Common Quantatative Analysis Questions and Answers.

Below are common Quantitative Analysis related Questions and solutions

What is the nature of causation and what are the criteria for assessing causation?

The nature of causation is often to show that causes differ in significance and thus warrant further investigation and examination in details. This investigation often falls in three distinct categories to determine the necessary cause, sufficient cause and contributory cause. To show causation it must be obvious that the variables are statistically related. Causation cannot be prove with complete certainty.
Causation also differs in the timing of their influence.

The major criteria for assessing causation are whether the conditions for the cause do exist in the first place. Causation always makes it possible to conduct further investigation on correlation between variables.
It is important to know whether the causation is linear in nature or reciprocal in order to assess it. The linear causality assumes causal influence is in one direction. While reciprocal depict the causal influence is bidirectional or sometimes circular.
In quantitative research analysis, correlation can never in its entirety show causation because it does not imply causation.

Discuss the difference between experimental and correlation research?

Experimental researches involve processes that clearly define the design approach used in conducting the research in question. In a way, it emphasized how data is collected for final analysis. In experimental research, the researcher can manipulate his or her experiment and check the effect or results of the initial manipulation.
While correlation research involves analysis tools used in conducting the research. Variables are not influenced but only the relationship between them is measured and show by the researcher. In addition, sometimes there is correlations in experimental research but typically one of the variables is manipulated.

Finally, experimental research is only about experiment while correlation research is common in archival research, naturalistic observation, survey research, and even in case study.

 

How can surveys be designed to elicit the most valuable responses?

Surveys generally accomplish two main purposes to know new thing about the research and to be able to generalize from the survey data. Therefore in order to elicit adequate and most valuable responses from respondents, it is essential to use a design approach that is consistent with the following:
-The quality of survey questions must be well tailored to the focus group.
-The questions should be well designed, meaningful and sample size manageable.
- Appropriate survey techniques such as mailing, email, telephone must be employed and used in the circumstance under consideration.
- Appropriate language of communication that is comfortable with the audience especially during face to face encounter must be used.
- Survey questions must be devoid of ambiguity. It is important to test the questions.
- A face to face encounter with survey respondents often provides more tenable answers.
- The researchers must provide enough clarification being asked by survey respondents.
- There is no basis for being bias because a biased sample will produce biased results.

When do ethical issues become important in experimental research?

The ultimate aim of any experimental research work is to advance and contribute to the body of knowledge. Thus serious ethical issues become concerns when for instance there is glaring abuse in selection process of research participants. A good selection process must be undertaking and encouraged. It must be deemed fair and balance and not tainted to favor a group or set of people.
Informed consent must be afforded to all research participants partly to address any of their concerns as well as afford them the opportunity to know what the research is all about. If otherwise, the experimental research could be said to fail one of the ethical standards recognized worldwide in research projects.
The risks and benefits of the research must be well spell out and clarified to  subjects or research participants in order to alleviate issues that bother on ethical concern. Which I believe can later jeopardize the whole research efforts.

Discuss threats to validity.

Bias will increasingly be recognized as one of the most important threats to validity that must be addressed in the design, conduct and interpretation of research. Not removing bias can in the long run compromised the final interpretation of research result hence its validity.
Also, the low reliability of measures which can be attributed to factors such as poor question wording, bad instrument design or layout, illegibility of field notes are great threat to research validity.
Passing communication among research subjects who are not suppose to know it until the research exercise is over can affect the validity of the research in a negative way. This will greatly reduce the measurable effects of the research program because the information so obtain indirectly cast doubt on the true validity of the research results.

How can computer output for t-tests and confidence intervals be interpreted?

The t-tests output can be interpreted in the context of regression analysis whether a regression coefficient (b) is significantly different from zero. However, in the context of experimental work, it is used to test whether the differences between the two observed means are significant different from zero. For instance, SPSS provides the exact probability that the observed value of “t” would occur if the value of “b” in the population were zero. In this regard, if the observed significance is less than .05 (p <0.05), we may conclude it reflect a genuine effect from the population.
Confidence intervals can be interpreted from means by looking at the range of values that contain with some probability (95%) of the true value from the population. In a computer output, it represents the 95% of the difference between the lower and upper limits by estimate. This mean that the true value of the difference in population is somewhere between the limits and we can be 95% confidence of this value.

Are regression and ANOVA antagonistic methods.

The answer is affirmative no and undoubtedly, Regression and ANOVA are both based on the general Linear model (GLM). The techniques are recognized as two most useful statistical analysis tools be it in business, psychology or sociology circles. In fact, they are complementary tools and are not antagonistic as some school of thoughts argued. For instance and in most researches, when it is desirable to test a multiple regression equation for the statistical significant factors or measureable variables, it is imperative to start with ANOVA.
In the same token, when using ANCOVA for analysis the degree of covariance and in order to improve on an ANOVA technique it is usually incumbent on the researcher to use regression methods. The regression method applied in this way enable on the spot adjustment for a control variable before embarking on ANOVA. Reliability of statistical analysis and results in such research undertaking are further reinforced. So I submit to this forum that it depends on what the statistical research want to achieve and the information the researcher wants to convey to the public.
It is important to stress that for any statistical output to be considered validated when using ANOVA or regression techniques; certain assumptions must be fulfilled and tested. For example in ANOVA, measurable independent variables must be categorical.

Reference:

Field P.A. (2009). Discovering Statistics Using SPSS, 3rd edition.  Thousand Oaks, CA: Sage.

Vogt, W. P. (2007). Quantitative research methods for professionals. Boston: Pearson Education, Inc.