sampling technique-ppt -3 powerpoint presentation

screening techniques used

1. SAMPLING TECHNIQUES BY: Dr VANI H C  GUIDE: Dr LALITHA K   6/18/2013 1 SAMPLING TECHNIQUES

2. CONTENTS Introduction Definitions Need for sampling Major requirements for a sample Generalisation/ External validity Sample size & precision Reliable sample 6/18/2013 2 SAMPLING TECHNIQUES

3. CONTENTS Sampling techniques Sampling errors Advantages & limitations of sampling Conclusion References 6/18/2013 3 SAMPLING TECHNIQUES

4. Introduction A major reason for having an insight into the science of epidemiology & research methodology is that we always study a ‘ sample’ Concerned with the selection of representative sample, especially for the purposes of statistical inference. Idea of sampling is very old & from time immemorial, people have used it in day-to-day life. For example: 6/18/2013 4 SAMPLING TECHNIQUES

5. Introduction On the basis of a sample study, we can predict & generalise the behaviour of the population. Most researchers come to a conclusion of their study by studying a small sample from the huge population or universe. Census VS sampling 6/18/2013 5 SAMPLING TECHNIQUES

6. DEFINITIONS Population- aggregate of units of observations either animate or inanimate about which certain information is required. Sample- word used to describe a portion chosen from the population For sampling purpose, the population has to be divided into smaller units - sampling unit URL:http ://www.google.co.in/images?rls=ig&hl=en&source=imghp&biw=1024&bih=651&q=population+and+sample&gbv=2&aq=4&aqi=g1&aql=&oq=population+and+sam&gs_rfai= 6/18/2013 6 SAMPLING TECHNIQUES

7. DEFINITIONS Sample size- number of units in a sample Sampling frame - list of each and every individual in the population Variable: any quality or quantity liable to show variation from one individual to the next in the same population Variate: individual observations of any variable 6/18/2013 7 SAMPLING TECHNIQUES

8. DEFINITIONS Statistic / datum - measured or counted fact or piece of information stated as a figure Statistics/ data - field of study concerned with techniques of collection of data, classification, summarising, interpretation, drawing inferences, testing of hypothesis, making recommendations etc when only a part of data is used Biostatistics- when tools of statistics are applied to the data that is derived from biological sciences such as medicine. 6/18/2013 8 SAMPLING TECHNIQUES

9. DEFINITIONS Distinction Population Sample Definition Collection of items under consideration. Part of the population selected for study. Characteristics Parameter Statistics Symbols N= population µ = population mean σ = population standard deviation π = population percentage n = sample size x = sample mean s = sample standard deviation p = sample percentage 6/18/2013 9 SAMPLING TECHNIQUES

10. Need for sampling Complete enumerations are practically impossible when the population is infinite. When the results are required in a short time. When the area of survey is wide. When resources for survey are limited particularly in respect of money and trained persons. 6/18/2013 10 SAMPLING TECHNIQUES

11. Major requirements for a sample To draw conclusions about population from sample, there are two major requirements for a sample. Sample has to be selected appropriately so that it is representative of the population. It should have all the characteristics of the population. The sample size should be adequately large 6/18/2013 11 SAMPLING TECHNIQUES

12. Sampling Terminology 6/18/2013 12 SAMPLING TECHNIQUES

13. Sampling & representativeness 6/18/2013 13 SAMPLING TECHNIQUES

14. How Do We Generalize? Population Sample draw sample draw sample generalize back generalize back PROBLEMS OF GENERALISATION 6/18/2013 14 SAMPLING TECHNIQUES

15. Calculating the sample size : depends upon precision which in turn depends upon significance level & allowable error. Depends upon the kind of data: Problems with very large or small sample size Qualitative data n = 4pq/ L 2 Quantitative data n= 4 σ 2 L 2 6/18/2013 15 SAMPLING TECHNIQUES

16. Reliable sample There are 8 basic requisites for a reliable sample: 1. Efficiency 2. Representativeness 3. Measurability 4. Size 5. Coverage 6. Goal orientation 7. Feasibility 8. Economy and cost efficiency 6/18/2013 SAMPLING TECHNIQUES 16 SOURCE:World Health Organization.Health Research Methodology – a guide for training in research methods.First edition.New Delhi:Oxford university press;1993.p.77-94.

17. Types of Sampling Methods Cluster Sampling Non-Probability Sampling Convenience Probability Sampling Simple Random Systematic Stratified Purposive 6/18/2013 17 SAMPLING TECHNIQUES

18. Simple Random Sampling (SRS) Here each unit in the population has equal chance or probability to be selected in the sample. Two types: SRS with replacement & SRS without replacement. Situations where random sampling can be done: Sampling frame is available. When the population is small. Parameters to be estimated -homogeneously distributed in population. Units should be readily available- ex: patients in wards 6/18/2013 18 SAMPLING TECHNIQUES

19. Simple Random Sampling (SRS) The procedure involved in Random Sampling: Preparing a sampling frame Deciding the size of the sample to be chosen. To select the required number of units at random Random samples can be drawn by: lottery method - random number tables- using calculators or computers- 6/18/2013 19 SAMPLING TECHNIQUES

20. RANDOM NUMBER TABLE Standard tables Steps to use table SOURCE: Training for mid-level managers. 7. The EPI coverage survey. [Serial online] 2008 [Cited 2013 June 6] Available from URL: http://whqlibdoc.who.int/hq/2008/WHO_IVB_08.07_eng.pdf 6/18/2013 20 SAMPLING TECHNIQUES

21. 1 2 3 4 5 6 7 8 9 10 Simple Random Sampling ex: 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 N =30 n = 10 6/18/2013 21 SAMPLING TECHNIQUES

22. Merits of using random numbers: 1. Personal bias is eliminated 2. It is in general a representative sample for a homogenous population. 3. There is no need for the thorough knowledge of the units of the population. 4. The accuracy of a sample can be tested by examining another sample from the same universe when the universe is unknown. 5. This method is also used in other methods of sampling. 6/18/2013 22 SAMPLING TECHNIQUES

23. Limitations of SRS: 1. Preparing lots or using random number tables is tedious when the population is large. 2. When there is large difference between the units of population, the simple random sampling may not be a representative sample. 3. The size of the sample required under this method is more than that required by stratified random sampling. 4. It is generally seen that the units of a simple random sample lie apart geographically. The cost and time of collection of data are more. 6/18/2013 23 SAMPLING TECHNIQUES

24. SYSTEMATIC RANDOM SAMPLING Commonly employed technique, when complete and up to date list of sampling units is available. Procedure: 1.Prepare the list of population (sampling units) 1 to N. 2. Decide on the n (sample size) that you want or need. 3. Calculate sampling fraction/ sampling interval (k) k= N/n where N = population size & n = sample size 4. Randomly select an integer between 1 to k th . 5. Add to this the sampling interval to get required sample. Then take every k th unit. 6/18/2013 24 SAMPLING TECHNIQUES

25. SYSTEMATIC RANDOM SAMPLING Ex: in PPI 15 out of 150 houses have to be selected 6/18/2013 25 SAMPLING TECHNIQUES

26. SYSTEMATIC RANDOM SAMPLING Merits : Simple and convenient to adopt. Time and labour involved in the collection of sample is relatively small. If the population is sufficiently large, homogenous & each unit is numbered, this method can yield accurate results. Limitations: The sample may exhibit a pattern or periodicity Systematic sampling may not represent the whole population. There is a chance of personal bias of the investigators. 6/18/2013 26 SAMPLING TECHNIQUES

27. STRATIFIED RANDOM SAMPLING Preferred when the population is heterogeneous with respect to characteristic under study. The complete population is divided into homogenous sub groups - ‘ Strata’ & then a stratified sample is obtained by independently selecting a separate simple random sample from each population stratum. Gives equal chance to the units in each stratum to be selected as sample. The total sample is the addition of samples of each stratum 6/18/2013 27 SAMPLING TECHNIQUES

28. Stratified random sampling: Ex: STAFF PG OTHER 6/18/2013 28 SAMPLING TECHNIQUES

29. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 10 2 8 4 1 6 12 8 4 2 1 6 Non-Proportional stratified sampling 6/18/2013 29 SAMPLING TECHNIQUES

30. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 10 2 8 4 1 6 12 8 4 2 1 6 Proportional stratified sampling 6/18/2013 30 SAMPLING TECHNIQUES

31. STRATIFIED RANDOM SAMPLING Merits: 1. It is more representative. 2. It ensures greater accuracy 3. It is easy to administer as the universe is sub - divided. 4. Greater geographical concentration reduces time and expenses. 5. When the original population is badly skewed, this method is appropriate. 6. For non – homogeneous population, it may yield good results. 6/18/2013 31 SAMPLING TECHNIQUES

32. STRATIFIED RANDOM SAMPLING Limitations: 1. To divide the population into homogeneous strata, it requires more money, time and statistical experience which are a difficult one. 2. Improper stratification leads to bias, if the different strata overlap such a sample will not be a representative one 6/18/2013 32 SAMPLING TECHNIQUES

33. LOT QUALITY ASSURANCE SAMPLING Originated in manufacturing industry for quality control purposes Only outcome - “acceptable” or “not acceptable” Two types of risk ( i ) Risk of accepting a “bad” lot - Type I Error, (ii) Risk of not accepting a “good” lot - Type II Error The advantage over a traditional stratified sampling design: the response for each lot is binary (acceptable or not) & therefore smaller sample sizes can be used 6/18/2013 33 SAMPLING TECHNIQUES

34. CLUSTER SAMPLING Used when the population is heterogeneous & when sampling frame is not available at individual level Clusters are formed by grouping units on the basis of their geographical locations. Obtained by selecting clusters from population on the basis of SRS From the selected clusters each and every unit is included for study Special form of cluster sampling - “ 30 cluster sampling ” for field studies in assessing vaccination coverage 6/18/2013 34 SAMPLING TECHNIQUES

35. CLUSTER SAMPLING Section 4 Section 5 Section 3 Section 2 Section 1 6/18/2013 35 SAMPLING TECHNIQUES

36. 6/18/2013 36 SAMPLING TECHNIQUES

37. The immunization coverage in the target area to be evaluated (coastal region of a hypothetical country) All cities, towns and villages of the coastal region have been listed. The cumulative population given. Calculate sampling interval & identify 1-5 clusters given the random number= 12,762 6/18/2013 37 SAMPLING TECHNIQUES

38. Total population to be surveyed / 30 clusters = Sampling interval 800000/30 = 26666.66= Sampling interval random number= 12,762 1st cluster= cumulative population of 12,762 (village 1) 2nd cluster = 12,762+26666.66= 39428.66 (village 9) Next cluster = Number which identified the location of the previous cluster + Sampling interval 3rd cluster= 39428.66+26666.66= 66095.32 (village 14) 6/18/2013 38 SAMPLING TECHNIQUES

39. CLUSTER SAMPLING Design effect - loss of variation in a sample that occurs as a consequence of using cluster sampling, as opposed to any other probability method Advantages : - Only need to obtain list of units in the selected clusters. - Cost-effective. Disadvantages : - Not intended for calculation of estimates from individual clusters. - Less precise than simple random sample. 6/18/2013 39 SAMPLING TECHNIQUES

40. MULTISTAGE SAMPLING Sampling procedures carries out in several stages using random sampling techniques. When the sampling frame is rarely available, or if such a list is available, it may be too large and unwieldy to use. To overcome such a problem, multi-stage sampling procedures are often employed. Each point of sampling is called a “stage” and the term “multi-stage sampling procedure” is generally used to refer to a sample selection process that has at least two stages. Any of the probability sampling techniques may be used at each stage of a multi-stage procedure STAGE 1 6/18/2013 40 SAMPLING TECHNIQUES

41. Example for multistage sampling Sample size = 3000 41 STAGE 1 STAGE 2

42. MULTIPHASE SAMPLING Part of information is collected from the whole sample & part from the sub sample. Number in the sub samples in 2 nd & 3 rd phases will become successively smaller & smaller. Survey by such methods will be less costly, less laborious & more purposeful. 6/18/2013 42 SAMPLING TECHNIQUES

43. MULTIPHASE SAMPLING Ex: In a tuberculosis survey First phase- physical examination or mantoux test done in all cases of the sample Second phase-x-ray of the chest done in mantoux positive cases & in those with clinical symptoms Third phase -sputum may be examined in X-ray positive cases 6/18/2013 43 SAMPLING TECHNIQUES

44. TABLE: COMPARITIVE PERFORMANCE OF VARIOUS RANDOM SAMPLING METHODS Method of random sampling Desired size of target population Reliability of conclusions for fixed sample size Economy Remarks Simple Small Very good Expensive Requires full sampling frame Systematic Small Good Economical sampling frame not needed but the size of the target population is needed Stratified Medium Good Expensive Good for non-homogenous population Cluster Large Poor Very economical Very convenient for geographically diverse population Multi stage Very large Medium economical Requires sampling frame only for each nested unit SOURCE: Indrayan A, Satyanarayana L. Simple biostatistics. `3 rd ed. Academia publishers: 2009; Delhi 6/18/2013 44 SAMPLING TECHNIQUES

45. NON RANDOM SAMPLING The sampling is purposive when cases that serve specific purpose are chosen. Results based on non-random samples are not generalizable yet are useful in some situations in providing a clue about the status of a phenomenon. Non-Probability Sampling Convenient sampling Snowball sampling Convenient groups Purposive sampling Volunteers for phase 1 of a trial Delphi method Pilot study & pre-testing 6/18/2013 45 SAMPLING TECHNIQUES

46. Snowball sampling: A few identified members of a rare population are asked to identify other members of the population, those so identified are asked to identify others Hard-to-reach, or equivalently hidden populations. Constructing sampling frame using methods such as household surveys is infeasible when the population is small relative to the general population geographically dispersed when population membership involves stigma group has networks that are difficult for outsiders to penetrate Ex: people exposed to sex workers or those injecting drugs in the context of HIV. 6/18/2013 46 SAMPLING TECHNIQUES

47. Convenient groups : many studies are done on medical students just because they are available in captivity & would generally provide correct response. Ex: Famous Doll & Hill study on smoking and lung cancer done on physicians 6/18/2013 47 SAMPLING TECHNIQUES

48. Purposive sampling Volunteers for phase 1 trial : generally done on volunteers to provide useful information regarding the safety and side effects of a treatment regimen Delphi method : the least expensive method to generate data is to ask colleagues as to what do they think about a particular problem. They will narrate their experience which will vary from person to person Pilot study & pre-testing : small scale study as a forerunner 6/18/2013 48 SAMPLING TECHNIQUES

49. Differences b/w Non-probability samples & probability samples 6/18/2013 SAMPLING TECHNIQUES 49 URL:http://www.chsbs.cmich.edu/fattah/courses/empirical/22.html Key terms Non-probability samples Probability samples Sampling frame Does not exist or inaccurate Accurate and up-to-date Sampling error Cannot be calculated Can be calculated Sample size Matter of convenience Determined by sampling theory Level of generalizability Illustrative Representative.

50. Errors in sampling Sampling errors - definition -how is it measured -how can it be reduced Non sampling errors Coverage errors Observational errors Processing errors 6/18/2013 50 SAMPLING TECHNIQUES

51. Advantages of Sampling There are many advantages of sampling methods over census method. They are: 1. Saves time and labour. 2. Results in reduction of cost in terms of money and man-hour. 3. Ends up with greater accuracy of results. 4. Has greater scope. 5. Has greater adaptability. 6. If the population is too large, or hypothetical sampling is the only method to be used. 6/18/2013 51 SAMPLING TECHNIQUES

52. Limitation of Sampling Sampling is to be done by qualified and experienced persons. Otherwise, the information will be unbelievable. Sample method may sometimes give the extreme values There is the possibility of sampling errors. Census survey is free from sampling error 6/18/2013 52 SAMPLING TECHNIQUES

53. CONCLUSION Whenever a scientific study is planned it may not always be feasible to study the entire population. In such situations we need to apply some sampling technique to select our samples and it’s better to select probability sampling techniques. Selecting a sampling method depends upon: Population to be studied ( size & heterogeneity with respect to variables) Level of precision required Resources available Importance of having a precise estimate of the sampling error 6/18/2013 53 SAMPLING TECHNIQUES

54. REFERENCES Bhalwar R, Vaidya R, Tilak R, Gupta R, Kunte R. Text book of public health and community medicine. Pune : Department of Community Medicine Armed Forces Medical College; 2009 Rao NSN, Murthy NS. Applied statistics in health sciences. 2 nd ed. New Delhi: Jaypee ; 2010. Mahajan BK. Methods in biostatistics. 6 th ed. New Delhi: Jaypee ; 2006 Abramson JH. Survey method in community medicine. Edinburg: Churchill Livingstone: 1974 6/18/2013 54 SAMPLING TECHNIQUES

55. REFERENCES Indrayan A, Satyanarayana L. biostatistics for medical, nursing and pharmacy students. New Delhi: Prentice-Hall of India; 2006 Varalakshmi V, Suseela N, Sundaram TG, Ezhilarasi S & Indrani B. Statistics. HIGHER SECONDARY – FIRST YEAR. Chennai: TAMILNADU TEXTBOOK CORPORATION; 2005 Training for mid-level managers. 7. The EPI coverage survey. [Serial online] 2008 [Cited 2013 June 6] Available from URL: http://whqlibdoc.who.int/hq/2008/WHO_IVB_08.07_eng.pdf 6/18/2013 55 SAMPLING TECHNIQUES

56. REFERENCES Woodard SH. Description and comparison of the methods of cluster sampling and lot quality assurance sampling to assess immunization coverage. [Serial online] 200 [Cited 2013 June 6] Available from URL: http://www.who.int/vaccines-documents/DocsPDF01/www592.pdf Heckathorn DD Snowball versus respondent driven sampling. [Serial online] 2005 [Cited 2013 June 6] Available from URL:http:// www.ncbi.nlm.nih.gov / pmc /articles/PMC3250988/ pdf /nihms319788.pdf 6/18/2013 56 SAMPLING TECHNIQUES

57. REFERENCES Handcock MS, Gile KJ. On the concept of snowball sampling. [Serial online] 2011 [Cited 2013 June 6] Available from URL:http:// arxiv.org / pdf /1108.0301.pdf Sethi D, Habibula S, McGee K, Peden M, Bennett S, Hyder AA, Klevens J, Odero W, Suriyawongpaisal P. Guidelines on conducting community surveys on injuries and violence. [Serial online] 2004 [Cited 2011 Aug 21]. Available from URL: http://www.bvsde.paho.org/bvsacd/cd20/conducting.pdf. Accessed on 21/08/11 Seminar notes on sampling techniques by Dr Gautam S. Seminar notes on sampling techniques by Dr Chetana T 6/18/2013 57 SAMPLING TECHNIQUES

58. Thank you 6/18/2013 58 SAMPLING TECHNIQUES