No, Professor.

You can not claim -statistically- that "people with diabetes" is the population. The patiens sampled were exclusively from "diabetic patients [that] are admitted in the medical department of the hospital" (and I further assume that Hamna was talking about a particular hospital). Since the sampling was done within a particular time window, the statistical population is actually the group of diabetic patients that are admitted in the medical department of this hospital within this period of time.

Diabetic patient that were not admitted to the hospital, or admitted to other hospitals, or admitted years before or after the study had zero probability to be sampled.

The pure statistical measures therefore are related only to this specific population.

Surely a generalization beyond this population may be possible, but the level of confidence of such generalizations can not be infered by the data, only by common sense and substantive argumentation. One has to consider knowledge about the bigger picture, information than can not be extracted from the data.

Example: If the particular hospital and its catchment area is "typical" for a greater class of hospitals (at least w.r.t. all relevant parameters regarding diabetes), then it is sensible to generalize the findings to "such hospitals" (and their patients). eneralizations can become very inaccurate when this particular hospital (and its ctchment area) was some kind of a "special case". Also generalizations over larger periods of time are always difficult, because the whole environment changes over time, as does the care, therapy and drugs...

You should use the N that corresponds to the size of the source from which you draw your sample, so that sounds like it would indeed be the number of total patients visiting the medical department in the specific time of conducting research

I agree with the answer of Prof. David Morgan.

Your population is people with diabetus. Thus, you should limit your sample to them only and study their QoL. If you need help in calcualting the sample size, send me your research quesiton or hypothesis at: [email protected]

good luck

No, Professor.

You can not claim -statistically- that "people with diabetes" is the population. The patiens sampled were exclusively from "diabetic patients [that] are admitted in the medical department of the hospital" (and I further assume that Hamna was talking about a particular hospital). Since the sampling was done within a particular time window, the statistical population is actually the group of diabetic patients that are admitted in the medical department of this hospital within this period of time.

Diabetic patient that were not admitted to the hospital, or admitted to other hospitals, or admitted years before or after the study had zero probability to be sampled.

The pure statistical measures therefore are related only to this specific population.

Surely a generalization beyond this population may be possible, but the level of confidence of such generalizations can not be infered by the data, only by common sense and substantive argumentation. One has to consider knowledge about the bigger picture, information than can not be extracted from the data.

Example: If the particular hospital and its catchment area is "typical" for a greater class of hospitals (at least w.r.t. all relevant parameters regarding diabetes), then it is sensible to generalize the findings to "such hospitals" (and their patients). eneralizations can become very inaccurate when this particular hospital (and its ctchment area) was some kind of a "special case". Also generalizations over larger periods of time are always difficult, because the whole environment changes over time, as does the care, therapy and drugs...

The sample design is not limited to the size of the sample, also includes the sampling strategy. There are 2 issues that matter. First of all diabetics are supported by this department so theoretically all diabetics (that are supported in a window of time to be determined) will be at this point. Where people who participate in the study at the Department or the entire hospital selected?

Hamna, the population of size N must be that to which we like to infer from the sample of size n. As you need to infer about the "diabetic patients of your hospital", this is the finite population in an instant or period of time. And the number N is obtained from your list of patients with diabetes.

The number of sampled patients can be obtained from specifications of precision of estimators, or required confidence level. But, in practice, it is better to be realist "the number of beds for the sample is the number of available beds", say n in the simple random sample without replacement. From these data: N, n, and the sample observations you can give unbiased estimation of the interest parametric function, and unbiased estimation of the variance of such preceding estimation.

The number of beds for admitting patients must be taken by the needs of the patients, never from scientific considerations. The dignity of the patients is always over the possible advancement of science (i. e. quality of life knowledgement).

Hamna, I honestly believe that the decision on who your population is, and t what the size of your population is should have been done prior to the determination of the sample size and selection of the sample elements ; and the collection of data. Eddie

As per your statement you are conducting a research on diabetic people's quality of life. I presume for this you are doing a case control type study? The quality of life of diabetic patients cannot be assessed without considering normal healthy controls. So my question is apart from your sample size consideration, have you considered any controls?  Your sample size consideration should be based on the characteristics of diabetic patients which you want to study? Say you are interested in studying the obesity and its relationship with diabetics? Another problem may be frequency of morbidity which occurs among them as compared to normal healthy subjects? The third problem may be related to their food habits whether it is different as compared to that seen in  controls? So my suggestion is that your sample size consideration should be based on the characteristics which you are going to study. Once your design of study is finalized, the sample size can be calculated easily.  

Hello, Hamma.  I think most people who have answered your question are speculating and actually making things more confusing for you.  So far, the answer given by Dr, David L. Morgan is the most accurate and parsimonious.

Rather confusing but you should have done sample size estimation prior to collection of data. It doesn't mean your data is useless but you can still analyse your data and report it.

Just as says colleague Sarangapany, an operational  hypothesis-based sample size calculation should have lead to a proper sample size meeting the design –and hypothesis- requirements before the data collection.  Nevertheless, you can analyze your data and attempt to publishing the results somewhere.

Yet, the real problem is that you are facing several classical sources of bias, given that your study population is restricted to that from just one hospital. The list of potential biases  is too long so I would suggest you read Sackett DL.”Bias in analytical research”.  J. Chron. Dis. 1972; 32:51-63 an old, yet easy as well as straightforward text to start with the subject.

Best wishes,

Yes N would be the medical department's patients and specifically the total no. of patients visited during your research period let say it 2011 to 2013.

Since no previous hypothesis seem to have existed, the study is a descriptive one  (description of some traits of your community of diabetic patients in your particular hospital), and since you already have such diabetic group I would advise you to forget a calculated sample size. Just proceed on your analysis whose resulting figures will have their proper 95% confidence limits, this way bringing good information about theirs precision, although any aim of their  generalization  will be absolutely hampered by being a non representative sample (regarding the nature of the sample).

The problem with accepting a smaller sample size and the standard errors that it generates is that those standard errors will be large -- in other words, the tests will lack "power" to find statistically significant results. 

your N will be the total number of units in your target population, at that time of your survey, within your desired frame(  i think your units here are diabetic patients and with it as your target population within that particular hospital, admitted within your desired period of time).

There are two issues here:

Also, in response to other comments in this thread, it's very true that many types of biases and potential errors exist from both a data gathering and data analysis point of view, and it takes a long time to learn about them. Hence, it's important to collaborate with knowledgeable colleagues. I think the best advice we can give is not about individual issues (although this is helpful), but rather a broader approach to learning about conducting research. Asking questions on the Internet is OK, but close in-person collaboration is far superior.

Good luck with your research,

John

I fully agree with Dr. John R. Richard., although it is true that sometimes the issue may be quite simple and easy as to to support young colleagues with our standpoints. 

If the population is well defined With N units, and you sample at random from it (due to every limitations stated above) You can correct the standard error by SQRT((N-n)/(N-1)). This is  Close to 1 (no correction) if N>>n 

Dear All

In order to answer this question/problem, several remarks have to be studied.

1. General remarks:

Research studies are usually carried out on sample of subjects rather than whole populations. The most challenging aspect of fieldwork is drawing a random sample from the target population to which the results of the study would be generalized. The key to a good sample is that it has to be typical of the population from which it is drawn. When the information from a sample is not typical of that in the population in a systematic way, we say that error has occurred. In actual practice, the task is so difficult that several types of errors, i.e. sampling error, non-sampling error, Response error, Processing error,…

In addition, the most important error is the Sampling error, which is statistically defined as the error caused by observing a sample instead of the whole population. The underlying principle that must be followed if we are to have any hope of making inferences from a sample to a population is that the sample be representative of that population. A key way of achieving this is through the use of “randomization”. There several types of random samples, Some of which are: Simple Random Sampling, Stratified Random Sampling, Double-stage Random Sampling... Moreover, the most important sample is the simple random sample which is a sample selected in such a way that every possible sample of the same size is equally likely to be chosen. In order to reduce the sampling error, the simple random sample technique and a large sample size have to be developed.

2. Specific remarks:

The following factors are highly affected the sample size and need to be identified:

• Population Size,

• Margin of Error,

• Confidence Level (level of significance) and

• Standard of Deviation.

Then, the sample size can be estimated by,

Necessary Sample Size = (z-score or t-value)2 * StdDev*(1-StdDev) / (margin of error)2 .

am in agreement with prof: Dr, David L. Morgan