===Special Interest Groups Quality Control=== If time and resources allow, there is a lot of work that can be done to clean the SIGs portion of the database. There are a few perpetual issues with SIGs that have compounded over the years, mostly due to the guidelines being followed inconsistently. Cleaning rating and endorsement data can vary in time and resources needed. It is good practice to do weekly quality control scans on common errors and problems to minimize the need for large-scale QC projects in the future. <<__Regular QC Scans__ - Check candidates and officials on the live site - Blank ratings - Unreleased ratings/endorsements/SIGs - Correct rating format - Extra characters in our_rating - Letter Grades with '+' & '-' << >>__Other QC ideas__ - Rating name spelling - Special characters in SIG description - State affiliates - Action fund, education fund, etc. duplicates - People endorsed under the wrong election - Standardizing historic NRA data across years>> //- Making sure all SIG's Release Statuses are live that should be: Any non-blank fields under release_status need to be further investigated. Look for notes in the SIG Description field to see if there is a reason we should not release that interest group. If there is no reason that you can find as to why it shouldn't be on the live web, release it. - Making sure all Ratings' Release Statuses are live that should be: Check for notes in the SIG Description field to see if there is a reason we should not release that interest group's ratings. If this rating is fairly new and is not done with checks, DO NOT RELEASE IT. If it looks like there's a problem with it and shouldn't be released, talk to another person about it. If there is no reason that you can find as to why the ratings shouldn't be on the live web, release them. Webcheck the ratings and make sure they appear. -Making sure all Endorsements' Release Statuses are live that should be: Check for notes in the SIG Description field to see if there is a reason we should not release that interest group's endorsements. If there is no reason that you can find as to why the endorsements shouldn't be on the live web, release them. Webcheck the endorsements and make sure they display.// There are several QC queries already written in the network drive. Using pre-written queries is a good starting point for QC projects. Note: In 2013 there was a large undertaking to examine all of the scorecards in our archives. There isn't much documentation on this project and it looks like they never finished the project. If you run into incomplete work (most likely found in the backlog boxes) with red and green tags you can find more information in the SIGs network drive folder under Quality Control > 2013 Clean-Up, Red&Green labels. Back to [[RatingsGuide|SIGs home]] ++This project requires every original rating scorecard in our database be reevaluated using the instructions below. While there are hard copies of many ratings scorecards in the archives, some are missing. We have outlined instructions for tracking old ratings down in the Ratings Clean Up Guide, but it may not be possible to find every missing scorecard. The whole point of this project is to ensure that we are conveying the correct ratings data in the clearest way possible to our users. This project requires that every scorecard be revisited and reevaluated for the conditions listed below. Federal ratings should be prioritized over state, but every scorecard in the archives will eventually need to be examined. For each individual scorecard, use the instructions outlined below. On paper scorecards check and make note of: 1. Key included 2. Methodology included 3. Voting record roll call with bill summaries if scores based on voting record 4. Questions asked if scores are based on survey 5. Type of scores: % (0-100, 0-100+), grade (and range ie. A to F, A+ to F- etc., open, string). 6. SIG math if SIG has numerical score (spot check 2-3 scores) 7. Are scores weighted, curved, are there bonuses/penalties 8. Are Absences, N/A, Excused, Passed, No Votes, or similar counted 9. Are there endorsements, if endorsements are on separate paper copy, pull those from folder so we can put into new folder specifically of SIG endorsements 10. See if SIG site has electronic copies of scorecard. If not, check SIG site @ http://archive.org and see if there is an archived version of scorecard. Save any found to pez:/public/research/ratings scorecards in specific folder (may need to create folders for new state/SIG. Download any new ratings to: pez:/public/research/ratings scorecards/0new. Some SIGS may split House and Senate ratings, be sure to specify in file name. If using archive.org to download old ratings found, use Firebug in Firefox to remove the upper archive.org site banner at the top of the page. 1. Install the Firebug add-on in Firefox if not already installed @ https://addons.mozilla.org/en-US/firefox/addon/firebug/?src=hp-dl-mostpopular. 2. To remove the archive.org banner, click on the Firebug button located at the lower-right of the Firefox window, which displays the HTML source of the page. 3. Click on the + sign to the left of the tag. 4. Click on the first line under the which should be a
tag to highlight it. 5. Use the delete key to delete the line, which should remove the archive.org banner at the top of the page. 6. Print the scorecard as a PDF file into the correct folder on pez:/public. 7. If you inadvertently delete the wrong line with Firebug, reload the page and repeat the process to delete the banner. In admin check and make note of: 1. Rating type. % scores should be type numeric, strings, open (non % points system, discrete rating choices with no letter grades) 2. For particular rating, on votesmart.org confirm that strings and letter grades are converted to numerical scores. If not, uncheck 'use sig rating' in admin. Note open type ratings. Groups like Planned Parenthood with ratings of pro-choice, mixed choice, and anti-choice should have scores of 100, 50, 0 respectively. Where groups rate on a positive/negative scale, unrelease rating in admin for now, highlight rating in yellow in section below describing Google Docs file. 3. Spot check 2-3 ratings to make sure they match up with actual scorecard, paying attention to strings where SIG's preferred positions not necessarily corresponding to official's votes in favor of particular legislation. 4. Verify that rating scorecard isn't repeated in admin On Google Docs tracking sheet: 1. Use 11022011_list_of_sig_and_ratings file 2. Enter in details in column Q: "comments", and column R: "in archive" using ! for 'no' or 'not' ex. if no electronic formatted scorecard found use "!E", if no methodology/key found use "!methodology/key", etc. 3. For % ratings with scores > 100, note in "comments" column with 100+. These will may have to be converted to a score type rating instead of %. 4. Note any ratings that need revisiting for reentering of data, or other issues in yellow. Note any new ratings, or ratings not yet entered into admin with teal. Note possible duplicate ratings with magenta. Note ratings to be ignored and unreleased with gray (currently only Gay Rights Information, as this is not a real SIG, just some random person with a website). Note in orange any rating where SIG's stance on grading is not clear and we will need to discuss standards for rating to clearly represent scores (ex. http://www.ethanallen.org/html/report_cards.html). 5. If particular rating is listed on this spreadsheet, without a paper copy, and an electronic version is found, use a P in column R to signify that we need to print the rating for our paper archive.++