Re: [R] Reporting binomial logistic regression from R results
Dear Bert, I understand and thanks for your recommendation. Unfortunately I do not have any possibility to contact a statistical expert at the moment. So this forum experts' recommendation would be crucial to me to understand how R works in relation to my question. I hope that someone could reply to my last questions. Best regards FJ On Mon, Nov 12, 2018 at 7:48 PM Bert Gunter wrote: > Generally speaking, this list is about questions on R programming, not > statistical issues. However, I grant you that your queries are in something > of a gray area intersecting both. > > Nevertheless, based on your admitted confusion, I would recommend that you > find a local statistical expert with whom you can consult 1-1 if at all > possible. As others have already noted, you statistical understanding is > muddy, and it can be quite difficult to resolve such confusion in online > forums like this that cannot provide the close back and forth that may be > required (as well as further appropriate study). > > Best, > Bert > > On Mon, Nov 12, 2018 at 11:09 AM Frodo Jedi < > frodojedi.mailingl...@gmail.com> wrote: > >> Dear Peter and Eik, >> I am very grateful to you for your replies. >> My current understanding is that from the GLM analysis I can indeed >> conclude that the response predicted by System A is significantly >> different >> from that of System B, while the pairwise comparison A vs C leads to non >> significance. Now the Wald test seems to be correct only for Systems B vs >> C, indicating that the pairwise System B vs System C is significant. Am I >> correct? >> >> However, my current understanding is also that I should use contrasts >> instead of the wald test. So the default contrasts is with the System A, >> now I should re-perform the GLM with another base. I tried to use the >> option "contrasts" of the glm: >> >> > fit1 <- glm(Response ~ System, data = scrd, family = "binomial", >> contrasts = contr.treatment(3, base=1,contrasts=TRUE)) >> > summary(fit1) >> >> > fit2 <- glm(Response ~ System, data = scrd, family = "binomial", >> contrasts = contr.treatment(3, base=2,contrasts=TRUE)) >> > summary(fit2) >> >> > fit3 <- glm(Response ~ System, data = scrd, family = "binomial", >> contrasts = contr.treatment(3, base=3,contrasts=TRUE)) >> > summary(fit3) >> >> However, the output of these three summary functions are identical. Why? >> That option should have changed the base, but apparently this is not the >> case. >> >> >> Another analysis I found online (at this link >> >> https://stats.stackexchange.com/questions/60352/comparing-levels-of-factors-after-a-glm-in-r >> ) >> to understand the differences between the 3 levels is to use glth with >> Tuckey. I performed the following: >> >> > library(multcomp) >> > summary(glht(fit, mcp(System="Tukey"))) >> >> Simultaneous Tests for General Linear Hypotheses >> >> Multiple Comparisons of Means: Tukey Contrasts >> >> >> Fit: glm(formula = Response ~ System, family = "binomial", data = scrd) >> >> Linear Hypotheses: >> Estimate Std. Error z value Pr(>|z|) >> B - A == 0 -1.2715 0.3379 -3.763 0.000445 *** >> C - A == 00.8588 0.4990 1.721 0.192472 >> C - B == 0 2.1303 0.4512 4.722 < 1e-04 *** >> --- >> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 >> (Adjusted p values reported -- single-step method) >> >> >> Is this Tukey analysis correct? >> >> >> I am a bit confused on what analysis I should do. I am doing my very best >> to study all resources I can find, but I would really need some help from >> experts, especially in using R. >> >> >> Best wishes >> >> FJ >> >> >> >> >> >> >> On Mon, Nov 12, 2018 at 1:46 PM peter dalgaard wrote: >> >> > Yes, only one of the pairwise comparisons (B vs. C) is right. Also, the >> > overall test has 3 degrees of freedom whereas a comparison of 3 groups >> > should have 2. You (meaning Frodo) are testing that _all 3_ regression >> > coefficients are zero, intercept included. That would imply that all >> three >> > systems have response probablilities og 0.5, which is not likely what >> you >> > want. >> > >> > This all suggests that you are struggling with the interpretation of the >> > regression coefficients and their role in the linear predictor. This >> should >> > be covered by any good book on logistic regression. >> > >> > -pd >> > >> > > On 12 Nov 2018, at 14:15 , Eik Vettorazzi >> wrote: >> > > >> > > Dear Jedi, >> > > please use the source carefully. A and C are not statistically >> different >> > at the 5% level, which can be inferred from glm output. Your last two >> > wald.tests don't test what you want to, since your model contains an >> > intercept term. You specified contrasts which tests A vs B-A, ie A- >> > (B-A)==0 <-> 2*A-B==0 which is not intended I think. Have a look at >> > ?contr.treatment and re-read your source doc to get an idea what dummy >> > coding and indicatr variables are about. >> > > >> > > Cheers >> > > >> > > >> > > Am
Re: [R] Reporting binomial logistic regression from R results
Generally speaking, this list is about questions on R programming, not statistical issues. However, I grant you that your queries are in something of a gray area intersecting both. Nevertheless, based on your admitted confusion, I would recommend that you find a local statistical expert with whom you can consult 1-1 if at all possible. As others have already noted, you statistical understanding is muddy, and it can be quite difficult to resolve such confusion in online forums like this that cannot provide the close back and forth that may be required (as well as further appropriate study). Best, Bert On Mon, Nov 12, 2018 at 11:09 AM Frodo Jedi wrote: > Dear Peter and Eik, > I am very grateful to you for your replies. > My current understanding is that from the GLM analysis I can indeed > conclude that the response predicted by System A is significantly different > from that of System B, while the pairwise comparison A vs C leads to non > significance. Now the Wald test seems to be correct only for Systems B vs > C, indicating that the pairwise System B vs System C is significant. Am I > correct? > > However, my current understanding is also that I should use contrasts > instead of the wald test. So the default contrasts is with the System A, > now I should re-perform the GLM with another base. I tried to use the > option "contrasts" of the glm: > > > fit1 <- glm(Response ~ System, data = scrd, family = "binomial", > contrasts = contr.treatment(3, base=1,contrasts=TRUE)) > > summary(fit1) > > > fit2 <- glm(Response ~ System, data = scrd, family = "binomial", > contrasts = contr.treatment(3, base=2,contrasts=TRUE)) > > summary(fit2) > > > fit3 <- glm(Response ~ System, data = scrd, family = "binomial", > contrasts = contr.treatment(3, base=3,contrasts=TRUE)) > > summary(fit3) > > However, the output of these three summary functions are identical. Why? > That option should have changed the base, but apparently this is not the > case. > > > Another analysis I found online (at this link > > https://stats.stackexchange.com/questions/60352/comparing-levels-of-factors-after-a-glm-in-r > ) > to understand the differences between the 3 levels is to use glth with > Tuckey. I performed the following: > > > library(multcomp) > > summary(glht(fit, mcp(System="Tukey"))) > > Simultaneous Tests for General Linear Hypotheses > > Multiple Comparisons of Means: Tukey Contrasts > > > Fit: glm(formula = Response ~ System, family = "binomial", data = scrd) > > Linear Hypotheses: > Estimate Std. Error z value Pr(>|z|) > B - A == 0 -1.2715 0.3379 -3.763 0.000445 *** > C - A == 00.8588 0.4990 1.721 0.192472 > C - B == 0 2.1303 0.4512 4.722 < 1e-04 *** > --- > Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > (Adjusted p values reported -- single-step method) > > > Is this Tukey analysis correct? > > > I am a bit confused on what analysis I should do. I am doing my very best > to study all resources I can find, but I would really need some help from > experts, especially in using R. > > > Best wishes > > FJ > > > > > > > On Mon, Nov 12, 2018 at 1:46 PM peter dalgaard wrote: > > > Yes, only one of the pairwise comparisons (B vs. C) is right. Also, the > > overall test has 3 degrees of freedom whereas a comparison of 3 groups > > should have 2. You (meaning Frodo) are testing that _all 3_ regression > > coefficients are zero, intercept included. That would imply that all > three > > systems have response probablilities og 0.5, which is not likely what you > > want. > > > > This all suggests that you are struggling with the interpretation of the > > regression coefficients and their role in the linear predictor. This > should > > be covered by any good book on logistic regression. > > > > -pd > > > > > On 12 Nov 2018, at 14:15 , Eik Vettorazzi wrote: > > > > > > Dear Jedi, > > > please use the source carefully. A and C are not statistically > different > > at the 5% level, which can be inferred from glm output. Your last two > > wald.tests don't test what you want to, since your model contains an > > intercept term. You specified contrasts which tests A vs B-A, ie A- > > (B-A)==0 <-> 2*A-B==0 which is not intended I think. Have a look at > > ?contr.treatment and re-read your source doc to get an idea what dummy > > coding and indicatr variables are about. > > > > > > Cheers > > > > > > > > > Am 12.11.2018 um 02:07 schrieb Frodo Jedi: > > >> Dear list members, > > >> I need some help in understanding whether I am doing correctly a > > binomial > > >> logistic regression and whether I am interpreting the results in the > > >> correct way. Also I would need an advice regarding the reporting of > the > > >> results from the R functions. > > >> I want to report the results of a binomial logistic regression where I > > want > > >> to assess difference between the 3 levels of a factor (called System) > on > > >> the dependent variable (called Response) taking two values,
Re: [R] Reporting binomial logistic regression from R results
Dear Peter and Eik, I am very grateful to you for your replies. My current understanding is that from the GLM analysis I can indeed conclude that the response predicted by System A is significantly different from that of System B, while the pairwise comparison A vs C leads to non significance. Now the Wald test seems to be correct only for Systems B vs C, indicating that the pairwise System B vs System C is significant. Am I correct? However, my current understanding is also that I should use contrasts instead of the wald test. So the default contrasts is with the System A, now I should re-perform the GLM with another base. I tried to use the option "contrasts" of the glm: > fit1 <- glm(Response ~ System, data = scrd, family = "binomial", contrasts = contr.treatment(3, base=1,contrasts=TRUE)) > summary(fit1) > fit2 <- glm(Response ~ System, data = scrd, family = "binomial", contrasts = contr.treatment(3, base=2,contrasts=TRUE)) > summary(fit2) > fit3 <- glm(Response ~ System, data = scrd, family = "binomial", contrasts = contr.treatment(3, base=3,contrasts=TRUE)) > summary(fit3) However, the output of these three summary functions are identical. Why? That option should have changed the base, but apparently this is not the case. Another analysis I found online (at this link https://stats.stackexchange.com/questions/60352/comparing-levels-of-factors-after-a-glm-in-r ) to understand the differences between the 3 levels is to use glth with Tuckey. I performed the following: > library(multcomp) > summary(glht(fit, mcp(System="Tukey"))) Simultaneous Tests for General Linear Hypotheses Multiple Comparisons of Means: Tukey Contrasts Fit: glm(formula = Response ~ System, family = "binomial", data = scrd) Linear Hypotheses: Estimate Std. Error z value Pr(>|z|) B - A == 0 -1.2715 0.3379 -3.763 0.000445 *** C - A == 00.8588 0.4990 1.721 0.192472 C - B == 0 2.1303 0.4512 4.722 < 1e-04 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 (Adjusted p values reported -- single-step method) Is this Tukey analysis correct? I am a bit confused on what analysis I should do. I am doing my very best to study all resources I can find, but I would really need some help from experts, especially in using R. Best wishes FJ On Mon, Nov 12, 2018 at 1:46 PM peter dalgaard wrote: > Yes, only one of the pairwise comparisons (B vs. C) is right. Also, the > overall test has 3 degrees of freedom whereas a comparison of 3 groups > should have 2. You (meaning Frodo) are testing that _all 3_ regression > coefficients are zero, intercept included. That would imply that all three > systems have response probablilities og 0.5, which is not likely what you > want. > > This all suggests that you are struggling with the interpretation of the > regression coefficients and their role in the linear predictor. This should > be covered by any good book on logistic regression. > > -pd > > > On 12 Nov 2018, at 14:15 , Eik Vettorazzi wrote: > > > > Dear Jedi, > > please use the source carefully. A and C are not statistically different > at the 5% level, which can be inferred from glm output. Your last two > wald.tests don't test what you want to, since your model contains an > intercept term. You specified contrasts which tests A vs B-A, ie A- > (B-A)==0 <-> 2*A-B==0 which is not intended I think. Have a look at > ?contr.treatment and re-read your source doc to get an idea what dummy > coding and indicatr variables are about. > > > > Cheers > > > > > > Am 12.11.2018 um 02:07 schrieb Frodo Jedi: > >> Dear list members, > >> I need some help in understanding whether I am doing correctly a > binomial > >> logistic regression and whether I am interpreting the results in the > >> correct way. Also I would need an advice regarding the reporting of the > >> results from the R functions. > >> I want to report the results of a binomial logistic regression where I > want > >> to assess difference between the 3 levels of a factor (called System) on > >> the dependent variable (called Response) taking two values, 0 and 1. My > >> goal is to understand if the effect of the 3 systems (A,B,C) in System > >> affect differently Response in a significant way. I am basing my > analysis > >> on this URL: https://stats.idre.ucla.edu/r/dae/logit-regression/ > >> This is the result of my analysis: > >>> fit <- glm(Response ~ System, data = scrd, family = "binomial") > >>> summary(fit) > >> Call: > >> glm(formula = Response ~ System, family = "binomial", data = scrd) > >> Deviance Residuals: > >> Min 1Q Median 3Q Max > >> -2.8840 0.1775 0.2712 0.2712 0.5008 > >> Coefficients: > >> Estimate Std. Error z value Pr(>|z|) > >> (Intercept)3.2844 0.2825 11.626 < 2e-16 *** > >> SystemB -1.2715 0.3379 -3.763 0.000168 *** > >> SystemC0.8588 0.4990 1.721 0.085266 . > >> --- > >> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05
Re: [R] Reporting binomial logistic regression from R results
Yes, only one of the pairwise comparisons (B vs. C) is right. Also, the overall test has 3 degrees of freedom whereas a comparison of 3 groups should have 2. You (meaning Frodo) are testing that _all 3_ regression coefficients are zero, intercept included. That would imply that all three systems have response probablilities og 0.5, which is not likely what you want. This all suggests that you are struggling with the interpretation of the regression coefficients and their role in the linear predictor. This should be covered by any good book on logistic regression. -pd > On 12 Nov 2018, at 14:15 , Eik Vettorazzi wrote: > > Dear Jedi, > please use the source carefully. A and C are not statistically different at > the 5% level, which can be inferred from glm output. Your last two wald.tests > don't test what you want to, since your model contains an intercept term. You > specified contrasts which tests A vs B-A, ie A- (B-A)==0 <-> 2*A-B==0 which > is not intended I think. Have a look at ?contr.treatment and re-read your > source doc to get an idea what dummy coding and indicatr variables are about. > > Cheers > > > Am 12.11.2018 um 02:07 schrieb Frodo Jedi: >> Dear list members, >> I need some help in understanding whether I am doing correctly a binomial >> logistic regression and whether I am interpreting the results in the >> correct way. Also I would need an advice regarding the reporting of the >> results from the R functions. >> I want to report the results of a binomial logistic regression where I want >> to assess difference between the 3 levels of a factor (called System) on >> the dependent variable (called Response) taking two values, 0 and 1. My >> goal is to understand if the effect of the 3 systems (A,B,C) in System >> affect differently Response in a significant way. I am basing my analysis >> on this URL: https://stats.idre.ucla.edu/r/dae/logit-regression/ >> This is the result of my analysis: >>> fit <- glm(Response ~ System, data = scrd, family = "binomial") >>> summary(fit) >> Call: >> glm(formula = Response ~ System, family = "binomial", data = scrd) >> Deviance Residuals: >> Min 1Q Median 3Q Max >> -2.8840 0.1775 0.2712 0.2712 0.5008 >> Coefficients: >> Estimate Std. Error z value Pr(>|z|) >> (Intercept)3.2844 0.2825 11.626 < 2e-16 *** >> SystemB -1.2715 0.3379 -3.763 0.000168 *** >> SystemC0.8588 0.4990 1.721 0.085266 . >> --- >> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 >> (Dispersion parameter for binomial family taken to be 1) >> Null deviance: 411.26 on 1023 degrees of freedom >> Residual deviance: 376.76 on 1021 degrees of freedom >> AIC: 382.76 >> Number of Fisher Scoring iterations: 6 >> Following this analysis I perform the wald test in order to understand >> whether there is an overall effect of System: >> library(aod) >>> wald.test(b = coef(fit), Sigma = vcov(fit), Terms = 1:3) >> Wald test: >> -- >> Chi-squared test: >> X2 = 354.6, df = 3, P(> X2) = 0.0 >> The chi-squared test statistic of 354.6, with 3 degrees of freedom is >> associated with a p-value < 0.001 indicating that the overall effect of >> System is statistically significant. >> Now I check whether there are differences between the coefficients using >> again the wald test: >> # Here difference between system B and C: >>> l <- cbind(0, 1, -1) >>> wald.test(b = coef(fit), Sigma = vcov(fit), L = l) >> Wald test: >> -- >> Chi-squared test: >> X2 = 22.3, df = 1, P(> X2) = 2.3e-06 >> # Here difference between system A and C: >>> l <- cbind(1, 0, -1) >>> wald.test(b = coef(fit), Sigma = vcov(fit), L = l) >> Wald test: >> -- >> Chi-squared test: >> X2 = 12.0, df = 1, P(> X2) = 0.00052 >> # Here difference between system A and B: >>> l <- cbind(1, -1, 0) >>> wald.test(b = coef(fit), Sigma = vcov(fit), L = l) >> Wald test: >> -- >> Chi-squared test: >> X2 = 58.7, df = 1, P(> X2) = 1.8e-14 >> My understanding is that from this analysis I can state that the three >> systems lead to a significantly different Response. Am I right? If so, how >> should I report the results of this analysis? What is the correct way? >> Thanks in advance >> Best wishes >> FJ >> [[alternative HTML version deleted]] >> __ >> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see >> https://stat.ethz.ch/mailman/listinfo/r-help >> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html >> and provide commented, minimal, self-contained, reproducible code. > > -- > Eik Vettorazzi > > Department of Medical Biometry and Epidemiology > University Medical Center Hamburg-Eppendorf > > Martinistrasse 52 > building W 34 > 20246 Hamburg > > Phone: +49 (0) 40 7410 - 58243 > Fax: +49 (0) 40 7410 - 57790 > Web: www.uke.de/imbe > -- > > _ > >
Re: [R] Reporting binomial logistic regression from R results
Dear Jedi, please use the source carefully. A and C are not statistically different at the 5% level, which can be inferred from glm output. Your last two wald.tests don't test what you want to, since your model contains an intercept term. You specified contrasts which tests A vs B-A, ie A- (B-A)==0 <-> 2*A-B==0 which is not intended I think. Have a look at ?contr.treatment and re-read your source doc to get an idea what dummy coding and indicatr variables are about. Cheers Am 12.11.2018 um 02:07 schrieb Frodo Jedi: Dear list members, I need some help in understanding whether I am doing correctly a binomial logistic regression and whether I am interpreting the results in the correct way. Also I would need an advice regarding the reporting of the results from the R functions. I want to report the results of a binomial logistic regression where I want to assess difference between the 3 levels of a factor (called System) on the dependent variable (called Response) taking two values, 0 and 1. My goal is to understand if the effect of the 3 systems (A,B,C) in System affect differently Response in a significant way. I am basing my analysis on this URL: https://stats.idre.ucla.edu/r/dae/logit-regression/ This is the result of my analysis: fit <- glm(Response ~ System, data = scrd, family = "binomial") summary(fit) Call: glm(formula = Response ~ System, family = "binomial", data = scrd) Deviance Residuals: Min 1Q Median 3Q Max -2.8840 0.1775 0.2712 0.2712 0.5008 Coefficients: Estimate Std. Error z value Pr(>|z|) (Intercept)3.2844 0.2825 11.626 < 2e-16 *** SystemB -1.2715 0.3379 -3.763 0.000168 *** SystemC0.8588 0.4990 1.721 0.085266 . --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 (Dispersion parameter for binomial family taken to be 1) Null deviance: 411.26 on 1023 degrees of freedom Residual deviance: 376.76 on 1021 degrees of freedom AIC: 382.76 Number of Fisher Scoring iterations: 6 Following this analysis I perform the wald test in order to understand whether there is an overall effect of System: library(aod) wald.test(b = coef(fit), Sigma = vcov(fit), Terms = 1:3) Wald test: -- Chi-squared test: X2 = 354.6, df = 3, P(> X2) = 0.0 The chi-squared test statistic of 354.6, with 3 degrees of freedom is associated with a p-value < 0.001 indicating that the overall effect of System is statistically significant. Now I check whether there are differences between the coefficients using again the wald test: # Here difference between system B and C: l <- cbind(0, 1, -1) wald.test(b = coef(fit), Sigma = vcov(fit), L = l) Wald test: -- Chi-squared test: X2 = 22.3, df = 1, P(> X2) = 2.3e-06 # Here difference between system A and C: l <- cbind(1, 0, -1) wald.test(b = coef(fit), Sigma = vcov(fit), L = l) Wald test: -- Chi-squared test: X2 = 12.0, df = 1, P(> X2) = 0.00052 # Here difference between system A and B: l <- cbind(1, -1, 0) wald.test(b = coef(fit), Sigma = vcov(fit), L = l) Wald test: -- Chi-squared test: X2 = 58.7, df = 1, P(> X2) = 1.8e-14 My understanding is that from this analysis I can state that the three systems lead to a significantly different Response. Am I right? If so, how should I report the results of this analysis? What is the correct way? Thanks in advance Best wishes FJ [[alternative HTML version deleted]] __ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Eik Vettorazzi Department of Medical Biometry and Epidemiology University Medical Center Hamburg-Eppendorf Martinistrasse 52 building W 34 20246 Hamburg Phone: +49 (0) 40 7410 - 58243 Fax: +49 (0) 40 7410 - 57790 Web: www.uke.de/imbe -- _ Universitätsklinikum Hamburg-Eppendorf; Körperschaft des öffentlichen Rechts; Gerichtsstand: Hamburg | www.uke.de Vorstandsmitglieder: Prof. Dr. Burkhard Göke (Vorsitzender), Prof. Dr. Dr. Uwe Koch-Gromus, Joachim Prölß, Marya Verdel _ SAVE PAPER - THINK BEFORE PRINTING __ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Reporting binomial logistic regression from R results
Dear Petr, thank you very much for your feedback. Can anyone in the list advise me if the way I report the results is correct? Kind regards FJ On Mon, Nov 12, 2018 at 1:02 PM PIKAL Petr wrote: > Hi Frodo > > > > I do not consider myself as an arbiter in statistical results and their > presentation. Again your text seems to as good as any other. > > > > You should keep responses to mailing list as others could have another > opinion. > > > > Cheers > > Petr > > > > > > *From:* Frodo Jedi > *Sent:* Monday, November 12, 2018 1:48 PM > *To:* PIKAL Petr > *Subject:* Re: [R] Reporting binomial logistic regression from R results > > > > Dear Petr, > > many thanks for your reply. I was wondering whether in your opinion it is > correct to report in a journal the following text: > > > > > > “A logistic regression was performed to ascertain the effects of the > system type on the likelihood that participants report correct > identifications. The logistic regression model was statistically > significant, χ2(3) = 354.6, p < 0.001, indicating an overall effect of the > system type on participants' identification performances. The Wald test was > used to compare the model coefficients related to the three systems. > Results showed that participants’ accuracy was significantly lower for the > system B compared to both the system C (χ2(1) = 22.3, p < 0.001) and the > system A (χ2(1) = 58.7, p < 0.001), as well as that the system C led to > significantly higher identification accuracies than the system A (χ2(1) = > 12, p < 0.001).” > > > > > > Best wishes > > > > FJ > > > > > > > > > > > > On Mon, Nov 12, 2018 at 10:05 AM PIKAL Petr > wrote: > > Dear Frodo (or Jedi) > > The results seems to confirm your assumption that 3 systems are different. > How you should present results probably depends on how it is usual to > report such results in your environment. > > BTW. It seems to me like homework and this list has no homework policy > (Sorry, if I am mistaken). > > Cheers > Petr > > -Original Message- > > From: R-help On Behalf Of Frodo Jedi > > Sent: Monday, November 12, 2018 2:08 AM > > To: r-help@r-project.org > > Subject: [R] Reporting binomial logistic regression from R results > > > > Dear list members, > > I need some help in understanding whether I am doing correctly a binomial > > logistic regression and whether I am interpreting the results in the > correct way. > > Also I would need an advice regarding the reporting of the results from > the R > > functions. > > > > I want to report the results of a binomial logistic regression where I > want to > > assess difference between the 3 levels of a factor (called System) on the > > dependent variable (called Response) taking two values, 0 and 1. My goal > is to > > understand if the effect of the 3 systems (A,B,C) in System affect > differently > > Response in a significant way. I am basing my analysis on this URL: > > https://stats.idre.ucla.edu/r/dae/logit-regression/ > > > > This is the result of my analysis: > > > > > fit <- glm(Response ~ System, data = scrd, family = "binomial") > > > summary(fit) > > > > Call: > > glm(formula = Response ~ System, family = "binomial", data = scrd) > > > > Deviance Residuals: > > Min 1Q Median 3Q Max > > -2.8840 0.1775 0.2712 0.2712 0.5008 > > > > Coefficients: > > Estimate Std. Error z value Pr(>|z|) > > (Intercept)3.2844 0.2825 11.626 < 2e-16 *** > > SystemB -1.2715 0.3379 -3.763 0.000168 *** > > SystemC0.8588 0.4990 1.721 0.085266 . > > --- > > Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > > > > (Dispersion parameter for binomial family taken to be 1) > > > > Null deviance: 411.26 on 1023 degrees of freedom Residual deviance: > > 376.76 on 1021 degrees of freedom > > AIC: 382.76 > > > > Number of Fisher Scoring iterations: 6 > > Following this analysis I perform the wald test in order to understand > whether > > there is an overall effect of System: > > > > library(aod) > > > > > wald.test(b = coef(fit), Sigma = vcov(fit), Terms = 1:3) > > Wald test: > > -- > > > > Chi-squared test: > > X2 = 354.6, df = 3, P(> X2) = 0.0 > > The chi-squared test statistic of 354.6, with 3 degrees of freedom is > associated > > with a p-value < 0.001 indica
Re: [R] Reporting binomial logistic regression from R results
Hi Frodo I do not consider myself as an arbiter in statistical results and their presentation. Again your text seems to as good as any other. You should keep responses to mailing list as others could have another opinion. Cheers Petr From: Frodo Jedi Sent: Monday, November 12, 2018 1:48 PM To: PIKAL Petr Subject: Re: [R] Reporting binomial logistic regression from R results Dear Petr, many thanks for your reply. I was wondering whether in your opinion it is correct to report in a journal the following text: “A logistic regression was performed to ascertain the effects of the system type on the likelihood that participants report correct identifications. The logistic regression model was statistically significant, χ2(3) = 354.6, p < 0.001, indicating an overall effect of the system type on participants' identification performances. The Wald test was used to compare the model coefficients related to the three systems. Results showed that participants’ accuracy was significantly lower for the system B compared to both the system C (χ2(1) = 22.3, p < 0.001) and the system A (χ2(1) = 58.7, p < 0.001), as well as that the system C led to significantly higher identification accuracies than the system A (χ2(1) = 12, p < 0.001).” Best wishes FJ On Mon, Nov 12, 2018 at 10:05 AM PIKAL Petr mailto:petr.pi...@precheza.cz>> wrote: Dear Frodo (or Jedi) The results seems to confirm your assumption that 3 systems are different. How you should present results probably depends on how it is usual to report such results in your environment. BTW. It seems to me like homework and this list has no homework policy (Sorry, if I am mistaken). Cheers Petr > -Original Message- > From: R-help > mailto:r-help-boun...@r-project.org>> On Behalf > Of Frodo Jedi > Sent: Monday, November 12, 2018 2:08 AM > To: r-help@r-project.org<mailto:r-help@r-project.org> > Subject: [R] Reporting binomial logistic regression from R results > > Dear list members, > I need some help in understanding whether I am doing correctly a binomial > logistic regression and whether I am interpreting the results in the correct > way. > Also I would need an advice regarding the reporting of the results from the R > functions. > > I want to report the results of a binomial logistic regression where I want to > assess difference between the 3 levels of a factor (called System) on the > dependent variable (called Response) taking two values, 0 and 1. My goal is to > understand if the effect of the 3 systems (A,B,C) in System affect differently > Response in a significant way. I am basing my analysis on this URL: > https://stats.idre.ucla.edu/r/dae/logit-regression/ > > This is the result of my analysis: > > > fit <- glm(Response ~ System, data = scrd, family = "binomial") > > summary(fit) > > Call: > glm(formula = Response ~ System, family = "binomial", data = scrd) > > Deviance Residuals: > Min 1Q Median 3Q Max > -2.8840 0.1775 0.2712 0.2712 0.5008 > > Coefficients: > Estimate Std. Error z value Pr(>|z|) > (Intercept)3.2844 0.2825 11.626 < 2e-16 *** > SystemB -1.2715 0.3379 -3.763 0.000168 *** > SystemC0.8588 0.4990 1.721 0.085266 . > --- > Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > > (Dispersion parameter for binomial family taken to be 1) > > Null deviance: 411.26 on 1023 degrees of freedom Residual deviance: > 376.76 on 1021 degrees of freedom > AIC: 382.76 > > Number of Fisher Scoring iterations: 6 > Following this analysis I perform the wald test in order to understand whether > there is an overall effect of System: > > library(aod) > > > wald.test(b = coef(fit), Sigma = vcov(fit), Terms = 1:3) > Wald test: > -- > > Chi-squared test: > X2 = 354.6, df = 3, P(> X2) = 0.0 > The chi-squared test statistic of 354.6, with 3 degrees of freedom is > associated > with a p-value < 0.001 indicating that the overall effect of System is > statistically > significant. > > Now I check whether there are differences between the coefficients using again > the wald test: > > # Here difference between system B and C: > > > l <- cbind(0, 1, -1) > > wald.test(b = coef(fit), Sigma = vcov(fit), L = l) > Wald test: > -- > > Chi-squared test: > X2 = 22.3, df = 1, P(> X2) = 2.3e-06 > > > > # Here difference between system A and C: > > > l <- cbind(1, 0, -1) > > wald.test(b = coef(fit), Sigma = vcov(fit), L = l) > Wald test: > -- > > Chi-squared test: > X2 = 12.0, df = 1, P(> X2) = 0.00052 > > > > # Here difference between system A and
Re: [R] Reporting binomial logistic regression from R results
Dear Frodo (or Jedi) The results seems to confirm your assumption that 3 systems are different. How you should present results probably depends on how it is usual to report such results in your environment. BTW. It seems to me like homework and this list has no homework policy (Sorry, if I am mistaken). Cheers Petr > -Original Message- > From: R-help On Behalf Of Frodo Jedi > Sent: Monday, November 12, 2018 2:08 AM > To: r-help@r-project.org > Subject: [R] Reporting binomial logistic regression from R results > > Dear list members, > I need some help in understanding whether I am doing correctly a binomial > logistic regression and whether I am interpreting the results in the correct > way. > Also I would need an advice regarding the reporting of the results from the R > functions. > > I want to report the results of a binomial logistic regression where I want to > assess difference between the 3 levels of a factor (called System) on the > dependent variable (called Response) taking two values, 0 and 1. My goal is to > understand if the effect of the 3 systems (A,B,C) in System affect differently > Response in a significant way. I am basing my analysis on this URL: > https://stats.idre.ucla.edu/r/dae/logit-regression/ > > This is the result of my analysis: > > > fit <- glm(Response ~ System, data = scrd, family = "binomial") > > summary(fit) > > Call: > glm(formula = Response ~ System, family = "binomial", data = scrd) > > Deviance Residuals: > Min 1Q Median 3Q Max > -2.8840 0.1775 0.2712 0.2712 0.5008 > > Coefficients: > Estimate Std. Error z value Pr(>|z|) > (Intercept)3.2844 0.2825 11.626 < 2e-16 *** > SystemB -1.2715 0.3379 -3.763 0.000168 *** > SystemC0.8588 0.4990 1.721 0.085266 . > --- > Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > > (Dispersion parameter for binomial family taken to be 1) > > Null deviance: 411.26 on 1023 degrees of freedom Residual deviance: > 376.76 on 1021 degrees of freedom > AIC: 382.76 > > Number of Fisher Scoring iterations: 6 > Following this analysis I perform the wald test in order to understand whether > there is an overall effect of System: > > library(aod) > > > wald.test(b = coef(fit), Sigma = vcov(fit), Terms = 1:3) > Wald test: > -- > > Chi-squared test: > X2 = 354.6, df = 3, P(> X2) = 0.0 > The chi-squared test statistic of 354.6, with 3 degrees of freedom is > associated > with a p-value < 0.001 indicating that the overall effect of System is > statistically > significant. > > Now I check whether there are differences between the coefficients using again > the wald test: > > # Here difference between system B and C: > > > l <- cbind(0, 1, -1) > > wald.test(b = coef(fit), Sigma = vcov(fit), L = l) > Wald test: > -- > > Chi-squared test: > X2 = 22.3, df = 1, P(> X2) = 2.3e-06 > > > > # Here difference between system A and C: > > > l <- cbind(1, 0, -1) > > wald.test(b = coef(fit), Sigma = vcov(fit), L = l) > Wald test: > -- > > Chi-squared test: > X2 = 12.0, df = 1, P(> X2) = 0.00052 > > > > # Here difference between system A and B: > > > l <- cbind(1, -1, 0) > > wald.test(b = coef(fit), Sigma = vcov(fit), L = l) > Wald test: > -- > > Chi-squared test: > X2 = 58.7, df = 1, P(> X2) = 1.8e-14 > > My understanding is that from this analysis I can state that the three systems > lead to a significantly different Response. Am I right? If so, how should I > report > the results of this analysis? What is the correct way? > > > Thanks in advance > > Best wishes > > FJ > > [[alternative HTML version deleted]] > > __ > R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. Osobní údaje: Informace o zpracování a ochraně osobních údajů obchodních partnerů PRECHEZA a.s. jsou zveřejněny na: https://www.precheza.cz/zasady-ochrany-osobnich-udaju/ | Information about processing and protection of business partner’s personal data are available on website: https://www.precheza.cz/en/personal-data-protection-principles/ Důvěrnost: Tento e-mail a jakékoliv k němu připojené dokumenty jsou důvěrné a podléhají tomuto právně závaznému prohláąení o vyloučení odpovědnosti: https://www.precheza.cz/01-dovetek/ | This email and any documents attached to it may be confidential and are subject to the legally binding disclaimer: https://www.precheza.cz/en/01-disclaimer/ __ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
