Working with Your Tentative Quantitative Research Question - Assignment Example

? Tentative Quantitative Research Question. al affiliation: One of the crucials in understanding the internal validity is the ability to recognize that, when it is associating with studies it deals with how perfect the research was conducted The variables determined include the research design, definitions of the operations used, variable measurability, and the confidentiality of the independent and the dependent variables. Internal validity answers the effect of subject variance in the experimental and control group. Additionally, in one-subject research it resolves the cause of a treatment in the behavior of the subject compared to other factors. In descriptive research, is concerned with the study’s quality and accuracy. There are various extraneous variables that jeopardize and control the internal validity of an experiment (Acheson, 2010). They include history, maturation, testing, instrumentation, selection, statistical regression, experimental mortality, and selection interaction. In selection interaction, the methods of selection interact with single or multiple threats, thereby biasing the results of the study. In experimental mortality, the variables drop out of the research. If one group incurs a greater subject level of mortality and withdrawal than others, then the differences between the subjects become questionable. Consequently, the elements in comparison need to be equivalent at the research start. Their equivalence makes the difference between the subjects measured by the dependent variables at the research end are caused by independent variables rather than orgasmic ones. The research is biased if the comparison groups differ from one another during the start of the research (Cochran, 1977). Statistical regression occurs when the study subjects are chosen as the subjects because they over performed or underperformed on some performance measurement. When the subjects are retested, they produce a varied score distribution, and the mean performance measurement will near the population. For instance, if the selected subjects had high scores at the beginning, the groups mean on the retest will be minimum than before. Subsequently, when the measurement method is altered during the research, the measured subject is also affected. Also, when the human observers are used, the observer’s judgment changes every time rather than the performance of the subjects. Consequently, the subjects are always pretested to find out that the subjects are beginning the study at an equal level. An after math of pretesting protocols is that they can change the performance of the subjects on later tests that measure similar domains above the effects that are caused by the same treatment. However, the degree to which the result of the study can be applied or generalized to other people shows its external validity (DuPont & Plummer, 1990). The group research that employs randomization will possess greater external validity than the research that does not employ random assignment. The factors that affect the external validity of a study include: interaction, pretesting, setting, and interventions. The interaction between two variables shows their treatment and selection. If the elements are not selected from the population randomly, then their orgasmic features may be performance biased and the results of the study may not apply you the other \group that represents the features of the population (Faull et al., 2007). Pretesting a subject causes them to either react less or more strongly to treatment than if pretest lacked. In some instances the researcher cannot conclude that population members who did not under pretest would perform in same manner to those subjects in the study. The subject performance is a more reaction or product to the setting of an experiment than in variables that are independent. For instance, subjects knowing they are participants in a research, or who knows of being observed, reacts differently than the experienced subjects who have know no knowledge. Finally, research using interventions have confined generalizability because the prior treatments have cumulative effects on the performance of the subject (Houser, 2007). In a group research, the primary methods that are used to attain external and internal validity are randomization, research design use and analysis of statistics to the collected data and the investigated questions. One-subject experimental research has greater internal validity because the elements act as their control but low with external validity. The studies that are single-subjects attain external validity via extension and replication process. The outcomes of the group studies are acceptable during replication. An example of external validity research question is whether male or female college student models put on make-up or not. The setting is selected group in a college class, the participants are male and female students, and the treatment is make up verses no makeup and was represented by 15 colored photos of the participants (Kalton, 1983). Assuming internal validity is a primary cause, the research problem can be to find out the effect of campaign on the behavior of voters. The hypothesis is the propaganda’s influence on the voting behavior. The intention of voters is compared after and before the voters is exposed to propaganda. Random selection is drawing people’s sample from the population study. Random assignment is assigning the sample to various treatments in the study. It is possible to use random assignment and random selection in a research study. For instance, drawing a random sample of 100 customers from a population of 1000 clients is random sampling. Assigning the 50 clients for new additional treatment and other half to be control. That is random assignment. Random selection is connected to sampling and thus external validity. Random assignment is connected to design. When we assign participants randomly to control and treatment then we have experimental design, thus internal validity. Random assignment is crucial because it assure that the control group as similar before treatment (McCready, 2006). There are various ways of taking a sample. The alternative sampling can either be probability or non-probability sampling technique. In probability sampling every subject in the population has an equal chance of being chosen. Simple random is the common known probability sample where every element has equal chances. The probability sampling is used if the sample representativeness is crucial on wider generalisability (Mayr et al., 2007). When other factors apart from generalisability is critical, non-probability sampling is used. In non-probability, the elements chosen are not known. Their selection is arbitrary because the researchers dwell on the personal judgment. This sampling method has no statistical technique for measuring the random sampling. However, there are events when the samples of non-probability are best for researcher’s purpose. Therefore, researchers collect data using different kinds of methods, from experimental designs to surveys (Piasta & Justice, 2010). The methods involve experimental subject’s choice. For instance, the researcher may need to select the patient under examination in a medical research or the interviewed respondent in a survey. The choices can either be probability based where the selection is a mechanical procedure that involves random numbers or equivalent. Other methods known as non-probability can also be used involving element judgment. When using G*Power to calculate one tailed t-test, click tests, then mean, and lastly one group. Alternatively, the t test is selected under the test family, means. One sample case will be the difference from the constant under the statistical test and cA priori. Given the alpha, effect size, and power, then the sample size can easily be generated. An example of the power analysis is as shown in the below figure. One-tailed test is chosen and then provide the significance level, statistical power and the effect size. The G*Power effect size is Cohen’s d computed by dividing effect by the standard deviation. G*Power report the crucial value of one-tailed test at a level equal to 0.01 under null hypothesis Ho. The value should correspond to the original t distribution. The outcome at no centrality parameter is the t statistic in alternative hypothesis Ha (Selkirk, 1978). Assuming the research study is on teaching methods in mathematics to improve the math score standardization in classrooms. The research study includes various teaching methods and the subjects from fourth grade who are sampled randomly from an urban county school and then randomly assigned to the various teaching methods. In answering the questions, assumptions are needed on the data. The first assumption is standardized standard deviation and national value. The program needs an F-family test as shown: Calculating the sample size: In calculating the sample size we assume that Alpha is 0.05, Beta 0.2, small effect size, and t-test is one tailed with two independent groups of the same size. Additionally, the pre-specified statistical power is set at .8. The two means of the independent groups are first specified 0 for the first group and 10 for the second group. The mean difference will be 10. The pooled standard deviation ( square root of the mean of the 2 SD). Here it is assumed to be 16.04 after the calculation. The power is set at .8. Therefore, the corresponding sample size is calculated as Power t-test (d= (0-10/16.04, power is .8, the significance level is 0.05 for Alpha and .2 for Beta. The t-test calculation will be n= 41.4 d= 0.63 and the significance level is .05 for Alpha and .2 for Beta. N is the sample size of the two groups. The calculations indicate that the groups need a sample size of 42. Computing beta and alpha: When the sample size is halfed the corresponding beta and alpha will be doubled accordingly as .4 and .1 respectively. The study is worth using smaller sample because it saves cost and time. Using a smaller sample size can be researched in a limited time and the total cost is very small. Also, it is reliable because the inference of the target group parameters is easier when a smaller sample is used. Consequently, a smaller sample checks the census data accuracy. Calculating the sample size: The sample size can be computed using the MacAnova as power (C*, g, EI,n). The balance and C* is assumed to be the estimation for non-centrality parameter for n being equal to 1. Therefore, to compute sample size, sample size ( C*, g, EI, 1-EII). Maintaining the same assumption and C*. The outcome of the command will be the per group sample size. Supposing we test the ANOVA null hypothesis for 0.05 for all I at a level of .2. The deviations from the null of a certain size with a specific power need to be detected. This is in reference to the analogy in the means of population and the t and z tests. If the alternative hypothesis Ha is true, the F-statistic that is computed to test the null do not follow the F-distribution. Areas below the distribution for alpha convey the test's power. When the sample size is halved, beta and alpha will also be doubled.Therefore, the selected beta and alpha will be .4 and .1 accordingly. A smaller sample is worthy because it is reliable, saves cost and time. Sample size effect power: Computing sample size effect power. Effect size: Reference Acheson, A. (2010). "Sample Size." Encyclopedia of Research Design. SAGE Publications. Cochran, W. G. (1977). Sampling techniques (3d Ed.). New York: Wiley. DuPont, W. D., & Plummer Jr., W. D. (1990). Power and sample size calculations: A review and computer program. Controlled Clinical Trials, 11 (2), 116-128. Faull, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39(2), 175-191 Houser, J. (2007). How many are enough? Statistical power analysis and sample size estimation in clinical research. Journal of Clinical Research Best Practices, 3(3), 1-5. Kalton, G. (1983). Introduction to survey sampling. Beverly Hills: Sage Publications. Mayr, S., Erdfelder, E., Buchner, A. & Faul, F. (2007). A short tutorial of Gpower [2]. Tutorials in Quantitative Methods for Psychology, 3(2), 51-59. McCready, W. (2006). "Applying Sampling Procedures." The Psychology Research Handbook. SAGE Publications. Piasta, S.B., & Justice, L.M. (2010). "Cohen's d Statistic." Encyclopedia of Research Design. SAGE Publications. Selkirk, K. E. (1978). Sampling. Nottingham] ([University Park, Nottingham NG7 2RD]): [University of Nottingham School of Education]. Read More