Epssa methods workshop overly honest mixed methods - 4-11-13

#overlyhonestmethods:Mixed Methods step-by-step

Cat BiddleEPSSA Methods Workshop

April 11th, 2013

Img source: http://runt-of-the-web.com/best-overly-honest-methods

The project

What role does place play in defining school purpose?

• Tradition of local control of schooling in America • Neo-institutional theory suggests normative,

coercive and mimetic forces may play a role in shaping school purpose

• Schools are local/national institutions – so, who/what determines school purpose?

Data and definitions

• Data = school district mission statements

• Place = urbanicity using NCES locale codes

Method selection

• Content analysis of district mission statements-Data must be durable in nature-used to understand specified characteristic of messages

• Emergent coding structure: – Randomly selected 25 mission statements and coded

them by hand, compared our coding and developed initial coding structure

– Coded 50 mission statements by hand using initial coding schematic, reconciled any differences through more discussion of what codes signified

Coding

Using Nvivo in MM

Step 1: Import spreadsheet

-External Data Dataset

-Differentiate between codable and classifying fields in your spreadsheet (this creates attributes)

-Create nodes for each “case” in your spreadsheet (this is just a good idea in any project) through

Using Nvivo in MM

Using Nvivo in MM

Step 2: Code the data and reconcile coding divergence-Merge Kai’s coding file with my version

-Compare Kai’s and my coding by using the “Show coding stripes” function (and selecting “users”)

-View Coding Stripes “Selected Items” Users select desired codes

Using Nvivo in MM

Using Nvivo in MM

• Matrix queries allowed us to see preliminary results that led us to pursue more analysis

• Exported a spreadsheet of a Matrix Query of all districts for all codes (with presence of code = 1 and absence of code = 0)

• Save query right click on it and “Export list” to Excel

Using Nvivo in MM work

Method Selection

• Chi-square tests of bivariate association – By urbanicity

• Binary logistic regression using CCD– Coded themes as dichotomous outcome variables– Variables included students eligible for

free/reduced price lunch, PSSA score of advanced on reading and math, % non-white students*, district drop out rate*, and student-teacher ratio

*log-transformed to correct for non-normal distribution

Responding to reviewers

• Questions about the unit of analysis led us to collect additional data

• Randomly selected – 75 elementary schools– 75 middle schools – 75 high schools

Compared these mission statements (if they existed) to the district mission statement