Hi everyone!

In this week's reading, by Melody Denny & Lindsay Clark, authors Melody Denny and Lindsay Clark discuss the logic and ethics behind collecting data to answer a research question. The article begins on the note that "how you ask the question determines the type of data you collect" (4). Depending on whether you ask closed-ended or open-ended questions, the methods of data analysis will slightly differ. Closed ended questions require less steps, as you merely compare the frequencies or percentages of certain answer selections, and then ask yourself questions as to why the results are as they are. In open-ended question data analysis, it's necessary that the researcher organizes. Organizing can take form through categorizing, in which in the next step of data analysis, the researcher will search for patterns in the categories they created. This could be through "patters in words, letters, numbers, phrases, colors, or combinations of these" (9). Moreover, one example I found useful in the reading was summarizing the open ended responses in all capital letters beside the responses, and creating codes that "create a description using categories, themes, settings, or people" (12).

One aspect of this week's reading I found really interesting was the discussion of avoiding bias in data analysis. For example, "cherry picking, [or] making conclusions based on thin (or not enough) data or focusing on data that’s not necessarily representative of the larger dataset (Morse)" (14). The point of collecting data is to find evidence that something is applicable to the entire population, as data relates us all together. However, when only a small part of the story is used, or when only a few select data points are used instead of all the data from the representative sample, the conclusions can be incorrect. The image I included in this blog post illustrates how confirmation bias could play a dangerous role in our data analysis. If we already have expectations for how our data will turn out, we need to make sure we don't overvalue the data that does correspond with our pre-existing beliefs, and undervalue all the other data, that doesn't necessarily confirm our pre-existing beliefs.