Applied Research Methods
Applied Research Methods
Assignment Instructions: Module 5 – SLP
Variables or Constructs: DSP Chapter 3 Applications
Based on your proposed DSP topic of interest, and recommended research methodology selected in Case 4, write a draft for the Chapter 3 – Methods section, Data Analysis Procedures.
Module 5 – Resources
Variables or Constructs: DSP Chapter 3 Applications
By now you should have a solid selection for your proposed DSP research method and design.
Research Method
Research Design refer to page in Leedy & Ormrod (2019) text
Instrumentation
Comments
Descriptive (non-experimental)
Survey [p. 153+]
1 survey for each dependent variable.
Quantitative data analysis. Access to large sample size is common >100.
Descriptive (non-experimental)
Retrospective [p. 211+]
Data already collected but not analyzed.
Quantitative data analysis. Access to large dataset is common >100.
Action and Participatory
Prospective Program Evaluation [p. 282+]
Multiple options include surveys, data, Likert scales, outcome measures, etc.
Quantitative data analysis. Researcher must be involved in the program being evaluated. Program is NOT designed with input from participants. Pre and post evaluation data is most commonly collected and compared. Moderate sample size is common >30.
Action and Participatory
Participatory Action Research [p. 285+)
Multiple options include surveys, data, Likert scales, outcome measures, etc.
Quantitative data analysis. Researcher must be involved in the program being evaluated. Program MUST have at some input from participants. Pre and post evaluation data is most commonly collected and compared. Moderate sample size is common >30.
Mixed
Convergent [p. 261+]
Requires 1 quantitative (e.g., surveys, data, Likert scales, outcome measures, etc.) source and 1 qualitative (e.g., personal interviews, group interviews, observation notes, etc.) source of data.
Total of 2 instrumentation options for Quantitative and Qualitative data analysis. Small sample size is common 12-15.
Mixed
Case Study [p.230+]
Requires 3 total sources of data, with at least 1 being quantitative (e.g., surveys, data, Likert scales, outcome measures, etc.) and 1 being qualitative (e.g., personal interviews, group interviews, observation notes, etc.).
Total of 3 instrumentation options for Quantitative and Qualitative data analysis. Small sample size is common 12-15.
One section of the DSP that is particularly important is related to the Data Analysis Procedures and Variable Identification and Mapping that corresponds with your proposed topic and construct or variable(s) of interest. Considerations for this section of the Chapter 3 Methodology of the DSP are as follows:
Variable Identification and Mapping
As cited in DHA600, the 4 categories of variables are applied as follows:
Nominal – Yes/No, or Male/Female for example. Any dichotomous set or pair of options.
Ordinal – Rank order data like Strongly Disagree = -2, Disagree = -1, Neutral = 0, Agree = 1, Strongly Agree = 2
Interval – Data without an absolute 0 like a survey score representing a construct and ranging from 20-50.
Ratio – Data with an absolute 0 which are often expressed in percentages, or averages including the use of decimals.
A table listing all variables for the DSP will be constructed. Independent variables will be identified, as will dependent variables. Confounding and/or control variables will also be specified and separated from demographic variables used to create a demographic profile. *Note: Some research books will use the terms mediating or moderating variables instead of control or confounding variables (MyClassRoom, 2015). Using our research sample topics of interest and an action research method, a table may look like the following:
Table 1
The intervention in this scenario (4-week Burnout Prevention Program participation by a group of nurses for Acme Health System) allows the researcher to ultimately test the effectiveness, or lack thereof, of the program or intervention. While additional demographic variables may be collected, for simplicity we have only identified 2 that could influence the post-intervention results (tenure – number of years in nursing, and annual income). It would be hypothesized and likely supported in the Research Literature that tenure (more years of nursing service) would decrease the program intervention effectiveness and income (higher income) would increase the program intervention effectiveness. The difference between pre- and post-intervention burnout levels will ultimately determine how effective the program is with a specific group of nurse participants.
A variable map (figure) can then represent visually the relationships between variables of interest. Word is a simple program to use. Click on Insert, Shapes, Text Box, and create.
Data Analysis Procedures:
Much like the Data Collection Procedures of Chapter 3, this is a very standard and required section. Just outline how all of the variables (data) will be analyzed once IRB approval is attained.
All data begins with a database (Excel or SPSS, commonly) coding procedure that is communicated. Code to ensure anonymity and privacy of participants, but also in a way that can be tracked for hypothesis testing. P1…P200 is common. P stands for “Participant.” Number the first to last participant so all other data is tied to a specific P number.
Each demographic variable (data) is evaluated descriptively (DHA620 will cover this in detail) to include mean/average, standard deviation, min-max values, or others of interest (e.g., mode, median, variance, kurtosis, etc.).
The same descriptive analysis process is also applied to all independent and dependent variables of interest. In the case of a survey instrument as above, you would want to include the entire list of questions for the instrument, and then provide an aggregate (composite value as instructed by the instrument authors). For example, the MBI has 40 questions which would be individually reported in Table or Figure format with descriptive statistics and then a separate Table or Figure would represent the overall score that represents the variable of interest (Burnout) for hypothesis testing.
Inferential data analysis is then required to test study hypotheses and answer guiding research questions. Commonly, these may include Pearson correlation coefficients, chi-square tests, t tests, ANOVA, MANOVA, and other statistical calculations (more in DHA620). For our example above, we would do a paired-comparison t test statistic for pre- and post-Burnout Prevention Program MBI scores. The results would be summarized in Table and/or Figure format with the average values reported, t scores, and most importantly P values. P values represent the Type 1 error. Standardly, we allow a 5% or less P value as the gold standard to support Ha (Alternate Hypothesis) that there IS a statistically significant difference, relationship, ability to predict with at least 95% confidence in that conclusion. If p values exceed 5%, we typically conclude that the Ho (Null Hypothesis) is supported and there is NOT a statistically significant difference, relationship, ability to predict – at least not at a 95% confidence level.
In our example, as with most DSP topics, we also have Control Variable analysis to complete. It is considered inferential in nature and would utilize a Pearson correlation coefficient from which to determine if there IS or is NOT a significant impact on Burnout change from the control variables of tenure and annual income. The sample P value rules apply in order to tell us whether or not these control variables have a significant influence on the Burnout Program effectiveness and results. The same process is applied if you have confounding variables identified. These may include things that are unanticipated like the Covid-19 pandemic where an unexpected increase in workload and job stress is likely to skew the results of the study.
*Predictive data analysis may be applied in your DSP depending on the variables used, and goals (hypotheses). In the example provided here, predictive data analysis does not apply. Multiple regression, linear regression, and binomial regression are the most common statistical analytics used in research. Examples will be provided in DHA620.
*Qualitative data analysis may also be applied in your DSP (Mixed Methods). In the example provided here, this does not apply. However, the standard process of analyzing qualitative data (e.g., interview transcripts, observation notes, etc.). The distillation process begins with open coding, then axial coding, then selective coding. Most commonly, a computer program assists with this process of coding or funneling information into a cogent representation for the entire sample of participants. At the end of the process the researcher will generate themes from the selective coding information from which to answer the Research Question of interest. Also keep in mind that themes are informed by quantitative data as well, which is part of triangulation common in mixed-methods study design. More on this in DHA620.
Data Analysis Procedures provide a natural segue to Chapter 4 (Data Analysis) once you have obtained IRB approval.
Required Reading
Allen, J. (2019). Writing a research proposal. The Productive Graduate Student Writer (1st ed., Chapt 31). Routledge. 7 pgs. Available in the Trident Online Library.
Core concepts in research. Variables (2018). Shortcuts TV. [Video] Retrieved from Trident University Library.
Elliott, V. (2018). Thinking about the coding process in qualitative data analysis. The Qualitative Report, 23(11), 2850-2861. Retrieved from Trident University Library.
Mediator (mediating) variable (2008). In Miquel Porta (Ed.), A Dictionary of Epidemiology, (5th ed.). Oxford University Press. Retrieved from Trident University Library.
Organizing quantitative data. (2005). In Films On Demand. Films Media Group. Available in the Trident Online Library. 37 min
Answer preview to Applied Research Methods
APA
1200 WORDS
