Bioequivalence and Bioavailability Forum • Rounding or not to rounding

Rounding or not to rounding [🇷 for BE/BA]

posted by yjlee168 – Kaohsiung, Taiwan, 2010-05-04 23:27 (5880 d 06:25 ago) – Posting: # 5287
Views: 30,506

Dear Helmut and D. Labes,

Sorry about this. I didn't notice that this thread was about bear at the beginning until I saw the edited footnote by Helmut. Thanks you all for all messages here.

❝ Just discovered a little problem. bear log-transforms raw values and rounds them to three significant digits. That's nasty, [...]

Yes. Then I tested bear (v2.5.0, not released yet!) to compare the runs between the pre-transformed ~~manually~~ Cmax to 3 digits and internally transformed with the model (lnCmax<- lm(log(Cmax) ~ seq + subj:seq + prd + drug , data=TotalData)) as suggested by D. Labes, using a 2x2x2 crossover demo dataset in bear (single-dosed). I got the same results as follows. I think two methods are taking the same values to calculate under R, not manually pre-transformed values. That's why no difference is observed.

- Pre-transformed ~~manually~~---

  Statistical analysis (ANOVA(lm))    

--------------------------------------------------------------------------

  Dependent Variable: lnCmax                                              



Type I SS

Analysis of Variance Table



Response: lnCmax

          Df   Sum Sq  Mean Sq F value Pr(>F)

seq        1 0.000690 0.000690  0.0388 0.8472

prd        1 0.018169 0.018169  1.0205 0.3323

drug       1 0.036238 0.036238  2.0354 0.1792

subj(seq) 12 0.283676 0.023640  1.3278 0.3155

Residuals 12 0.213641 0.017803               



Type III SS

Single term deletions



Model:

lnCmax ~ seq + subj:seq + prd + drug

          Df Sum of Sq     RSS     AIC F value  Pr(F)

<none>                 0.21364 -104.52               

prd        1  0.018169 0.23181 -104.23  1.0205 0.3323

drug       1  0.036238 0.24988 -102.13  2.0354 0.1792

subj(seq) 12  0.283676 0.49732 -104.86  1.3278 0.3155



Tests of Hypothesis for SUBJECT(SEQUENCE) as an error term



Error: subj

          Df  Sum Sq   Mean Sq F value Pr(>F)

prd:drug   1 0.00069 0.0006903  0.0292 0.8672

Residuals 12 0.28368 0.0236397               



Error: Within

          Df   Sum Sq  Mean Sq F value Pr(>F)

prd        1 0.018169 0.018169  1.0205 0.3323

drug       1 0.036238 0.036238  2.0354 0.1792

Residuals 12 0.213641 0.017803               



Intra_subj. CV = 100*sqrt(abs(exp(MSResidual)-1)) = 13.4025 %

Inter_subj. CV = 100*sqrt(abs(exp((MSSubject(seq)-MSResidual)/2)-1)) = 5.4059 %

    MSResidual = 0.01780339 

MSSubject(seq) = 0.0236397 

[...]
**************** Classical (Shortest) 90% C.I. for lnCmax ****************



  Point_estimate CI90_lower CI90_upper

1        107.460     98.223    117.566

[...]

--- internally transformed method ---

  Statistical analysis (ANOVA(lm))    

--------------------------------------------------------------------------

  Dependent Variable: lnCmax                                              



Type I SS

--- internally transformed method ---

Analysis of Variance Table



Response: log(Cmax)

          Df   Sum Sq  Mean Sq F value Pr(>F)

seq        1 0.000690 0.000690  0.0388 0.8472

prd        1 0.018169 0.018169  1.0205 0.3323

drug       1 0.036238 0.036238  2.0354 0.1792

subj(seq) 12 0.283676 0.023640  1.3278 0.3155

Residuals 12 0.213641 0.017803               



Type III SS

Single term deletions



Model:

log(Cmax) ~ seq + subj:seq + prd + drug

          Df Sum of Sq     RSS     AIC F value  Pr(F)

<none>                 0.21364 -104.52               

prd        1  0.018169 0.23181 -104.23  1.0205 0.3323

drug       1  0.036238 0.24988 -102.13  2.0354 0.1792

subj(seq) 12  0.283676 0.49732 -104.86  1.3278 0.3155



Tests of Hypothesis for SUBJECT(SEQUENCE) as an error term



Error: subj

          Df  Sum Sq   Mean Sq F value Pr(>F)

prd:drug   1 0.00069 0.0006903  0.0292 0.8672

Residuals 12 0.28368 0.0236397               



Error: Within

          Df   Sum Sq  Mean Sq F value Pr(>F)

prd        1 0.018169 0.018169  1.0205 0.3323

drug       1 0.036238 0.036238  2.0354 0.1792

Residuals 12 0.213641 0.017803               



Intra_subj. CV = 100*sqrt(abs(exp(MSResidual)-1)) = 13.4025 %

Inter_subj. CV = 100*sqrt(abs(exp((MSSubject(seq)-MSResidual)/2)-1)) = 5.4059 %

    MSResidual = 0.01780339 

MSSubject(seq) = 0.0236397               

[...]
**************** Classical (Shortest) 90% C.I. for lnCmax ****************



  Point_estimate CI90_lower CI90_upper

1        107.460     98.223    117.566

[...]

—
All the best,
-- Yung-jin Lee
bear v2.9.6:- created by Hsin-ya Lee & Yung-jin Lee
Kaohsiung, Taiwan https://www.pkpd168.com/bear
Download link (updated) -> here

Complete thread:

F value for sequence effect Pakawadee 2010-05-03 11:48 [🇷 for BE/BA]
- F value for sequence effect AA 2010-05-03 13:38
- F value for sequence effect Ohlbe 2010-05-03 14:16
  - F value for sequence effect Pakawadee 2010-05-04 03:48
    - F value for sequence effect ElMaestro 2010-05-04 08:48
      - F value for sequence effect Pakawadee 2010-05-04 12:12
      - F value; rounding Helmut 2010-05-04 13:00
        
        log rounding machine d_labes 2010-05-04 13:35
        Rounding or not to roundingyjlee168 2010-05-04 21:27
        
        Rounding or not to rounding ElMaestro 2010-05-04 23:23
        
        Rounding or not to rounding yjlee168 2010-05-05 01:17
        
        Rounding or not to rounding ElMaestro 2010-05-05 12:15
        
        Rounding or not to rounding yjlee168 2010-05-05 12:59
        
        Rounding or not to rounding ElMaestro 2010-05-05 13:15
        
        Rounding or not to rounding yjlee168 2010-05-05 13:28
        Type III SS sequence effect -revised yjlee168 2010-05-05 23:58
        
        Type III SS sequence effect -revised ElMaestro 2010-05-06 07:27
        
        SAS-like Type III SS sequence with bear? yjlee168 2010-06-03 13:04
        
        New dataset? d_labes 2010-05-05 11:40
        
        New dataset? yjlee168 2010-05-05 12:31
        
        log to 3 decimals guess work d_labes 2010-05-05 16:00
        
        log to 3 decimals guess work yjlee168 2010-05-05 20:13
        
        rounding and presentation of data Helmut 2010-05-05 20:29
        
        rounding and presentation of data yjlee168 2010-05-05 21:05
        
        rounding and presentation of data in R martin 2010-05-06 07:53
        
        Canadian logs d_labes 2010-05-06 10:12
        
        Canadian logs ElMaestro 2010-05-06 10:21
        Canadian logs Helmut 2010-05-06 14:57
        
        Canadian lost d_labes 2010-05-06 15:38