Chapter 6 SAMPLING DESIGNS Whether a research deSign is descriptive or experimen- tal, Sempling is necessary, especially if the population of the study is too large where the 4 M’s(Man, Money, Material, and Machinery) resources of the investigator are limited. And it is advantageous for him to use sample survey rather than the total population However, the use of the total population is advisable if the number of subjects under study is less than 100. But if the total population is equal to 100 or more, it is advisable to get a sample in order to be effective, efficient and economical in gathering data, provided however, that the sample is a representative cross-section of the population and is scientifi- cally selected. Sampling may be defined as the method of getting a representative portion of a population. The term, population, is the aggregate or total of objects, persons, families, species or orders of plants or of animals Several studies have been conducted using a representa- tive sample of a population based on the theory that “a value caleulated from a sample distribution is called statistic and the corresponding value from a population distribution is called a parameter.” (Parel et al., 1978). The symbol used for a statistic is in Roman letters, e.., X for ‘arithmetic mean; for parameter, Greek letters, e:g., u for population mean. SU is used to refer to the standard deviation or astatistic (sample distribution), and o for parameter (popu- lation distribution), Any other statistics calculated on a sample is an estimate of a corresponding populatian parameter. However, in “most 92situations ‘SAMPLING DESIGNS fol tea “the parameters ‘are unknown and must be esti- mated in Same mannér from the sample data.” (Ferguson and Takane, 1989). On the other hand, “the body of statistical methods con- “cerned with the making of statements about population ‘pa- rameters from sample statistics is called sampling statistics.” (Ferguson and Takané, 1989). Hence, the data gathered in the study from a sample distribution need to be converted into sampling statistics for better analysis and interpretation of results. Advantages of Sampling ‘The following are the advantages of sampling, 4, It saves time, money and effort. The research save time, money and effort because the number of subjects involved is small. There are only 2 small number to be collected, tabulated, presented, analyzed and interpreted, but the use of sample gives a comprehensive information of the results of the study. It is more effective. Sampling is more effective if every individual of the population without bias has an equal chance of being included in the sample and data are scientifically collected, analyzed and inter- preted. It is faster and cheaper. Since sample is only a“drop in a bucket,” the collection, tabulation, presenta- tion, analysis and interpretation of data are rapid and less expensive because of the small number of subjects. It is more accurate. Fewer errors are made due to the small size of data involved in collection, tabu- lation, presentation, analysis and interpretation Itgives more comprehensive information. Since there is a thorough investigation of the study due to a small sample, the results give more comprehensive information because all members of the population34 METHODS OF RESEARCH AND THESIS WRITING have an equal chance of being included in the sam- ple. Limitations of Sampling If sampling design has its strength, it also has its weak- nesses. The following are the disadvantages of sampling de- sign: vr w o Sample data involve more care in preparing de- tailed subclassifications because of a small number of subjects. If the sampling plan is not correctly designed and followed, the results may be misleading. Sampling requires an expert to conduct the study in an area. If this is lacking, the results could be erroneous. ‘The characteristic to be observed may occur rarely in a population, e.g., teachers over 30 years of teach- ing experience. Complicated sampling plans are laborious to pre- pare. Planning a Sample Survey In planning a sampling survey, the steps are as follows: 1. 2 3. 5. 6. State the objectives of the survey; Define the population; Select the sampling individual; Locate and select the source list of particular indi- viduals to be included in the sample; “Decide the sampling design to be used: Determine the sample size by using the formula, ENV + [Se%(1'= p)} ’s° NSe + (Vp = py NSe + [V*p(1 — p)] tion, 8, is the sample aes) Vis the standard ¥ where N is the popula- ieee:: SAMPLING DESIGNS reliability level; and p, the largest possible propor- tion (0.50); sue 7. Sclect the method in estimating the reliability of the sample either test-retest, split-half, parallel-forms or internal consistency; . 8. Test the reliability of the sample in a pilot institu- tion; and 9, Interpret the reliability of the sample Determination of Sample Size Determining a sample’ size is basically 2 researcher's decision. Some investigators have no idea of determining the sample size in a given population scientifically, and arbitrarily select the majority criterion (50 per cent plus one) as sufficient for their study. Others would choose a proportion lesser or greater than the majority criterion or 51% of the population under study. But these ideas are not scientifically-oriented. As stated earlier, samplirig is advisable if the population is equal to or more than 100, but it is inapplicable to popu- lation less than 100. The use of the total population is advis- able when the population is less than 100 due to categoriza- tion, For instance, the study is on the problems met by science and mathematics instructors and professors. First, the sub- jects are categorized as a whole; second, they are categorized into science and mathematics instructors and professors; third, into qualified and nonqualified, and so on. It is desirable to have a bigger number in each categorization of samples to arrive at reliable results. . To have a scientific determination of sample size, the following formula is suggested: NV + (Se? (1 - pil (6.1) s “Ses 1" = ot = 2 z NSe + (V7 x pil - pil Sample size S. S, N° = Total number of population Y= The standard value (2.58) of 1 per cent level of probability with 0.99 reliability96 METHODS OF RESEARCH AND THESIS WRITING Se = Sampling error (0.01) p = The largest possible proportion (0.50) For an illustration cf the foregoing formula, the steps are as follows: Step 1. Determine the total population (N) as assumed subjects of the study. (Step 2. Get the value of V (2.58), Se (.01), and p (0.50). Step 3. Compute the sample size using the foregoing formula 6.1. For instance, the total population is 500, the standard value at 1 per cent level of probability is 2.58 with 99% re- liability and has a sampling error of 1% or 0.01 and the proportion of a target population is 50% or 0.50. Then, the sample size is computed as follows: Given: N = 500 V = 2.58 Se = 0.01 p = 0.50 s, = -NV+iSe? x =p) $s“ UNSe + (V)? x p(1 — p) 500(2.58) + (0.01)? x (1 ~ 0.50) 500(.01) + (2.58) x 0.50(1 — 0.50) _ _1290 + 0.0001 x 0.50 “5 + 6.6564 x 0.50(0.50) 1290 + 0.00005 6.6641. 09005, 6.6641 S,. = 193.57 or [194 _ The sample size for’ a, population of 500 is*194. This _sample (194) will represent the subjects of the dy. Table”SAMPLING DESIGNS 97 6.1 shows the computed sample sizes for different population (N) at 0.01 level of probability with 0.99 reliability to a pro- portion of 0.50. Table 6.1. Computed Sample Sizes for Different Population (N) at 0.01 Level of Probabil- ity to a Proportion of 0.50 ioe Sample Size _N Sample Size 100 — 97 300 — 166 25 m1 oo qm 100 122 350 175 i 132 375 — 179 200 — 148 425. 185 205 4 148, 42500 185 200 — 155 450 — 188 275 — 161 415 191 5254 196 900 218 550 =_ » 198 925 os 219 575 — 200 950 — 220 600 — 202 1000 — 221 62 — 204 1100 — 224 650 — 205 1500 — 232 675 — 207 1700 — 235 70. — 208 2000 — 238 995% 210 2500. 242 m0 — 2u1 3000 — 244 71 — 212 3500 — 244 116. = 212 4000 — 248 8250 215° . 4500. — 249 850 Comal 216 5000 aad 250. 875 217 The foregoing computed sample sizes in Table 6.1 appears that the smaller the population, the higher the percent- age of the sample size; the larger the population, the lower is the percentage of the sample size. In other words, popula- tion is inversely proportional to the percentage of the sample size, To prove, in a population of 100, the sample size is 97 To get the percentage, the formula is:98 METHODS OF RESEARCH AND THESIS WRITING . Os oe x 100 (6.2) where: {% = Percent S, = Sample size N' = Total number of population To substitute formula 6.2, consider the following compu- tation: , 8 " » 3 8 0.97 x 100 = 97% In other examples, the percentage of a sample size of 194 from a 500 population is 194/500 x 100 cquals 28.8%; 208 sample size from a 700 population is 208/700 x 100 equals 29.7%; a sample size of 221 from a 1000 population is 221/1000 x 100 equals 22.1%; a sample size of 238/2000 x 100 equals 11.9%; and so on. Hence, perceniage is nut upplicable in de termining sample size because population:is inversely propor- tional to the percentage of sample size. Sampling Designs By and large, there are two kinds of sampling, namely, (1) scientific sampling and (2) non-scientific sampling. In scientific sampling, each member in the population is given an equal chance of being included in the sample. In non- scientific sampling, not all of the members in the population are given an equal chance of being included in the sample. Scientific Sampling 1. Restricted random sampling. This sampling involves certain restrictions intended to improve the validity of the sample. However, this design is applicable only when the population being investigated is homogeneous. Restricted random samples drawn from a homogeneous population are likely to arrive at accurate values of the population characteristics.SAMPLING DESIGNS 99 Unrestricted random sampling. This is the best random sampling design because no restriction is imposed, and every member in the population has an equal chance of inclusion in the sample Restricted or unrestricted random sampling is chosen by using either of two techniques, namely, lottery or table of random numbers. In the former technique, each individual in the population is as- signed a number which is written on a piece of paper. For example, if the population is 100, there are 100 pieces of paper numbered 1 to 100. The pieces of papers are rolled and placed in a bottle, i.e., one liter coke bottle, and are mixed thoroughly. ‘The rolled pieces of paper are drawn from the bottle one at a time. Drawn numbers from 1 to 97 are chosen as samples because the sample size for a population of 100 is 97. This technique is useful when the population is small. ‘The table of random numbers consists of digits so selected that no systematic relation exists be- tween any sequence of digits in the table, regardless of whether the table is read vertically, horizontally, right or left, or in any other way. To draw a sample by using a table of random numbers, a technique must be used for assigning a number to each indi- vidual in the population. For instance, the popula- tion consists of 3,000 individuals. Each member must be assigned an identifying number ranging from 0001 to 3000. The identifying number may corré- spond to the total number of the population (3000). To choose a sample of 244 from a populatién of 3,000, four-digit numbers either vertically, horizon- tally, right or left, until 244 individuals are identi- fied The table of random numbers technique is ap- plicable to a large number of population Stratified random sampling. This design divides first the population into two or more strata. For each stratum, the sample items are drawn at rai100 METHODS OF RESEARCH AND THESIS WRITING For example, an institution may have 4,000:stu- dents, of which 1,000 are males and 3,000 are females. A sample of 221 males are drawn by a random technique from the subpopulation, or stra- tum, of males; and 244 females from the subpopulation, or stratum, of females. Systematic sompling. This is a design in which all individuals in the population are arranged in a methodical manier, i.c., alphabetical or chronologi- cal (age, experience, academic rank), and the nth name may be selected in the construction of the sample. For instance, there are 1,500 in a popula- tion. The 1,500 may be arranged in alphabetical or chronological order. They are numbered 1, 2, 3, 4, and 5 for the first set; 1, 2, 3, 4, & 5, for the second set; 1, 2,3, 4, & 5 for the third set; and so on until the desired population (1,500) is covered. Every Sth of the set is sclected a3 part of the sample until 232 is reached. Multistage sampling. This design is done in several stages. It can be two-stage, three-stage, four-stage, five-stage, ete., depending on the number of stages of sampling to be used. In this design, the popula- tion individuals are grouped into a hierarchy of units, and sampling is done consecutively. For in- stance, in a regionwide study, the provinces are chosen as the first stage; municipalities in each of the selected provinces constitute the second stage; barangays within the sample municipalities as the third stage; and sample rural ulks within the selected sample barangays as the fourth stage. Hence, the sampling is in four stages. Cluster sampling. In this design, the popu grouped into clusters or small units, for instance, blocks or districts, in a municipality or city, com- posed of population individuals and are selected either by random sampling or systematic sampling. This design is advantageous when indi uals in. _ the districts or blocks belong to heterogeneous group,SAMPLING DESIGNS = 301 Nonscientific Sampling In nonscientific sampling, subjectivity on the part of the researcher occurs because not all of the individuals in a popu- lation are given an equal chance of being included in. the sample. The nonscientific sampling designs are as follows: 1. Purposive sampling. This design is based on choos- ing individuals as samples according to the pur- poses of the researcher as his controls. An indi vidual is chosen as part of the sample because of good evidence that he is a representative of the total population. By and large, statisticians do not use this design because not all of the individuals in a population are given an equal chance to be included in the sample since the criteria are based on the purposes of the researcher. Incidental sampling. This design is “applied to tho: samples which are taken because they are the mo: able.” (Guilford and Fruchter, 1973). The re searcher simply takes the nearest individuals subjects of the study till the sample reaches ¢! desired size Quota sampling. This design is popular in the field of opinion research because it is done by merely looking for individuals with the requisite character- istics. The sampling technique is usually prepared by the main office with instructions to the field researchers to collect data from samples that meet the prescribed criteria or characteristics. * EXERCISES I. Multiple Choice: Choose the best answer among th options. Write the letter only at the blanks on the right column. 1. The best random sampling design 1 because every member in the popu-102 v METHODS OF RESEARCH AND THESIS WRITING Jation hag equal chance of inclusion in the sample is b. ce d. Stratified random sampling Simple random sampling Restricted random sampling Unrestricted random sampling Sampling design in which all indi- viduals in the population is arranged in methodical manner either in al- phabetical or chronological order and the nth name may be chosen in the construction of the sample is sc c. d Systematic sampling Stratified random sampling Unre ricted random sampling Restricted random sampling Sampling design is based on the selection of the individuals as sam- ples according to the criteria of the researcher which serve as controls is Quota sampling Incidental sampling Purposive sampling Cluster sampling Sampling design intended to improve the validity of the sample and is applicable when, the population beingstudied is homogencous is called Unrestricted random sampling Restricted random sampling Stratified random sampling Systematic random sampling bSAMPLING DESIGNS 5. A population of 900 has a sample _ size of a 218 Bi 217 c 219 d. 220 6. Sampling is inapplicable to popula- tion of a. 100 b. 110 c 99 d. 105 Which of the following does not be- long to the group? x a. Quota Sampling b. Purposive sampling c. Incidental sampling d. Cluster sampling 8. The sample size of a population of 750 is 210 21 c. 208 d. 212 9. A population of 2,000 has a sample size of a. 235 b. 238 ic. 232 ds 12287 ~ 103,104 ueTHODS OF RESEARCH AND THESIS WRITING 10, Sampling design in which the population is grouped into small pa its such as blocks or districts is a. Purposive sampling b. Quota sampling c. Cluster sampling d. Multistage sampling 11, Which of the following does not be- long to the group? a. Purposive sampling b. Multistage sampling c. Cluster sampling gd. Systematic sampling 12. Sampling design in which the re- searcher simply takes the closest individuals as subjects of the study because they are most avail- able 1s a, Quota sampling b. Purposive sampling c. Cluster sampling gd. Incidental sampling 13. A population of 300 has a sample size of a. 161 b. 168 -e. 165 d.), 163 14. A sampling design popular in the field of opinion research is 10. uw 13. 14,ee SAMPLING DESIGNS “ios a, Incidental sampling b. Cluster sampling © Quota sampling d. Purposive sampling 15. The sample size of a population of 15 550 is a. 196 b. 194 C .. 192; d. 198 Il. Essay 1. Differentiate samp.c distribution from population distribution, * 2. How do you conduct lottery te hnique and table of random numbers technique? Give examples. 2 In your own opinion, which is better sampling de- sign, scientific or nonscientific? Why? 4, Why is sampling not applicable to small popula- tion? o Explain briefly the advantages and disadvantages of sampling III. Compute the sample of the following population. . 1. 2a0 | 2. 340 3. 570 ] 4, 890 : } 5. 2,300106 Answers to Exercises Multiple Choice 1 : 2. 3. 4. 5. METHODS OF RESEARCH AND THESIS WRITING d a Seana co ao 1. 12, 13. 14. 15. cap