Title
Turnout, campaign visits, and outcomes in the 2016 U.S. presidential election
Journal:
Open Quantitative Sociology & Political Science
Abstract
This study tests two distinct hypothesis regarding the effect of two different factors-turnout and campaign visits-on the results of the 2016 U.S. presidential election at the state level. No evidence was found that either factor was related to the outcome of the election in each U.S. state.
Authors:
Jinkinson Smith
Key words:
(missing)
Length:
(missing)
Files:
https://www.authorea.com/users/225424/articles/284266-testing-hypotheses-about-determinants-of-electoral-outcomes
(OSF missing)
Back to [Archive] Withdrawn submissions
Hi Jinkinson,
Your submission needs some revision to meet the standards. Have a look at a recent publication's submission post:
https://openpsych.net/forum/showthread.php?tid=300&pid=4326#pid4326
I've edited your first to meet the usual standards as far as I could. Please fix the remaining parts, i.e. add proposed keywords and make an OSF repository and put the relevant study files there. http://osf.io/ Remember to set it to "public" when you've uploaded stuff.
---
For the submission, there are some issues.
1. References should use proper academic style. There's no particular rule about which style you prefer, but it's recommended that you follow some standard approach such as APA. You can use Zotero to help you easily manage references in the proper format.
2. In relation to the above, please archive all non-permanent url sources using your preferred method, e.g. archive.org, archive.is, or a saved html copy. This is to avoid link rot and later changes to pages. https://en.wikipedia.org/wiki/Link_rot
3. You should preferably report correlations, not r². r² is a misleading measure of effect size and also is directionless. Try reading e.g. criticism of it in Schmidt and Hunter's book. http://gen.lib.rus.ec/search.php?req=hunter+schmidt+methods&lg_topic=libgen&open=0&view=simple&res=25&phrase=0&column=def
4. You tested 5 things, so theoretically, one might want to correct your hypothesis tests for multiple testing. Of your correlations, none were p < .05, but 2 were close. Of course, since the sample size is so small, you had not much chance to spot any small relationship. For instance, one of your correlations was r = .27, quite a normal sized effect for individual-level research. See e.g. https://psyarxiv.com/2epc4/
5. You only perform bivariate analyses. Consider doing some multiple regressions (can't be done in Excel). It's possible that one effect obscures the effect of another (suppression effect, see e.g. https://openpsych.net/paper/54 for a related study).
I've attempted to replicate your findings here. Not entirely successful. I also added the obvious regression models. Didn't show anything.
http://rpubs.com/EmilOWK/OQSPS_turnout_us_election
Your submission needs some revision to meet the standards. Have a look at a recent publication's submission post:
https://openpsych.net/forum/showthread.php?tid=300&pid=4326#pid4326
I've edited your first to meet the usual standards as far as I could. Please fix the remaining parts, i.e. add proposed keywords and make an OSF repository and put the relevant study files there. http://osf.io/ Remember to set it to "public" when you've uploaded stuff.
---
For the submission, there are some issues.
1. References should use proper academic style. There's no particular rule about which style you prefer, but it's recommended that you follow some standard approach such as APA. You can use Zotero to help you easily manage references in the proper format.
2. In relation to the above, please archive all non-permanent url sources using your preferred method, e.g. archive.org, archive.is, or a saved html copy. This is to avoid link rot and later changes to pages. https://en.wikipedia.org/wiki/Link_rot
3. You should preferably report correlations, not r². r² is a misleading measure of effect size and also is directionless. Try reading e.g. criticism of it in Schmidt and Hunter's book. http://gen.lib.rus.ec/search.php?req=hunter+schmidt+methods&lg_topic=libgen&open=0&view=simple&res=25&phrase=0&column=def
4. You tested 5 things, so theoretically, one might want to correct your hypothesis tests for multiple testing. Of your correlations, none were p < .05, but 2 were close. Of course, since the sample size is so small, you had not much chance to spot any small relationship. For instance, one of your correlations was r = .27, quite a normal sized effect for individual-level research. See e.g. https://psyarxiv.com/2epc4/
5. You only perform bivariate analyses. Consider doing some multiple regressions (can't be done in Excel). It's possible that one effect obscures the effect of another (suppression effect, see e.g. https://openpsych.net/paper/54 for a related study).
I've attempted to replicate your findings here. Not entirely successful. I also added the obvious regression models. Didn't show anything.
http://rpubs.com/EmilOWK/OQSPS_turnout_us_election
Thanks Emil. I'm sorry about the many things in this submission that need to be fixed. I'm new to this and will try to bring the submission up to standards.
Thanks Emil. I'm sorry about the many things in this submission that need to be fixed. I'm new to this and will try to bring the submission up to standards.
I tried to address some of these concerns (specifically 1 2 and 3) and you can view my updated version here: https://www.authorea.com/users/225424/articles/284266-turnout-campaign-visits-and-outcomes-in-the-2016-u-s-presidential-election If any of you think there's anything else I need to change before this can be published please specify it below.
Hi Jinkinson,
Sorry for the wait. I've been a little busy. Feel free to just give me a heads up when you reply on the forum in case I miss it.
Regarding the points.
(1, 2, 3) seems to be okay now.
You didn't reply to my points labeled 4 and 5. Why?
6. There is also my lack of ability to replicate the all findings analytically. This may be a mistake somewhere on my part (coding error), but it needs to be resolved. Please look into it.
7. OP only publishes PDF files (technical limitations). I don't see a way to generate a nicely formatted PDF file from your link. If you want, you can pay my associate to have him give it a pretty formatting like Elsevier journals have (example here: https://openpsych.net/paper/53).
Sorry for the wait. I've been a little busy. Feel free to just give me a heads up when you reply on the forum in case I miss it.
Regarding the points.
(1, 2, 3) seems to be okay now.
You didn't reply to my points labeled 4 and 5. Why?
6. There is also my lack of ability to replicate the all findings analytically. This may be a mistake somewhere on my part (coding error), but it needs to be resolved. Please look into it.
7. OP only publishes PDF files (technical limitations). I don't see a way to generate a nicely formatted PDF file from your link. If you want, you can pay my associate to have him give it a pretty formatting like Elsevier journals have (example here: https://openpsych.net/paper/53).
I did not respond to point 4 because I honestly wasn't sure how to correct for that. But since my last post here I have modified the paper (which you can still read at the same Authorea link above) to account for multiple hypothesis testing. Lastly, with respect to point 5, I don't know how to use any programs other than Excel (I'm still learning how to use R). I did try to re-do the same analyses using the Analysis ToolPak in Excel as described here: https://www.wikihow.com/Run-a-Multiple-Regression-in-Excel
I will work on addressing points 6 and 7 soon.
I will work on addressing points 6 and 7 soon.
I more recently did multiple regressions in Excel using the Analysis ToolPak as I noted above. The results of them are now included in the updated authorea document: https://www.authorea.com/users/225424/articles/284266-turnout-campaign-visits-and-outcomes-in-the-2016-u-s-presidential-election
I am looking at your attempted replication of my results and it looks like you did replicate them insofar as you tried to test the same things: for example the correlation between turnout and MOV was .27 for both my and your analysis, and that between turnout and shift was .137 in mine and .14 in yours (presumably a rounding difference). But the correlation between VAI and MOV was -.03679 in mine and .072 in yours, apparently because I included all 50 states and it looks like you only included 26 for some reason? W/respect to PDF I think I should be able to rework the authorea document so the PDF will look right (maybe removing the charts).
I am looking at your attempted replication of my results and it looks like you did replicate them insofar as you tried to test the same things: for example the correlation between turnout and MOV was .27 for both my and your analysis, and that between turnout and shift was .137 in mine and .14 in yours (presumably a rounding difference). But the correlation between VAI and MOV was -.03679 in mine and .072 in yours, apparently because I included all 50 states and it looks like you only included 26 for some reason? W/respect to PDF I think I should be able to rework the authorea document so the PDF will look right (maybe removing the charts).
Ugh. I'm not excited about doing this, but I'd be even less excited if I didn't. I no longer feel comfortable associating myself with OpenPsych and therefore would like this submission to be withdrawn. I originally submitted it in the hope that I would be able to increase the credibility of this journal and cause it to be seen as better than a hereditarian echo chamber, but I am now highly doubtful that this can be successful. Also, I don't want my reputation to be damaged (additionally at least) from associating with any other members of this forum. I think I should try to submit my work to another journal, where I don't have to rationalize associating myself with the editors by telling myself I don't agree with everything they think or say. I appreciate Emil's feedback on this post.
Thread moved to withdrawn submissions.
© Copyright 2024 OpenPsych - All Rights Reserved