Collection of STATA scripts to aid in the analysis of survey data on consumer financial health:
The survey weight process in these scripts allows for a quick build, but still imposing rigor so as to avoid the weighting optimization landing on a local maxima.
The project was built with intent that beginner analysts can reuse scripts in a semi-automated fashion, to be later reviewed with subject matter expert in survey analysis and weighting design.
Critical steps of the process include:
- take survey responses and cleanse/standardize the data
- apply a financial health score to each respondent
- construct survey weights to assure a representative sample
- output data-dredged crosstabs.
| # | Script location | Description | Analyst Touch Points | Analyst action items for project team | Inputs? |
|---|---|---|---|---|---|
| Step 1 | Scripts\Step1_TTSurveyStandardization.do | Initial data processing of survey monkey data. File is imported, values are converted to numeric with labels, nulls are reported, & nulls are dropped | Analyst needs to provide: import command for Survey file, define the names of the variables. | Report survey nulls to project lead. | Survey Data |
| Step 2 | Scripts\Step2_TTSurveyStandardization.do | Create finhealth typing. Results from step 1 are typed into one of seven finhealth segments | None | Analyst needs to confirm that typing steps contain no errors as defined on worksheet "Typing Steps QC". Any errors should be consulted with data manager and corrected. Report summary of confidence % for typing. | "Outputs\Step1.dta" |
| Step 3 | Scripts\Step3_TTSurveyStandardization.do | Save benchmark file to ".dta" file type and rename, recode, or relabel any necessary labels. | Analyst needs to provide: import command for benchmark file, define the names of the variables, assigns codes & labels to text fields. | Report benchmark file nulls to project lead. | Benchmark Data |
| Step 4 | Scripts\Step4_TTSurveyStandardization.do | Create survey weights. | Analyst must define product and demographic variable to be used for weighting. | Analyst needs to provide reccomendation to project lead about which weight is best. Analyst must interpret results from the 7 default weight options. Differences in Finhealth outcomes should be a red flag that the weighting has peculiarities. | "Outputs\Step2.dta" & "Outputs\Step3.dta" |
| Step 5 | Scripts\Step5_TTSurveyStandardization.do | Create outputs for preliminary results report | None | Analyst must combine/format results according to project lead specification. Analyst should review results for errors, inconsistencies, interesting data points. | "Outputs/Step4.dta" |
| Step 6 | Scripts\Step6_TTSurveyStandardization.do | Creates banners | Analyst must define which variables are to be used for columns, and which variables are to be used for worksheets. | QC a few results by hand. Send results to project team. | "Outputs/Step4.dta" |
| Step 7 | Scripts\Step7_TTSurveyStandardization.do | Create charts for PPT deck | Analyst must define which variables are to be used in charts that compare CFHS to Survey. Analyst must define which CFHS metric to use (eg. West Banked). | QC a few charts by hand. Hit the refresh button inside PPT to update the charts. | "Outputs/Step4.dta" & "CFHS/CFHSDataSet.dta" |
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent