class: center, middle, inverse, title-slide .title[ # 3: Causation ] .subtitle[ ## Empirical Methods ] .author[ ### <large>J. Seawright</large> ] .institute[ ### <small>Northwestern Political Science</small> ] .date[ ### Sept. 23, 2025 ] --- class: center, middle <style type="text/css"> pre { max-height: 400px; overflow-y: auto; } pre[class] { max-height: 200px; } </style> ### How do citizen attitudes affect the survival of democracy? --- <img src="images/Chiopris.png" width="90%" style="display: block; margin: auto;" /> --- ### Causation <div class="grViz html-widget html-fill-item" id="htmlwidget-0792687314a550c1d396" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-0792687314a550c1d396">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.55918416199594,0.38376043367175!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.384223437385372,0.0421808891390072!\"] \n \"1\"->\"2\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Reverse Causation <div class="grViz html-widget html-fill-item" id="htmlwidget-38c726d285d13ae67abb" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-38c726d285d13ae67abb">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.558724090930535,0.616611669243774!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.0391835218668896,-0.189194268509953!\"] \n \"2\"->\"1\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Reciprocal Causation <div class="grViz html-widget html-fill-item" id="htmlwidget-3670aa7b232d9c5886d0" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-3670aa7b232d9c5886d0">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.200637003044044,0.0148655304511225!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.459547679327897,0.448868709026677!\"] \n \"1\"->\"2\" \n \"2\"->\"1\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Confounding <div class="grViz html-widget html-fill-item" id="htmlwidget-ef0ed4240deb0ac4d55f" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-ef0ed4240deb0ac4d55f">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.328069154070969,-1.09394922754298!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.425624413954447,-0.0948224004332598!\"] \n \"3\" [label = \"Social Media\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-1.24211599833985,-0.678871147344458!\"] \n \"1\"->\"2\" \n \"3\"->\"2\" \n \"3\"->\"1\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Post-Treatment <div class="grViz html-widget html-fill-item" id="htmlwidget-7fc76893a8d6b0c3ff08" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-7fc76893a8d6b0c3ff08">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.779383953075031,-0.947338950413368!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.876998389792383,0.0462369939956215!\"] \n \"3\" [label = \"Favorability\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"-0.0322708360934191,-0.366014395845813!\"] \n \"1\"->\"2\" \n \"3\"->\"2\" \n \"1\"->\"3\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Instrument <div class="grViz html-widget html-fill-item" id="htmlwidget-ac4883dced3febddff53" style="width:80%;height:504px;"></div> <script type="application/json" data-for="htmlwidget-ac4883dced3febddff53">{"x":{"diagram":"digraph {\n\ngraph [layout = \"neato\",\n outputorder = \"edgesfirst\",\n bgcolor = \"white\"]\n\nnode [fontname = \"Helvetica\",\n fontsize = \"10\",\n shape = \"circle\",\n fixedsize = \"true\",\n width = \"0.5\",\n style = \"filled\",\n fillcolor = \"aliceblue\",\n color = \"gray70\",\n fontcolor = \"gray50\"]\n\nedge [fontname = \"Helvetica\",\n fontsize = \"8\",\n len = \"1.5\",\n color = \"gray80\",\n arrowsize = \"0.5\"]\n\n \"1\" [label = \"Uncertainty\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.389208741037654,-0.55137112033844!\"] \n \"2\" [label = \"Backsliding\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.17712607295922,0.572912200957822!\"] \n \"3\" [label = \"Country Size\", fillcolor = \"#F0F8FF\", fontcolor = \"#000000\", pos = \"0.602013596355721,-1.67948287128656!\"] \n \"1\"->\"2\" \n \"3\"->\"1\" \n}","config":{"engine":"dot","options":null}},"evals":[],"jsHooks":[]}</script> --- ### Fundamental Problem of Causal Inference We can't usually tell by looking at data which of the graphs we just looked at reflects the real world, so we can't tell if we have a causal understanding or a distortion. --- ### Fundamental Problem of Causal Inference There are some questions we can ask to reason about that help us pinpoint what we know, or don't yet know, about a causal relationship. --- ### The Four Questions 1. Is there a relationship between the treatment and the outcome? --- <img src="images/Chiopris2.png" width="90%" style="display: block; margin: auto;" /> --- ### The Four Questions 1. Is there a relationship between the treatment and the outcome? 2. Could the outcome cause the treatment? --- <img src="images/Chiopris3.png" width="90%" style="display: block; margin: auto;" /> --- ### The Four Questions 1. Is there a relationship between the treatment and the outcome? 2. Could the outcome cause the treatment? 3. Is there evidence of a causal pathway from the treatment to the outcome? --- In our study, not too much; there isn't that much data about how people translated views about the policy into decisions. --- ### The Four Questions 1. Is there a relationship between the treatment and the outcome? 2. Could the outcome cause the treatment? 3. Is there evidence of a causal pathway from the treatment to the outcome? 4. Have confounding variables been ruled out? --- <img src="images/Chiopris4.png" width="70%" style="display: block; margin: auto;" /> --- ### Controlling for Confounders **Controlling for a confounder** is choosing a set of cases that all have the same score on the confounding variable and looking at the relationship between the treatment and the outcome in only those cases. --- ### Problems with Controlling for Confounders - You can only control for variables that you actually think of controlling for. - If you control for too many variables, you may not have very many cases left in each category. --- ### Do Mask Mandates Work? --- <img src="images/maskexp.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/maskexp2.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/maskexp3.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/masknews1.jpg" width="70%" style="display: block; margin: auto;" /> --- <img src="images/masknews2.jpg" width="70%" style="display: block; margin: auto;" /> --- <img src="images/maskcases.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/healthaffairs1.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/healthaffairs2.jpg" width="70%" style="display: block; margin: auto;" /> --- <img src="images/healthaffairs3.jpg" width="65%" style="display: block; margin: auto;" /> --- <img src="images/cepr1.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/cepr2.jpg" width="90%" style="display: block; margin: auto;" /> --- <img src="images/cepr3.jpg" width="90%" style="display: block; margin: auto;" />