Accuracy in Surveys - Assignment Example

Statistical analysis of the survey of the population sample-random members of the population 19-50 years old Statistical analysis of the population sample of the random members of the population in the age group of 19-50 years old were done applying the laws of probability by using the methods of standard deviation found out by research in a scientific approach. To find a solution to the problem by research requires identifying and analyzing the nature of the problem. Analysis largely depends on the collected data, and so the importance of the analysis lies with the quality of the data collected. Reliability and validity are necessary for accuracy. Gaining information from the collected is the science and art of statistics, empirical research is statistical reasoning, for this reasons, researchers from different fields are trained in for the application, confidence intervals, significance probabilities, hypothesis tests, or posterior probability distributions. Some ethical considerations were also discussed while conducting surveys. Introduction: Solution to the problem is of statistical methods in research. The present study is based on the scientific survey of sampling method and the sampling type is random sampling widely used for population sampling studies (2,8). In the present study methodology that was adopted for the experiment and the ethical considerations for surveying were discussed. Methodology: The methodology comprises collecting the data, analysing the data, and obtaining results and conclusions from the data. In the surveys where random sampling method is adopted, a particular criteria is adopted for selection, under this criteria different groups are selected for the same parameter, and by lottery procedure, a member is selected from each group for testing. This is the probability sampling method where the controversy on the selection procedure is less, the balance between the sampling frame and the population will be good by this procedure (3), (4). (5) A good designed experiment gives an idea of how, one variable responds to the changes in other variable in the controlled conditions of the experiment. While investigating the possibility of cause and effect relationship, the variable that is responsible for the effect is termed as ‘Dependant variable’ because this variable depends upon the causes, variables that represent causes are ‘Independent variables’. Confounding variable is a variable that correlates with the independent and dependent variable, thus it is not possible to determine whether the changes in the independent variable causes changes in the dependent variable or confounding variable, if the research data was not collected on confounder.(6),(9). The quality of the data determines the inferences. Interpreting the data that eliminates random error is important, laws of probability analyses the random error. The likely impact of the errors of chance can be determined by statistical methods like, standard errors, confidence intervals, significant probabilities, hypothesis tests, or posterior probabilities.(12)(15) In the present study the criteria is selecting random members of the public in the age group of 19-50 years. The data was collected for 15 groups by different researchers at different times, 15 groups were directly interviewed, thus 15 data papers were non-interactive, and 15 papers were interactive. Experiment : Population size: 30 Interactive papers: 15 Non-interactive papers: 15 In the present experiment, 15 are interactive and 15 are non- interactive papers, by applying the law of probabilities, the following inference can be drawn. For clarity of the sampling procedure, the following parameters are calculated.(14) Estimator: The numeric characteristic of the population is estimated by using the estimator; a static computed from the sample data. In the example of the present experiment, the difference in getting information from interactive and non-interactive papers is used to estimate the corresponding difference in the rest of the population. Let us assume that the information has been obtained from five of the non-interactive papers and seven of the interactive papers. The sample rates are therefore 5/15 i.e., 33.3% for non-interactive papers and 7/15 i.e. 46.6% for interactive papers. The difference in sample rates then becomes 46.6%-33.3percentage=13.3. Standard error: The magnitude of the random error is determined by the standard error, the estimate percentage may be off from 13.3 to a lower level due to random error, estimate should be accompanied by random error whenever possible. In the present example, the standard error is about 7 points; the estimate percentage of 13.3 points is of by 6 points. It cannot be stated how far the estimate would be off, but the 7 points would be likely magnitude of the error. Confidence intervals: this parameter gives a precise idea; if the standard error is one, and it would be correct by 75%, Standard interval is abbreviated as SE and range of confidence interval is 75%.the interval then goes from 10-30 points estimate-1SE to estimate +1 SE if higher confidence levels are wanted then the interval must be widened say 97%, then the 97% confidence level is about estimate-2SE to estimate + 2SE the confidence level goes from 0-40 points, if confidence levels of still higher level are wanted then the interval is widened by 99.5% estimate-3SE to estimate + 3SE the confidence levels go from -10-50 percentage points. The confidence levels depend upon the standard error; if the error is small, the estimate is right, if the standard error is more the estimate may be wrong. The standard error and confidence levels generally ignore selection bias and response bias, which may happen with population sampling. P-values and Hypothesis tests: In the present experiment, 15 papers were non-interactive and 15 papers were interactive drawn randomly from a section of population in the age group of 19-50 years. An interview was conducted in these two types, as assumed; information has been obtained from five non-interactive papers and seven interactive papers, then the percentage of information obtained is 33.3% and 46.6% respectively. The sample difference in the obtained information rate is 13.3 percentage points. The p-value answers the following questions. In the whole population, from which the information has been obtained randomly the rates information obtained from two groups are 33.3 and 46.6 respectively. The difference in the rates of information obtained is 13.3; the p- value answers the following questions.(13) If the information obtained from the selected population is identical, then why there is a difference of 13.3% is observed in the two samples. The assertion that information obtained from the whole population is same is called ‘null hypothesis’. The null hypothesis asserts that there in no difference in the information obtained from the random members and the difference in percentage in sampling is due to the luck of the draw process. The p-value is the probability of getting as extreme as or more extreme than, given that the null hypotheses is true, P = Pr (extreme data/ null hypothesis in model) If the null hypothesis is true there is only the chance of getting a difference of 5% in the percentages of 13-20%, the p-value for the discrepancy would be 5%or 0.05. in the examples of this type, small p-values are evidence of disparate impact and larger values are against the disparate impact. Significance level: Statistical is determined by comparing a p-value to a preestablished value, the significant level. If the difference in the p-value to the value of significance is less, null hypothesis is accepted and if the difference is more the hypothesis is rejected, that is termed the power of tests. Normally a value of 0.05 is accepted for population studies. Discussion: Surveys normally provide a means to measure the characteristics of the population, behaviour as told by them and observed awareness and attitudes or opinions. Information obtained like this helps to evaluate and implement any governmental programmes and policies.(11). In a valid statistical survey of population’s significance, a haphazard selection of the members for interview should not be practised. Practical possibility of getting the information, as to the willingness of the members to give the response should be analysed if the survey is to be conducted. The appropriate time and the available time should be calculated for collecting information. While surveying ethical considerations must be taken into account, that is wherever required, the confidentiality of the data should be maintained. While taking the information the anonymity of the respondent should be preserved. If the identifiable information of the respondent is required for the follow up of the research process, then consent should taken from the respondent. The adverse effect on the respondent’s participation in the survey should be checked. There should be some procedures to check the identity of the researchers by the respondent, while survey is conducted respondents should believe that, it is voluntary and not compulsory. If it is necessary to take the interview of children under the age of fourteen, the consent from the parents or the guardian of the children should be taken. The above factors must be taken into consideration while developing a good sample design and survey method. With the assistance from the office of the government statistician, depending on the expertise of the agency or department, survey can be conducted effectively(6). References: Accuracy in surveys and the frequency of the surveys need to be checked. (6) 1. course on statistical quality control : prepared by the research Reference guide on statistics : David.H.Kaye, David.A.Freedman 2. statistical quality control : M.Mahajan division 3. M.A.Flatratrone.et.al., exercise training and nutritional 38 4. Journal of prosthetics and orthotics, 1995, vol.7 (3), pp 105-112, Research forum--The Research Sample, Part I: Sampling, Thomas R. Lunsford, and Brenda Rae Lunsford. 5. Queensland government, office of the economic and statistical research, data search papers, Conduct of Sample Surveys by Queensland Government Agencies 6. David.S.Moore and George.P.McCabe, introduction to the practice of statistics (202) second edition, 1993. 7. David freedman et.al., Statistics ( 2nd edition) (1991) 8. reference guide on statistics , David.H.Kaye and David.A.freedman 9. the health benefits of exercise, a critical reappraisal 328, New. Eng. Jour.med 574-575 10. Isaac 5, Michael W. Handbook in research and evaluation, 2nd ed. San Diego: Edits Pub.; 1990:189. 11. Schlesselman JJ. Case-control studies: Design, conduct, analysis. New York: Oxford University Press; 1982. 12. Dominowski RL. Research methods. New Jersey: PrenticeHall; 1980 13. resampling based multiple testing: examples and methods for p-value adjustment(1993): Joseph.L.Gastwirth and Samuel.w.greenhouse 14. Gregory.A.Kimble How to use statistics, (1978) 15. Micheal oakes , Statistical interference : a commentary for the social and behavioural sciences (1986) Read More