Are you looking for which statistical methodology will be the most appropriate for your doctoral thesis? After many years in academic and corporate statistical consulting, we find that researchers are often perplexed about the differences between quantitative research designs and in what settings they should generally be utilized. Finding an adequate quantitative approach for your dissertation depends on several factors, including the purpose of your study, type of data being used, and any associated cost or time involved in collecting data.

ARC’s mission is to make sure that our academic clients are perfectly poised to tackle data collections after the  research methodology approval. Often times, utilizing secondary data will be the simplest approach as we’ll be able to apply a pre-existing dataset to explore new associations between variables. Many clients, however, still opt for a primary survey approach utilizing upwards of sometimes up to 200-300 participants to draw meaningful statistical inferences regarding their thesis research questions.

With that said, one big consideration in selecting an appropriate methodology is to identify the target sample size for your study. We are often able to draw meaningful statistical inferences from some limitations for data samples as low as 20-25 participants, although many PhD committees require that the chosen sample size will yield statistical power of 80% or greater.

Starting with our pre-dissertation writing assistance where we will be fully involved in your Topic Approval, we will be able to help at any stage prior to the actual Data Analysis to make sure that you’re working with a valid methodological approach for your thesis research and that we will have no problem satisfying your reviewers with this aspect of your Introduction, Literature Review, and Methodology work.

For doctoral-level statistical research methodology, we will also need to fully describe the population that you’re focusing on for your data analysis as well as the specific participant selection approach–for instance, a type of random or non-random (or convenience) sampling procedure. We may also need to describe why the chosen research design is the most appropriate for your doctoral research,  and finally, the data collections procedures (including the IRB Process) and the data analysis plan involving your descriptive and inferential analyses. Below are some of the most commonly used quantitative research designs and settings where each design should be applied:

In a correlational study, our aim is to discover the relationship between the independent and dependent variables. Most commonly, we can compute the correlation coefficients for each bivariate relationship of interest, utilizing the most common types including Kendall, Pearson, and Spearman coefficients. However, it is also important to note that other approaches such as multiple regression analyses and path analyses are possible in case your study contains multiple variables that may be connected through complex underlying statistical associations.

Notably, there is a tendency to confuse the correlational research design with correlational analysis. However, it’s important to note that correlation can simply be described as the way change in one variable affects another.

In a causal-comparative study, our aim is to suggest a causal basis for the association between the independent variable of focus (or predictor variable) and the dependent variable (or outcome variable). Although we can never “prove” causation using statistical inference, we can get close by being able to describe the relationship between the two main variables in detail while addressing any potential issues of confounding.

The causal-comparative approach is generally recommended in settings where a researcher wants to suggest that a dependent variable is a direct consequence of the main independent variable. Our recommendation is to incorporate a regression-based methodology consisting of the chosen variables for your research.

An experimental design can be very simple or complex depending on what your experiment involves. Typically, many of our clients design intervention studies in clinical settings where control and treatment groups are subject to different levels of treatment. Experimental studies are more than possible in most areas of research, since the core requirement is just that there needs to be random assignment of participants into experimental and control groups. One feature is that the independent variable is generally modeled as binary or categorical variables, with the ANOVA and ANCOVA approaches being the most popular in cross-sectional settings.

Approval works with world-leading biostatisticians and statisticians to provide you with full support—from the development of the experimental design for your dissertation research to the final completion of the data analysis phase of your experiment.

A quasi-experimental research design is very similar to an experimental design, except that treatment groups are not randomly assigned. One example of this is when the researcher does not have a choice of which sampling scheme to implement, and therefore, will need to work with pre-specified inclusion criteria and group assignments. The main potential issue with lack of random assignment is that a researcher risks the treatment groups not being balanced in terms of their baseline characteristics, although this can always be checked through descriptive statistics and easily remedied by procedures such as including additional control variables to take care of confounding.

Do you think you might be working with a quasi-experimental design? Our world-leading PhD statistical consultants will be able to instantly determine what the best suitable design will be for your doctoral research and set up a personalized plan based on your doctoral requirements.

Often considered the most involved of the simpler statistical methodologies, SEM is used the most often in the Social Sciences when there is an interest in analyzing structural relationships between the covariates in your study. The procedure normally involves combining factor analyses with regression-based methodologies to discover the structures governing the relationship between covariates in your study as well as the latent constructs that we are able to infer. In the world of more rigorous and theoretical Statistics, SEM is understood to be a type of applied causal inference approach where we seek to clarify all dependencies between the endogenous and exogenous variables in the model.

Approval often ends up taking the most difficult statistical consulting projects in the academic consulting industry today for obvious reasons. We work closely with biostatisticians and statisticians who have years of expertise in both academic and corporate settings, and you will immediately find that we are the most passionate statisticians in the world.

Seeking professional guidance for the topic approval process?