Research Paper

Introduction

Methodology

What are the units of analysis? What is the sample design? Where are you going to get the data? Secondary Data Analysis? Existing data from agencies? Data from other studies? Your own survey? (Include the survey in an appendix to the paper) Field observation? (Include the instrument in the appendix) Experiment (Include description of apparatus in the methodology section). How did you measure your variables? Dependent Variable? Level of measurement. Categories of responses. Question or item on form. Quote literally. Can give descriptive statistics such as means and standard deviations for interval or ratio data, and frequencies or percentages for nominal data. Evidence of reliability and validity? Independent Variable? Level of measurement. Categories of responses. Question or item on form. Quote literally. Can give descriptive statistics such as means and standard deviations for interval or ratio data, and frequencies or percentages for nominal data. Evidence of reliability and validity? Repeat for each independent variable. How did you control for rival hypotheses? What are some possible threats to internal or external validity? Control variables? How measured. Research design? Is the research design an Experiment, Quasi-experiment, Statistical study? How is it controlling for rival hypotheses? What statistical analysis will be used? What is the logic of the method? What are the assumptions of the method? Why is appropriate for your data? There can be more than one method used, so this might be discussed for several methods.

Example

For Chi Square, the logic is that it examines whether the observed data differs from what would be expected if there were know relationship. It assumes the data are measured at the nominal or ordinal level of measurement. It requires an expected value greater than 5 for 95 percent of the cells. Data with too many categories (such as interval or ratio data) generally fails to meet the expected value assumption. The null hypotheses is rejected if the probability that the results are greater than sampling error is less than 5 in 100

Findings

Organize the findings around the hypotheses being tested. There should be statistical tables for each hypothesis. Present the statistical tables. The tables should stand alone for a reader to understand without reading the text, and the text should stand alone for the reader to understand without looking that the tables.

The paragraph should begin with a reference to the table number to which it is referring. What is the hypothesis being tested? What is the statistical test used in this table?

First indicate that the test is valid. If it is not, eg., expected values are less than 5 for more than 5 percent of the cells, there is nothing more you can say. You need to refer to a different statistic.

Second, indicate whether the variable or variables examined are significant.

Third, describe the relationship. Is the direction direct or inverse? How strong is it? (Measures of association.)

Do you reject or not reject the null hypothesis and test hypothesis? Repeat this information for each table.

Discussion

Give explanations for your findings. Relate your findings back to the theories and research literature. Did they corroborate or challenge any existing theories? Are they similar to what has been found? Are they consistent with other research and findings? If not, why do you think your results are different, eg. better methodology, worse methodology, or methodological problems that you could not resolve in this study. Conclusions Summarize the findings. What has your study contributed to knowledge in the field. Are there any limitations to the study. No study is perfect. You should document limitations throughout the paper, but mention the key problems again here. What are some implications for future research? How could the study methodology be improved? What future questions could be addressed?

Implications. Given that your findings stand the test of replications and are intersubjective, are their any policy implications? Appropriate in some journals but not others.

Analyzing a Regression Table

What are the hypotheses? Each variable in the equation is one hypothesis. Which variables are significant? Examine the significance of the b coefficients with the t-test. How much does the dependent variable change when the independent variable changes by one unit? Refer to b, the unstandardized regression coefficient. Which variable is the most important? Refer to Beta, the standardized regression coefficient. How complete is the analysis? Refer to R Square, the proportion of the variation in the dependent variable that is explained by the variables in the regression equation. It ranges between 0 and 1. BE CAREFUL. R Square can be influenced by sample size, and the number of variables in the equation, whether they are significant or not. The number of variables cannot exceed the number of case or R Square will be perfect. If R Square is too high, start checking for artificial reasons for it. How much can we be wrong by? Refer to the Standard Error of the Estimate. Is the prediction of the equation better than we would expect by chance? Refer to the F-test. It is essentially a significance test of the R-Square.

Analyzing a Logistic Regession Table

What are the hypotheses? What variables are significant? Wald Test. How much does the log odds change given a change in the independent variable? B and EXP B How good is the equation? Model Chi Square and Percent Correctly Classified.