00:12:20 Christopher Patterson: doin ok, how are you 00:12:26 Silvana Larrea: Hi Kelsey! Doing okay :) 00:12:34 Jessica Fields (she/her/hers): Having wifi struggles in BWW! 00:12:45 Rachel Harvill: Studying for epi midterm 💀 00:12:51 Chitra Nambiar: celebrating Diwali! 00:13:09 Chitra Nambiar: you got it! thanks! 00:14:05 Sam Holland: I prefer lab first then review plz 00:14:05 Ekua-Yaaba Monkah: lab, then quiz 00:14:06 Phoenix Ding: Yes,please go over last week quiz in the end~ 00:14:06 Jessica Fields (she/her/hers): I’d love to do the lab first and then additional review after : ) 00:14:09 Michele Ko (she/her): I think lab first would be great! :) 00:14:12 Ijeoma Uche: I got an error saying is host is down 00:14:40 Sam Holland: i get that all the time, just refresh data hub ^ 00:16:24 Kelsey MacCuish: did that work Ijeoma? 00:17:27 Chitra Nambiar: same thing - host is down error 00:17:29 Kelsey MacCuish: hmm no its not 00:17:32 Kelsey MacCuish: oops 00:17:47 Kelsey MacCuish: chitra did you try restarting datahub? 00:18:18 Chitra Nambiar: ok, now it works- thanks 00:19:04 Silvana Larrea: prop_calc1 <- 14/38 se1 <- sqrt((prop_calc1*(1 - prop_calc1))/38) me1 <- 1.96*se1 lower_bound <- prop_calc1 - me1 upper_bound <- prop_calc1 + me1 00:19:23 Ekua-Yaaba Monkah: what does normal approximation mean? 00:21:44 Sam Holland: crit value for 95% CI 00:22:42 Ekua-Yaaba Monkah: yes! 00:23:20 Hiruni Jayasekera (she/her): less than 00:23:22 Rachel Harvill: Less than 95% of the time 00:23:22 Ala Koreitem: Lesss than 00:23:24 Ijeoma Uche: How do we know 00:23:39 Ala Koreitem: We’re using the large sample method 00:23:41 Sam Holland: It has poor coverage 00:23:42 Hiruni Jayasekera (she/her): because this method has poor coverage (?) 00:24:23 Leslie Giglio: it contains the true probability p less than 95% of the time? 00:24:41 Hiruni Jayasekera (she/her): because n is small? 00:26:56 Rachel Harvill: We can use prop.test(x = 14, n = 38, conf.level = 0.95) 00:26:57 Julia Hankin: prop.test(x = 14, n = 38, conf.level = 0.95) 00:27:45 Max Billat (any pronouns): why is p=.5? 00:27:48 Hiruni Jayasekera (she/her): ^ 00:28:06 Rachel Harvill: That is our best guess for the population proportion 00:29:39 Silvana Larrea: Kelsey, do we always have to specify the p = “x” in the R function? 00:30:29 Silvana Larrea: Oh okay! Thanks Kelsey! 00:31:23 Hiruni Jayasekera (she/her): it contains the true population proportion 95% of the time? 00:32:41 Ijeoma Uche: p.tilde <- (14+2)/(38+4) se <- sqrt((p.tilde*(1 - p.tilde))/(38+4)) plus_4_ci <-c(p.tilde- 1.96*se, p.tilde + 1.96*se) plus_4_ci 00:36:46 Sam Holland: all good 00:36:49 Jessica Fields (she/her/hers): I’m good to continue forward 00:36:49 Silvana Larrea: All good 00:36:50 Phoenix Ding: All good 00:36:50 Taylor Yoo: im good! 00:40:04 Ijeoma Uche: binom.test(x = 14, n = 38, conf.level = 0.95) exact_method_ci <- round(c( 0.2181250, 0.5400572),4) exact_method_ci 00:41:46 Hiruni Jayasekera (she/her): i’m good! 00:41:46 Annalisa Watson (she/her): Ready to move on 00:43:20 Phoenix Ding: contain the true population proportion more than 95% of the time 00:43:27 Hiruni Jayasekera (she/her): it’s wider 00:44:20 Justin Nguyen: so wider=lower confidence level? 00:45:28 Hiruni Jayasekera (she/her): does plus 4 and wilson score method include the true value more than 95% of the time? 00:45:30 Justin Nguyen: ah got it thanks 00:45:33 Ekua-Yaaba Monkah: yes 00:45:44 Hiruni Jayasekera (she/her): or is that just for the exact method? 00:47:29 Hiruni Jayasekera (she/her): 👍🏾 00:48:10 Ijeoma Uche: Is null value =0? 00:48:15 Ijeoma Uche: the^* 00:49:21 Jessica Fields (she/her/hers): For some reason when I run the code I dont see the data on the graph - any tips? 00:50:04 Silvana Larrea: Why would the point estimate is not the same for each method? Wouldn’t the point estimate be the p hat (14/38) independently of which method are we using to estimate the confidence intervals? 00:50:08 Jessica Fields (she/her/hers): Oooooh - yes thank you! 00:51:52 Hiruni Jayasekera (she/her): are they all the same except for plus 4? (14+2/38+4)? 00:51:58 Jessica Fields (she/her/hers): I put p.hat for all except the plus4 method. For the plus4 method I did p.tilde 00:53:12 Annalisa Watson (she/her): Null hypothesis? 00:53:30 Sam Holland: For some reason I don't have the points on the vertical lines 00:53:31 Annalisa Watson (she/her): We don’t have evidence to reject the null 00:53:31 Lian Hsiao: Is it just me that my graph looks different? (All points on 0 with no vertical lines) 00:54:05 Sam Holland: how do you get the points on the graph? 00:54:26 Ijeoma Uche: Mine still looks werod 00:54:29 Ijeoma Uche: werid 00:54:34 Rachel Harvill: I’m confused - is p6 supposed to be the same thing as step 2 00:54:42 Annalisa Watson (she/her): Yeah I think so rachel 00:54:46 Rachel Harvill: Thx!! 00:55:32 Ala Koreitem: Is y intercept = 50 or 0.50? 00:56:31 Hiruni Jayasekera (she/her): thanks! 00:56:48 Ala Koreitem: Oh I see thank you 00:57:27 Silvana Larrea: Plotting the confidence intervals for each method? 00:58:40 Aliza Adler: Could you go over the inside components again? 00:58:52 Aliza Adler: Like x, xend, etc. 01:00:10 Aliza Adler: Got it! Thanks! 01:00:57 Aliza Adler: So when we want something wider on the x-axis that isn’t a category, what would they be? 01:01:11 Aliza Adler: No prob just curious! 01:03:42 Annalisa Watson (she/her): No, there’s no difference since the CI includes the null of .5 01:04:28 Sam Holland: CI widths decrease? 01:04:44 Sam Holland: Larger sample size is more accurate 01:05:03 Annalisa Watson (she/her): Gets smaller 01:05:04 Sam Holland: decrease 01:05:04 Hiruni Jayasekera (she/her): it goes down? 01:05:41 Taylor Yoo: there is a difference between high stress groups and the normal population 01:06:12 Olufunke Fasawe: can you also go over the question explaining what geom_segment does? 01:06:31 Ijeoma Uche: Large and plus 4 01:07:09 Hiruni Jayasekera (she/her): p tilde 01:08:12 Annalisa Watson (she/her): 0 and 1 01:08:12 Ala Koreitem: 0 01:10:22 Olufunke Fasawe: can you also go over the question explaining what geom_segment does? 01:11:10 Ekua-Yaaba Monkah: do you mind scrolling to p6 please? 01:11:32 Ekua-Yaaba Monkah: thank u 01:13:38 Julia Hankin: Hiii can you remind me how to round the vector to 4 decimal points? 01:13:56 Kelsey MacCuish: yes! You can use round(blah, 4) 01:14:02 Kelsey MacCuish: where blah is your number. hahaha 01:14:09 Julia Hankin: Perfect thank you 01:14:41 Jessica Fields (she/her/hers): If helpful - I renamed things for number 8 “p.hat_low.stress” and things like that 01:14:47 Jessica Fields (she/her/hers): So I wouldn’t overwrite the previous variables 01:15:25 Julia Hankin: Still working on it, but happy to go over it 01:15:28 Olufunke Fasawe: more time 01:15:28 Ijeoma Uche: More time pls 01:15:29 Hiruni Jayasekera (she/her): i’m not done yet 😬 01:20:03 Ijeoma Uche: If the CI is exactly at the null do we still fail to reject 01:20:24 Ijeoma Uche: yeah 01:20:46 Ijeoma Uche: 3/4 01:21:56 Ijeoma Uche: large 01:22:06 Ijeoma Uche: Bad estimate 01:22:11 Ijeoma Uche: poor* 01:23:47 Hiruni Jayasekera (she/her): they all include the null 01:24:24 Cristal Escamilla: I get an error message...Error: unexpected symbol in: " ggplot" 01:24:32 Cristal Escamilla: ggplot(data = sex_CIs_low, aes(x = method, y = estimate)) + geom_point() + geom_hline(aes(yintercept = .50), lty = 2) + geom_segment(aes(x = method, xend = method, y = lower_CI, yend = upper_CI)) + labs(y = "Estimate with 95% CI") 01:24:34 Sam Holland: Support the alternative that there is a difference as CI's would shrink and possibly not contain 0.5 01:26:22 Jessica Fields (she/her/hers): One more sort of random question: for the R functions like prop.test that are running a full hypothesis test, is there a way to “pull” just the lower and upper bounds of the CI? Or is the only thing to do just copy-paste the CI lower and upper bounds from the output? 01:26:30 Silvana Larrea: Yes 01:26:32 Jessica Fields (she/her/hers): (And yes I heard you previously!) 01:26:55 Jessica Fields (she/her/hers): Ok thanks! Yeah that makes sense 01:27:25 Cristal Escamilla: yes 01:28:46 Sam Holland: I am all good and submitted, thank you Kelsey! 01:28:50 Julia Hankin: Can you review the large sample test for #8? 01:28:51 Jessica Fields (she/her/hers): I’m trying to figure out why the dots on an intermediate step are in weird places. Otherwise im good to go 01:28:55 Chitra Nambiar: good to go, thanks 01:29:00 Sam Holland: let me check 01:29:00 Michele Ko (she/her): I got the same thing for 8, thanks Kelsey! 01:29:14 Sam Holland: AUTOGRADER SCORE 4.9999 / 4.9999 01:29:18 Sam Holland: I believe all good! 01:29:26 Sam Holland: Thank you! 01:32:23 Julia Hankin: Yeah!! 01:34:19 Genesis Navarrete: Kelsey, what is the plan for lab next week since its a holiday? Can we attend other labs? 01:34:21 Ijeoma Uche: Quiz questions 01:35:19 Genesis Navarrete: okay sounds good, thank you!