Presentation on theme: "Writing Up Results POL 242. Overall Write for what your audience needs to know. Think of the 1-3 main points you want readers to learn from reading."— Presentation transcript:
Overall Write for what your audience needs to know. Think of the 1-3 main points you want readers to learn from reading your paper. Too much will cause the reader to lose track of your main points. Don’t forget that you also need to demonstrate competence and knowledge. Leave time to refine ideas and revise, revise, revise.
General Organization This is like other good papers you have written. Display an clear structure Burchinal (and others): "Tell them what you are going to tell them; tell them; and then tell them what you have told them.“ Employ headings judiciously See Gibson (p. 215) Short sentences – 10-15 words max (with few exceptions). Short paragraphs.
Abstract / Executive Summary Brief summary of study Key findings explained Write as if the reader will not read any more!
Introduction Describe and define the nature of your investigation. The problem you are seeking to explain. The riddle you are trying to interpret. The puzzle you want to elucidate. The question(s) you are asking. Provide primary rationale for research. Remember you are not doing advocacy. Let us know what you find – outline rest of report. Limitations of study (?) Limitations are okay- better you are upfront about them.
Literature Review Why is this study important? Interesting? Provides rationale for study. Organize by date or theme. Best: presents sides of a debate emphasizing the puzzle or riddle. Qualitative research findings would likely go here (or in Results).
Hypotheses One hypothesis or many hypotheses. Can be for entire model; Can be for each independent variable. Often want to pare down hypotheses to the ones you want to focus on. Provide rationale for each. May include additional or reiterated literature review. May include descriptive statistics. Do not hesitate to go back to hypotheses after finishing conclusion. Perfectly OK to reject a reasonable hypothesis. May want to add another hypothesis (or revise conclusion)
Results What does the reader need to know? May need to discuss research design or source of data (survey details). What makes your point clear? Descriptive Analytical
Descriptive Data Providing some descriptive statistics about the distribution of your dependent variable is strongly recommended. What is the variation you are trying to explain? See Gibson Always need detailed descriptions of indexes. Might include descriptions of IVs. Might include crosstab(s) or correlation(s) illustrating key relationships. Might just include key measures of association. Tabular or graphical presentation of results.
Qualitative findings Can go in literature review. Author Fred says this, my crosstab said that, and my focus group all thought they were bonkers. Can go in hypotheses. I hypothesize because my Uncle Dave said X in his interview. Can go in descriptive results Describing range of opinion. Preliminary discussion of relationship
Analysis Remember primary question regression is seeking to answer. What do you look at first???? But regression analysis can be used for other questions. Keep in mind your main hypotheses. Remember that you are looking at a relationship (or several relationships) May want separate section for each major variable or relationship (or sets of relationships).
Tables, Graphs and Diagrams “A picture is worth 1,000 words” You should explain the key results from the table, graph or diagram The illustration should be so clear that you should not need 1,000 words. If you need 1,000 words to explain a diagram… Be brief, let table provide details and your writing focus on big picture. Do not miss opportunities to tie into hypotheses.
Tables, Graphs and Diagrams Connect main point to what a specific illustrative format does best. Graphs are great for comparing different values. Arrow diagrams are great at providing a whole picture and/or highlighting specific relationships (like an IV that explains a large share of the variation). Standardized coefficients (betas) are often used in arrow diagrams because they allow one to make comparisons between variables. Some will also just use the sign of the variable. TWO OR THREE DIGITS MAXIMUM
Arrow Diagram Example Clearly shows relationship and strength of relationship. Can also put coefficients inside boxes Can highlight strong(est) relationship (see example on bottom)
Regression Tables Tables are great for comparisons, especially when you have much information to convey. Can put results of more than one regression in same table Coefficients (B) are always included Standard errors are often included as separate column or under coefficients in parentheses. Statistical significance (P < Z) is almost always included as separate column or as *** next to coefficient (one star for lowest level of significance, three or four for highest). R-squared and other modular statistics can be added as rows in the table. Does not mean you need to include all the possible information. In regressions with many IVs, you can exclude some controls from table if they are not important to hypotheses (note at bottom which variables are also in model)
Organizing Regression Table Put more important IVs at top or bottom. Group like-variables with sub-headings Ex: Demographics Women Age Income R-squared and other modular statistics can be included as additional rows at the bottom of the table. Source of data is usually included as a note under the table.
Visuals: Causal Explanation Visual clarity should match explanatory clarity. Colors or shading should match ordering of data. Present all relevant information, even if it may contradict your point. KISS
Match Chart and Comparison - 1 Component – Pie Use only when you are illustrating parts as a percentage of some whole. Very useful if you want to highlight share of one part. Difficult to compare one pie to another pie. http://www.sapdesignguild.org/resources/diagram_guidelines/index.html
Bar Charts Item – Bar Bars can be arranged in any order Great for categorical and nominal variables, especially with lengthy labels. Great for comparing values. Useful for showing ranges. Scale at top or bottom.
Frequencies Time Series and Frequency – Column or Line Unlike bar, both axes of column chart are ordered. Subdivided columns compare changes in parts of the whole better than multiple pie charts. Lines show trends and skews very well and smooth over slightly irregular distributions.
Conclusion Summarize main findings. Directly connect to hypotheses. Tie into puzzle. Generalize Implications for literature or public policy. Limitations Implications for future research.
More General Tips Watch your tense. Spell-check! Headings are your friends. You can use multiple levels of headings. Always be clear what your unit of measurements are. TWO OR THREE DIGITS MAXIMUM Be clear: percentage or percentage point change/difference.