For all participants
Please be informed that the purpose of the exercises through asking hand writing of some sentences is to let the participant to share and write the most important information included in the lecture. The purpose is not to give a rate or score but just to share. The system is not able to distinguish between correct or wrong answers in this particular aspect, therefore we deleted the rating or giving a score in the exercises that need hand writing. We apologize if we caused inconvenience to you by feeling doin wrong answers.
understood but little confuse between mean plus minus SD and CI. Could you please also brief about MCQ Question 4,9 and 10.
median, quartile and interquartile range??
Dr .Laxmi Ø Happy to hear from you. I am waiting your share since a long time.Ø For the difference between mean ±SD and CI, please read the questions and answers of the discussion below. I have answered this question clearly , just read the discussion very carefully.Ø For MCQ #4 · The number of patients is 100 · The mean weight is 70 · The SD is 10 · One SD is 10· Two SD is 20 · So , in a sample of 100 with mean ± SD weight of 70 ±10 : This means that two thirds of the patients (68 )have a weight ranging between one SD below and above the mean ,i.e. : 70-10 =60 and 70+10=80.And also means that 95 patients will have a weight ranging between 2SD below and above the mean, i.e. 70-20=50 and 70+20=90. Ø For MCQ #9 · The SEM deals with means and does not deal with true values. For example, in one study containing 100 patients, there is only one mean, but for calculation of SEM, the computer will imagine several calculations for other samples of 100 patients from the same population and calculates the means of those other imaginary patients and show the possible error among those imaginary means .· Therefore, SEM is used in cross sectional studies that deal with big number of cases because it deals with means and not with true values. It is imaginary calculation by the computer, but the study is only one study and the patients are the same patients.· Ø For MCQ #10 · Suppose you have the study of 20 patients who were subjected to a surgery.Suppose complications are noticed in 5 patients. When you present your results, you have to say the frequency of complications is 5/20 and do not say the frequency of complication is 25%. This is because in a small sample size, any number will give a very high percentage that may give false impression of very high good or bad results.
Dr. Laxmi
Please let me know if you are satisfied and the answers are clear or not. I waiting your response.
Another question please:
In week 2 exercise, I answered some of the questions in different way from the model answer but correct and finally I found my score 0% !!!..( my understanding 2 of them was wrong and the rest is correct).
Could you please double check if you don’t mind?
Thanks
Dr. Moustafa
Please be informed that the purpose of the exercises through asking hand writing of some sentences is to let the participant to share and write the most important information included in the lecture. The purpose is not to give a rate or score but just to share. The system is not able to distinguish between correct or wrong answer, therefore we deleted the rating or giving a score in exercises that need hand writing. You are correct and I do thank you very much for this cleaver note.
Prof Dr. Shokier
Thanks for your great lecture.
I have a question please:
As 95 % CI calculation depends on SEM not SD , so why when we present the normally distributed continuous data , we can say 95% of people expected to have values 2 SD values below and above the mean! so now saying 95% is the same like saying 95% CI , Am I correct?
sorry little bit confused with this point , needs more clarification please!
Dr. Moustafa
SD deals with the values themselves, while CI deals with means of the values not the values. In case of DS, 95% have values 2SD below and above the mean value.. In case of CI, 95% of the means of the values rage between 2SEM below and above the mean value.
Dr. Moustafa
Exactly Yes. This is correct regarding presentation of normally distributed continuous variable, but there are other uses of CI that you will know in the upcoming lectures.
Dr. Amr
There are 2 ways to present the normally distributed continuous data:
1- The first is by the use of mean ± SD. Under such situation, the values will be like that:
– 2/3 are expected to have values one SD below and above the mean.
– 95% are expected to have values 2 SD below and above the mean.
Please note that SD deals with true values and not with the means of the values.
2- The second is by the use of SEM and 95% of confidence interval (CI).
The CI is the range of values that fall within two SEMs above and below the mean, i.e. mean ±2 SEM.
Example:
– If a mean weight of a sample of 100 men is 75kg, and the SD is 10kg. The CI will SD/√n= 10/√10=1. So we can say the mean weight is 75kg (95% CI 73 to 77 kg).
– This means that if random samples of 100 men were repeatedly drawn from the same population, approximately 95% of these samples would be expected to have mean values between 73 and 77 kg.
Please note that SEM deals with means and not with true values.
It is only one study with only 100 patients with only one mean.
For calculation of the CI, the computer will generate a program to imagine that if random samples of 100 men were repeatedly drawn from the same population, we will be 95% confident the range of the mean values will fall between 73 and 77 kg.
Again, it is only one study with 100 patients with a mean of 75 kg, but there are 2 ways of presentation of the data of the same study:
– One is by the use of SD which deals with the true values.
– The other is by the use of SEM and CI which deals with the means of the values in an imaginary way generated by the computer.
In the usual studies containing up to hundreds of cases we use SD, but in cross-sectional studies containing thousands of cases we use CI.
Dr. Amr
Please let me know if the answer is clear or not.
Thank you very much. Please keep always in touch through asking questions and writing comments to know your impression and also the impression of all other participants.
Prof. Dr. Shokeir,
thanks for rich lecture
l’ve some queries …
In right-skewed data, can we say that most values are clustered around the left side of the distribution while right of distribution is longer? And if so! What is the significance of that?
Are we the ones who determine whether a study to be normally distributed or free-distributed, or the parameters?
In dispersion of distribution in example II, change of SD from 0.3 to 0.4. You mentioned that would change the shape of the curve without its position, what is the significance of that?
Is the center of distribution = center of gravity?
Dr. Amr
Thanks for rich lecture.
You are most welcome.
In right-skewed data, can we say that most values are clustered around the left side of the distribution while right of distribution is longer? And if so! What is the significance of that?Your statement is not correct. Suppose that we have a study including 100 patients, 95 patients have a weight between 70 to 90 kg and only 5 patients have a very high weight between 250 to 300 kg. Under such situation the mean value of the 100 patients will be high because the mean is the sum of weight of the 100 cases divided by 100. But the median will not be affected because the 5 cases with very high values will be calculated as only 5 cases above the median. Then, the mean will be bigger than the median and it will be located on the right side of the median on the curve and so called right skewed data.
The significance of that is to know the distribution, whether symmetric or normally distributed or asymmetric or non-normally distributed. Symmetric or normally distributed data are presented as mean +/- SD and we use for them parametric statistics, while asymmetric or non-normally distributed data are presented as median and range and we use for them non-parametric statistics..
These information will be more clarified in the upcoming lectures with very clear examples.
Are we the ones who determine whether a study to be normally distributed or free-distributed, or the parameters?The author will check the variables by asking the computer to calculate the mean +/- SD. If double SD is less than the mean, then the data are normally distributed or symmetric and we can use parametric statistics. On the other hand, if double SD is bigger than the mean, then the data are asymmetric and we have to use non-parametric statistics. This issue will be more clarified in the upcoming lectures.
This is a rough way to estimate the distribution of data that you can easily do by your self once you look at the mean and SD in any table while you are reading any article. Very frequent mistakes are done by the authors during presenting their data and this will result in using wrong statistics and wrong conclusion of the their studies.
There is a special program to check normality of distribution of data and will be explained latter on by Professor Ahmed Harraz.
In dispersion of distribution in example II, change of SD from 0.3 to 0.4. You mentioned that would change the shape of the curve without its position, what is the significance of that?It has no practical significance, just to know how change in the mean and the SD will affect the distribution of data. It is just a theoretical information of no practical value. Just to know.
Is the center of distribution = center of gravity?Yes, it is the mean.
What is the significance of SEM in a study?SEM is used in cross-sectional studies with thousands of cases to assure precision of the estimated mean of the population. It deals with multiple means, while SD deals with true values. in our usual studies containing hundreds of cases, we use SD not SEM.
This information will be more emphasized in the upcoming lectures.
Thank you Dr. Amr for your questions. It will give more chance for other participants to clarify the information. Next time, please ask the questions one by one to make it easy for me to reply.
greeting professor A shokair its an honar to be with you in such an important course. i need more details about confidence interval as i cannot understand its function and value,,, many thanks.
DR. Yasser
Thank you very much for the nice words.
More details about confidence interval will be given in the upcoming lectures.
Please do not worry. You will be completely satisfied through giving many examples. Just hear the next lectures and do the exercises and answer the MCQs.
Advice from the course director:
Before you answer the questions, please hear the slides several times ,one by one.
I am waiting comments from the participants.
Thank you doctor for this rich informative lectuer,
could you please expalin the interquartile rang and is there an equation to cululatie the median or it is just the mid point of any location
Dr .Basma – Yes , the median is the midpoint of any location (50th percentile) – The calculation of the median alone is not sufficient , we have to determine two other points , this could be :
1. Minimum – maximum.
2. The interquartile range (25th and 75th percentile of distribution)
3. The 10th and 90th percentile.
– For Example, in a sample of 100 patients if the median weight is 75 kg and the interquartile range is 62-88 kg (the example is virtual and the values are virtual).
– This means that the midpoint of distribution (50th percentile) is at 75 kg at the point of 62kg, 25% of the 100 patients will have weight less than 62 kg and 75% of the patients will have weight more than 62 kg.
At the point of 88kg, 75% of patients will have a weight less than 88 kg and 25% will have a weight more than 88kg.
All the calculations are done by computer. You just request the test and the answer will come automatically. I am just explaining the meaning of the statistical expressions you are dealing with. This basic information will help you during your entire scientific career.
Dr .Basma
Please let me know if the answer is clear or not.
For all participants
Please be informed that the purpose of the exercises through asking hand writing of some sentences is to let the participant to share and write the most important information included in the lecture. The purpose is not to give a rate or score but just to share. The system is not able to distinguish between correct or wrong answers in this particular aspect, therefore we deleted the rating or giving a score in the exercises that need hand writing. We apologize if we caused inconvenience to you by feeling doin wrong answers.
Good Afternoon Prof.
understood but little confuse between mean plus minus SD and CI. Could you please also brief about MCQ Question 4,9 and 10.
median, quartile and interquartile range??
Dr .Laxmi Ø Happy to hear from you. I am waiting your share since a long time.Ø For the difference between mean ±SD and CI, please read the questions and answers of the discussion below. I have answered this question clearly , just read the discussion very carefully.Ø For MCQ #4 · The number of patients is 100 · The mean weight is 70 · The SD is 10 · One SD is 10· Two SD is 20 · So , in a sample of 100 with mean ± SD weight of 70 ±10 : This means that two thirds of the patients (68 )have a weight ranging between one SD below and above the mean ,i.e. : 70-10 =60 and 70+10=80.And also means that 95 patients will have a weight ranging between 2SD below and above the mean, i.e. 70-20=50 and 70+20=90. Ø For MCQ #9 · The SEM deals with means and does not deal with true values. For example, in one study containing 100 patients, there is only one mean, but for calculation of SEM, the computer will imagine several calculations for other samples of 100 patients from the same population and calculates the means of those other imaginary patients and show the possible error among those imaginary means .· Therefore, SEM is used in cross sectional studies that deal with big number of cases because it deals with means and not with true values. It is imaginary calculation by the computer, but the study is only one study and the patients are the same patients.· Ø For MCQ #10 · Suppose you have the study of 20 patients who were subjected to a surgery.Suppose complications are noticed in 5 patients. When you present your results, you have to say the frequency of complications is 5/20 and do not say the frequency of complication is 25%. This is because in a small sample size, any number will give a very high percentage that may give false impression of very high good or bad results.
Dr. Laxmi
Please let me know if you are satisfied and the answers are clear or not. I waiting your response.
Dr. Laxmi
For the median and interquartile range, please also read the discussion.
Another question please:
In week 2 exercise, I answered some of the questions in different way from the model answer but correct and finally I found my score 0% !!!..( my understanding 2 of them was wrong and the rest is correct).
Could you please double check if you don’t mind?
Thanks
Dr. Moustafa
Please be informed that the purpose of the exercises through asking hand writing of some sentences is to let the participant to share and write the most important information included in the lecture. The purpose is not to give a rate or score but just to share. The system is not able to distinguish between correct or wrong answer, therefore we deleted the rating or giving a score in exercises that need hand writing. You are correct and I do thank you very much for this cleaver note.
Thanks Dear Prof for your follow up, reply and clarification
Dr. Moustafa
You are most welcome.
Prof Dr. Shokier
Thanks for your great lecture.
I have a question please:
As 95 % CI calculation depends on SEM not SD , so why when we present the normally distributed continuous data , we can say 95% of people expected to have values 2 SD values below and above the mean! so now saying 95% is the same like saying 95% CI , Am I correct?
sorry little bit confused with this point , needs more clarification please!
Thanks.
Dr. Moustafa
SD deals with the values themselves, while CI deals with means of the values not the values. In case of DS, 95% have values 2SD below and above the mean value.. In case of CI, 95% of the means of the values rage between 2SEM below and above the mean value.
Dr. Moustafa
Please read the answer of the question of Dr. Amr Muhammady below.
sure
will double ckeck
Thanks
Dr. Moustafa
You are most welcome.
Thanks Dr Shokier.
This means that CI should be used only with large sample size ( e.g. Thousands like in cross sectional studies) .
Dr. Moustafa
Exactly Yes. This is correct regarding presentation of normally distributed continuous variable, but there are other uses of CI that you will know in the upcoming lectures.
Thanks Prof Dr shokier
Dr. Moustafa
You are most welcome.
Prof. Dr. Shokeir
Regarding normally distributed continuous data, you mentioned 2/3 of men are expected to weigh 1 SD & 95% of men are expected to weigh 2 SD.
Yet you mentioned 2/3 of men is expected to be 1 SEM & 95% of men is expected to be 2 SEM
I just can’t process this, although I’ve read it over and over
Dr. Amr
There are 2 ways to present the normally distributed continuous data:
1- The first is by the use of mean ± SD. Under such situation, the values will be like that:
– 2/3 are expected to have values one SD below and above the mean.
– 95% are expected to have values 2 SD below and above the mean.
Please note that SD deals with true values and not with the means of the values.
2- The second is by the use of SEM and 95% of confidence interval (CI).
The CI is the range of values that fall within two SEMs above and below the mean, i.e. mean ±2 SEM.
Example:
– If a mean weight of a sample of 100 men is 75kg, and the SD is 10kg. The CI will SD/√n= 10/√10=1. So we can say the mean weight is 75kg (95% CI 73 to 77 kg).
– This means that if random samples of 100 men were repeatedly drawn from the same population, approximately 95% of these samples would be expected to have mean values between 73 and 77 kg.
Please note that SEM deals with means and not with true values.
It is only one study with only 100 patients with only one mean.
For calculation of the CI, the computer will generate a program to imagine that if random samples of 100 men were repeatedly drawn from the same population, we will be 95% confident the range of the mean values will fall between 73 and 77 kg.
Again, it is only one study with 100 patients with a mean of 75 kg, but there are 2 ways of presentation of the data of the same study:
– One is by the use of SD which deals with the true values.
– The other is by the use of SEM and CI which deals with the means of the values in an imaginary way generated by the computer.
In the usual studies containing up to hundreds of cases we use SD, but in cross-sectional studies containing thousands of cases we use CI.
Dr. Amr
Please let me know if the answer is clear or not.
Absolutely clear
Thank you, Prof. Dr. Shokeir
Thank you very much. Please keep always in touch through asking questions and writing comments to know your impression and also the impression of all other participants.
Prof. Dr. Shokeir,
thanks for rich lecture
l’ve some queries …
Dr. Amr
Thanks for rich lecture.
You are most welcome.
In right-skewed data, can we say that most values are clustered around the left side of the distribution while right of distribution is longer? And if so! What is the significance of that?Your statement is not correct. Suppose that we have a study including 100 patients, 95 patients have a weight between 70 to 90 kg and only 5 patients have a very high weight between 250 to 300 kg. Under such situation the mean value of the 100 patients will be high because the mean is the sum of weight of the 100 cases divided by 100. But the median will not be affected because the 5 cases with very high values will be calculated as only 5 cases above the median. Then, the mean will be bigger than the median and it will be located on the right side of the median on the curve and so called right skewed data.
The significance of that is to know the distribution, whether symmetric or normally distributed or asymmetric or non-normally distributed. Symmetric or normally distributed data are presented as mean +/- SD and we use for them parametric statistics, while asymmetric or non-normally distributed data are presented as median and range and we use for them non-parametric statistics..
These information will be more clarified in the upcoming lectures with very clear examples.
Are we the ones who determine whether a study to be normally distributed or free-distributed, or the parameters?The author will check the variables by asking the computer to calculate the mean +/- SD. If double SD is less than the mean, then the data are normally distributed or symmetric and we can use parametric statistics. On the other hand, if double SD is bigger than the mean, then the data are asymmetric and we have to use non-parametric statistics. This issue will be more clarified in the upcoming lectures.
This is a rough way to estimate the distribution of data that you can easily do by your self once you look at the mean and SD in any table while you are reading any article. Very frequent mistakes are done by the authors during presenting their data and this will result in using wrong statistics and wrong conclusion of the their studies.
There is a special program to check normality of distribution of data and will be explained latter on by Professor Ahmed Harraz.
In dispersion of distribution in example II, change of SD from 0.3 to 0.4. You mentioned that would change the shape of the curve without its position, what is the significance of that?It has no practical significance, just to know how change in the mean and the SD will affect the distribution of data. It is just a theoretical information of no practical value. Just to know.
Is the center of distribution = center of gravity?Yes, it is the mean.
What is the significance of SEM in a study?SEM is used in cross-sectional studies with thousands of cases to assure precision of the estimated mean of the population. It deals with multiple means, while SD deals with true values. in our usual studies containing hundreds of cases, we use SD not SEM.
This information will be more emphasized in the upcoming lectures.
Thank you Dr. Amr for your questions. It will give more chance for other participants to clarify the information. Next time, please ask the questions one by one to make it easy for me to reply.
Prof. Dr. Shokeir,
I’m so grateful and you really know how to bring out the best in us.
My apologies for bringing questions together
Dr. Amr
You are most welcome. I am happy by your discussion.
No worries.
greeting professor A shokair
its an honar to be with you in such an important course.
i need more details about confidence interval as i cannot understand its function and value,,, many thanks.
DR. Yasser
Thank you very much for the nice words.
More details about confidence interval will be given in the upcoming lectures.
Please do not worry. You will be completely satisfied through giving many examples. Just hear the next lectures and do the exercises and answer the MCQs.
I am ready to receive any quires.
Advice from the course director:
Before you answer the questions, please hear the slides several times ,one by one.
I am waiting comments from the participants.
Dear Participants
Please ask your questions one by one to make it easy for me to reply. Thank you very much.
Thank you doctor for this rich informative lectuer,
could you please expalin the interquartile rang and is there an equation to cululatie the median or it is just the mid point of any location
Dr .Basma
– Yes , the median is the midpoint of any location (50th percentile)
– The calculation of the median alone is not sufficient , we have to determine two other points , this could be :
1. Minimum – maximum.
2. The interquartile range (25th and 75th percentile of distribution)
3. The 10th and 90th percentile.
– For Example, in a sample of 100 patients if the median weight is 75 kg and the interquartile range is 62-88 kg (the example is virtual and the values are virtual).
– This means that the midpoint of distribution (50th percentile) is at 75 kg at the point of 62kg, 25% of the 100 patients will have weight less than 62 kg and 75% of the patients will have weight more than 62 kg.
At the point of 88kg, 75% of patients will have a weight less than 88 kg and 25% will have a weight more than 88kg.
All the calculations are done by computer. You just request the test and the answer will come automatically. I am just explaining the meaning of the statistical expressions you are dealing with. This basic information will help you during your entire scientific career.
Dr .Basma
Please let me know if the answer is clear or not.
yes thank you Dr Ahmed
Dr. Basma
You are most welcome. Please go to the next lectures and let me know your impression.