47:
68:
is that the collider blocks the association between the variables that influence it. Thus, the collider does not generate an unconditional association between the variables that determine it.
244:
Ali, R. Ayesha; Richardson, Thomas S.; Spirtes, Peter; Zhange, Jiji (2012). "Towards characterizing Markov equivalence classes for directed acyclic graphs with latent variables".
270:
445:
192:
389:
75:, stratification, experimental design, or sample selection based on values of the collider creates a non-causal association between
57:
The causal variables influencing the collider are themselves not necessarily associated. If they are not adjacent, the collider is
103:, potentially introducing associations where there are none. Colliders can therefore undermine attempts to test causal theories.
50:
158:
65:
455:
450:
335:
148:
118:
440:
31:
when it is causally influenced by two or more variables. The name "collider" reflects the fact that in
84:
340:
209:"Thinking Clearly About Correlations and Causation: Graphical Causal Models for Observational Data"
72:
36:
124:
Randomization and quasi-experimental research designs are not useful in overcoming collider bias.
249:
412:
309:
301:
293:
246:
Proceedings of the Twenty-First
Conference on Uncertainty in Artificial Intelligence (UAI2006)
188:
187:, Chapman & Hall/CRC monographs on statistics & applied probability, CRC, p. 70,
404:
345:
285:
224:
216:
87:). In the terminology of causal graphs, conditioning on the collider opens the path between
32:
35:, the arrow heads from variables that lead into the collider appear to "collide" on the
153:
434:
368:
349:
289:
229:
138:
24:
46:
363:
163:
143:
370:
Probabilistic reasoning in intelligent systems: networks of plausible inference
326:
Pearl, Judea (1986). "Fusion, Propagation and
Structuring in Belief Networks".
408:
107:
20:
416:
305:
297:
420:
220:
133:
313:
95:. This will introduce bias when estimating the causal association between
269:
Greenland, Sander; Pearl, Judea; Robins, James M. (January 1999),
254:
45:
117:
To detect and manage collider bias, scholars have made use of
61:. Otherwise, the collider is shielded and part of a triangle.
208:
39:
that is the collider. They are sometimes also referred to as
16:
Variable that is causally influenced by two or more variables
114:
be controlled for when estimating causal associations.
367:
110:variables. Unlike colliders, confounder variables
8:
390:"Collider bias in economic history research"
271:"Causal Diagrams for Epidemiologic Research"
183:Hernan, Miguel A.; Robins, James M. (2010),
339:
253:
228:
175:
64:The result of having a collider in the
106:Colliders are sometimes confused with
7:
383:
381:
14:
446:Independence (probability theory)
71:Conditioning on the collider via
397:Explorations in Economic History
290:10.1097/00001648-199901000-00008
207:Julia M. Rohrer (2018-07-02).
1:
350:10.1016/0004-3702(86)90072-x
388:Schneider, Eric B. (2020).
472:
409:10.1016/j.eeh.2020.101356
230:21.11116/0000-0006-5734-E
328:Artificial Intelligence
119:directed acyclic graphs
149:Directed acyclic graph
54:
221:10.31234/osf.io/t3qub
49:
73:regression analysis
53:model of a collider
423:on April 11, 2024.
374:. Morgan Kaufmann.
55:
27:, a variable is a
194:978-1-4200-7616-5
85:Berkson's paradox
463:
456:Causal inference
451:Graphical models
425:
424:
419:. Archived from
394:
385:
376:
375:
373:
360:
354:
353:
343:
323:
317:
316:
275:
266:
260:
259:
257:
241:
235:
234:
232:
204:
198:
197:
185:Causal inference
180:
33:graphical models
471:
470:
466:
465:
464:
462:
461:
460:
431:
430:
429:
428:
392:
387:
386:
379:
362:
361:
357:
325:
324:
320:
273:
268:
267:
263:
243:
242:
238:
206:
205:
201:
195:
182:
181:
177:
172:
130:
17:
12:
11:
5:
469:
467:
459:
458:
453:
448:
443:
433:
432:
427:
426:
377:
355:
341:10.1.1.84.8016
334:(3): 241–288.
318:
261:
236:
199:
193:
174:
173:
171:
168:
167:
166:
161:
156:
154:Selection bias
151:
146:
141:
136:
129:
126:
41:inverted forks
15:
13:
10:
9:
6:
4:
3:
2:
468:
457:
454:
452:
449:
447:
444:
442:
439:
438:
436:
422:
418:
414:
410:
406:
402:
398:
391:
384:
382:
378:
372:
371:
365:
359:
356:
351:
347:
342:
337:
333:
329:
322:
319:
315:
311:
307:
303:
299:
295:
291:
287:
283:
279:
272:
265:
262:
256:
251:
247:
240:
237:
231:
226:
222:
218:
214:
210:
203:
200:
196:
190:
186:
179:
176:
169:
165:
162:
160:
159:Path analysis
157:
155:
152:
150:
147:
145:
142:
140:
139:Causal graphs
137:
135:
132:
131:
127:
125:
122:
120:
115:
113:
109:
104:
102:
98:
94:
90:
86:
82:
78:
74:
69:
67:
62:
60:
52:
48:
44:
42:
38:
34:
30:
26:
25:causal graphs
22:
441:Epidemiology
421:the original
400:
396:
369:
364:Pearl, Judea
358:
331:
327:
321:
284:(1): 37–48,
281:
278:Epidemiology
277:
264:
245:
239:
212:
202:
184:
178:
123:
116:
111:
105:
100:
96:
92:
88:
80:
76:
70:
63:
58:
56:
40:
28:
18:
164:Bad control
144:Confounding
435:Categories
403:: 101356.
170:References
108:confounder
59:unshielded
21:statistics
417:0014-4983
336:CiteSeerX
306:484244020
298:1044-3983
255:1207.1365
248:: 10–17.
134:Causality
366:(1988).
213:PsyArXiv
128:See also
29:collider
314:9888278
415:
338:
312:
304:
296:
191:
112:should
393:(PDF)
274:(PDF)
250:arXiv
413:ISSN
310:PMID
302:OCLC
294:ISSN
189:ISBN
99:and
91:and
79:and
66:path
37:node
23:and
405:doi
346:doi
286:doi
225:hdl
217:doi
51:SEM
19:In
437::
411:.
401:78
399:.
395:.
380:^
344:.
332:29
330:.
308:,
300:,
292:,
282:10
280:,
276:,
223:.
215:.
211:.
121:.
43:.
407::
352:.
348::
288::
258:.
252::
233:.
227::
219::
101:Y
97:X
93:Y
89:X
83:(
81:Y
77:X
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
