith them but it was suggested by my barrister that I have the probablity as he will just say it is a coincidence.

    I have a feeling that it is something like, as there are 140 days in the 20 weeks, and 5 games on Sundays......

    5/140 x 4/139 x 3/138 x 2/137 x 1/136 as on his first day off sick he has 5 chances in 140, for the second he has 4 chances in 139, and so on.

    Does that make sense?

    If that is right, I'd like to show that in terms of 1 in y million, like the lottery odds.
     
  8. Otter

    Otter Gambling industry insider

    I wouldn't worry about probabilities, stick to facts and let the judge work it out. Being sick one Sunday is just one of those things but missing 5 when his football team played, you need to find out if he played, if he did then he obviously lied. If possible get the court to find out from his employer whether he was off sick on the Monday after any of these missed days.
     
  9. Jumbolina

    Jumbolina First Team

    Mine is over 1 in a billion I think, using 4 days average sick days, 262 working days, so 1.5% to the power of 5. And that's over a year as well.
     
  10. Hornpete

    Hornpete Squad Player

    Don't you need a teamsheet in any level of football, handed in to the ref, recorded in perpetuity?
     
  11. zztop

    zztop Eurovision Winner 2015

    Thanks Jumbo, but he doesn't work!
     
  12. zztop

    zztop Eurovision Winner 2015

    Thanks everyone that has taken a look but I think I have the answer. The reference to the lottery made me think that, in effect, it would be similar to picking the right number out of 140, 5 times in a row, so I found this website.

    http://www.easysurf.cc/lotodd.htm

    I stuck in 140 in the first box, 5 in the next two boxes and it came out with

    416,965,528 to 1.

    Sounds good enough to me.
     
  13. I Blame Pozzo

    I Blame Pozzo First Team

    He's lying.
    Best wishes to your daughter zz.
     
    zztop likes this.
  14. sydney_horn

    sydney_horn Squad Player

    I think you are the closest but it is impossible to calculate unless you know his actual specific level of health.

    If he is a particularly sick person then being ill on that number of Sundays is not that improbable.

    Or he could be a heavy weekend drinker who has self inflicted "illness" on Sundays.

    I think he is probably lying but proving he played football on those particular days, either through documentary or video evidence, is the only way to absolutely confirm that.
     
    nornironhorn likes this.
  15. Hornet4ever

    Hornet4ever WFC Forums Last Man Standing Winner 2018/2019

    Hope you get it sorted.

    Forgive me if I'm being a little too direct seeing as I don't know you, but your approach sounds like it could have turned into more of a personal vendetta & perhaps aimed at excluding the father further from his daughters life by claiming he was lying about being ill etc. You & your family would not have to deal with him so often, so whose purpose is this really for. (Rhetorical question).

    There might be some unstated reason you object with the principle, why not slowly let him take more of the parental responsibility in his childs life, seeing as he's the father & his overall intent is to see the child more, even though his execution of that intent has been severely lacking on occasions. Perhaps look at the framing of it differently maybe along lines of he's far from perfect but at least its starting point with a potential better future on the horizon for all concerned? He will always be her father like that or not.

    Like I say, hope it gets resolved.
     
  16. zztop

    zztop Eurovision Winner 2015

    I gave enough information to demonstrate the mathematical problem I wanted help with and I dont really want to go into it much more. But we are reluctant "parents" at our stage in life.

    But I will say that there is such a potential for conflict in my and Mrs ZZ's thought processes that we specifically asked the court to appoint a Childrens Guardian, via CafCass, to look after the interests of our grandaughter in Court, rather than us who currently have equal parental rights with both my daughter and the father.

    It's a horrible situation, and I wish it was over.
     
  17. sydney_horn

    sydney_horn Squad Player

    I don't think I explained myself properly last night.

    The basic calculation would be:

    A/B " A-1/B-1 * A-2/B-2 repeated until A = 0

    Where A is number of days the person is sick in a period and B is the period in days.

    For example, if someone is sick on 3 separate days over 3 weeks the calculation would be:

    3/21 " 2/20 * 1/19

    So choosing 3 specific days over a 3 week period is about a 1 in 10000 chance.

    It doesn't matter whether you are estimating the chances of that being 3 Sundays, 3 Saturdays or a Wednesday, Friday and Monday. The probability is the same.

    But illness doesn't work like that. It is far more likely that the days ill would be consecutive or near each other.

    Some illnesses like migraines can occur randomly for just one day. But most occur over multiple days.

    Injuries and drink related problems are more likely at weekends.

    To get an accurate probability these need to be weighted in to any calculation.

    But the thing you need to know the most is how many times was he ill during the 22 weeks other than the Sundays you know about.

    Without knowing that it is impossible to calculate the probability of those illnesses occurring on specific days.

    Bottom line is, that unless he is a very sick person, does activities on Saturdays that could lead to injuries or a heavy drinker on Saturdays, the chances of being sick 5 times on a Sunday in a 20 week period is very very remote.
     
    zztop likes this.
  18. nornironhorn

    nornironhorn Administrator Staff Member

    I've a degree in Actuarial Science and I currently work as a Data Scientist so statistics/probability is kind of my thing.

    Unfortunately though, you'll never get an overly accurate probability figure with the information you provided using just statistics.

    There are many other factors which will affect this figure, such as time of the year, weather, other sick days you don't know about (ie he could have been sick Friday, Saturday, Sunday, Monday and you would not know). His age, lifestyle, historic sick day records, line of work etc will also have an impact on the mathematical calculation, and there are many, many other factors too so you wouldn't be able to get a 'probability value' unfortunately. To use some of the values quoted in this thread in court would be madness. Most of these values assume sick days are independent and random events which they certainly aren't. It is certainly a very valid point to raise though, just don't try and attribute a figure to it.

    Taking a step back from the maths though, it looks pretty obvious what is happening. I imagine you have already gone down this avenue but it would be worthwhile checking if his local team puts match reports/photos online, either on facebook or their website. The team I play for play at a fairly low level of football yet all out lineups are put on our website each week.

    Apologies for being a dampener in terms of the maths. I hope the situation works out well for you, my best friend is going through a custody battle of his own so I have an insight into how awful they can be.
     
  19. Teide1

    Teide1 Squad Player

    Zztop from a fellow supporter, very sorry for you and you family you have the problem you have mentioned, very brave of you to share it with a group of people that you have probably never met! Many of us have grandchildren we are fortunate enough to " give back " when we have done our duties, unfortunately at this moment you are unable to do that, however let's hope things work themselves out and get easier for you!

    Hopefully our good season and results so far have given you some good moments away from your problems!
     
  20. zztop

    zztop Eurovision Winner 2015

    Thanks, you are right, I ahve already gone down the route of looking on the teams website. But no luck.

    What you say about all the parameters that could be taken into account is right, and so is what Sydney says along the same lines. But I was looking at it from a different, if simplistic, direction.

    If the team was going to play on 5 days over the 20 weeks, and if someone was to be sick on 5 days over that 20 weeks, what is the probablity that they would fall on exactly the same days. That makes it a purely mathematical calculation, I believe.
     
  21. zztop

    zztop Eurovision Winner 2015

    Thanks Teide1, but to be honest I didn't start by planning to say so much abaout the situation. But thanks anyway!
     
  22. Best wishes in your situation ZZ. My brother had similar issues with the deadbeat parents of their adopted child. I'm sure the courts will see through the BS.

    Now just get your head right on brexit.
     
    iamofwfc, wfcmoog and zztop like this.
  23. nornironhorn

    nornironhorn Administrator Staff Member

    Ah, that's unfortunate! Depending on how investigative you want to get, you could also look up the opposition websites or facebook pages for those matches and see if they have any match photos uploaded.

    In terms of the simple maths calculation, here is my working out:

    He was sick in 5 of the 20 Sundays he was supposed to have contact (we can't include 140 days as we have no idea if he was sick the other sick days of the week).

    In the 20 'match weeks' he has had, since gaining contact, 5 of those 20 matches have been on Sundays too. (Assuming he had a match each week, which may not have been the case).

    Assuming the day of a match is independent of the previous and future games (ie if a match is a Sunday one week, it isn't necessarily a Sunday the following week) then the probability of a match being on a Sunday is 1/4, or 25%, based on 5 games out of 20 being on a Sunday.

    The probability of him being sick, is also 25%, based on him being sick 1 in every 5 weeks.

    So, for a given match, there is a 25% chance it will be on a Sunday and a 25% chance he will be sick, based on the available information, which equates to a probability of 1/16 or 6.25% of both of these happening.

    However, given that the match is a Sunday, and using the above probabilities, there is a 25% chance he is sick that day. For this to occur 5 times in a row, we would simply multiply 25% to the power of 5 which gets a value of around 1/1024 or 0.1% roughly.

    A much higher value than you probably expected, and indeed a much higher value than is probably correct. However, this is the value that we get using the information available.
     
    lm_wfc likes this.
  24. zztop

    zztop Eurovision Winner 2015

    Yes, I got that. I see that I underestimated the chance of a game being played on Sunday. You are right, and that was my biggest mistake. Thereafter, I follow the rest.

    I reckon that saying " less than 1 in a thousand", sounds good.

    Thanks for your help, it is genuinely appreciated.
     
    wfcmoog and nornironhorn like this.
  25. Diamond

    Diamond First Team

    Sorry ZZ I didn't know about your daughter so my best wishes to you and your family. It must be a terrible situation. I'll PM you something about the football team.
     
  26. Beekayess

    Beekayess Reservist

    Interesting methodology Nornironhorn. Where have you taken account of the fact that he has been sick 100% of the time when the match was on a Sunday ?
     
  27. scummybear

    scummybear Reservist

  28. nornironhorn

    nornironhorn Administrator Staff Member

    I was calculating the probability of him being sick given a match is a Sunday, based on his overall sickness record (irregardless of day) in which I made the assumption that each sickness was independent.

    This value can then be compared to the actual value of 100%/1.0 to show the abnormality of his sickness record.

    Directly using Bayes' Theorem, the probability he is sick given it is a Sunday also works out at 25%/0.25.
     
  29. hornetgags

    hornetgags McMuff's lovechild

    Has the county FA affiliate got a website with the fixtures? Or can you write to the county FA and get the fixtures?

    If his absenteeism can be correlated to the Sunday fixtures then you'd have a chance.

    1 or 2 missed appointements maybe an anomaly however 5 is a pattern.
     
  30. Beekayess

    Beekayess Reservist

    On the basis that it is binomial (he's either sick or he isn't) and we are picking a specific 5 days out of 20, why isn't it :-

    (20 x 19 x 18 x 17 x 16) / (5 x 4 x 3 x 2 x 1) ?
     
  31. nornironhorn

    nornironhorn Administrator Staff Member

    Hmm I'm not too sure what you're doing here, are you assuming the probability he is sick and is not sick is equal? I.e P = 0.5 that he is sick? If so, this probability value isn't accurate.

    To use the binomial distribution, I'd imagine it would look something along these lines:

    P(sick) = 0.25
    5 sick days within 5 Sundays:

    P(5 sick) = (5! / [5! * 0 !]) * (0.25^5 )* (0.75 ^ 0)

    Not easy typing out a binomial distribution calculation and making it readable!
     
  32. HappyHornet24

    HappyHornet24 Crapster Staff Member

    I can't really add more than what has already been suggested re doing some detective work and finding proof that he was playing in matches on those Sundays. (Half term week at the moment means I have a bit more time on my hands in the evenings so just PM me if I can help at all in this respect - happy to do some digging). Regarding the probability aspect, there are clearly much more clever "maths" brains than mine on here to help with that.

    I'm really sorry you're going through such a stressful time, zz, wish you all the best and will keep my fingers crossed that common sense will prevail in the courts.
     
    zztop likes this.
  33. Jumbolina

    Jumbolina First Team

    Accepting that a person is sick 25% of Sundays just seems a wrong assumption. Surely that whole point is he isn’t really sick and is lying so you are trying to show the chances of an average person being sick on the exact days of his football matches (admittedly there is tons of missing information)?
     
  34. nornironhorn

    nornironhorn Administrator Staff Member

    I agree with what you said and that is what I outlined in my initial response. My subsequent calculations were based on the man is question's sickness record.

    Of course, you could find statistics for average amount of sick days taken per X number of days, but as you and I both said this most likely won't be a fair reflection on the individual as they could have a much worse than normal (or better for that matter) illness record. I feel you can only really get value from a population average if you are comparing it against another population, to reduce the impact of outliers, etc.

    My working out has been based on maths and nothing else, thus the assumption of the person being sick 25% of the time is accepted as it comes directly from the data. In reality, this number does seem very high, but I wanted to show what I felt was the purely mathematical way to calculate it, as opposed to using 'common sense'/bias.
     
  35. Jumbolina

    Jumbolina First Team

    Yeah not questioning the maths (you are more qualified than me) just seems a questionable first set of data (the claimed sick days by villain) to feed in initially.
     
    nornironhorn likes this.

Share This Page