Answer to exercise: Blood pressure lowering drugs

Size: px
Start display at page:

Download "Answer to exercise: Blood pressure lowering drugs"

Transcription

1 Answer to exercise: Blood pressure lowering drugs The data set bloodpressure.txt contains data from a cross-over trial, involving three different formulations of a drug for lowering of blood pressure: A: 50 mg tablet B: 100 mg tablet C: Sustained-release formulation capsule A total of 12 male volunteers were randomly divided into three groups (group=1,2,3), and the groups received each of the three formulations, but in different sequences (and with a wash-out period of one week), according to the scheme below: 1. A-B-C 2. B-C-A 3. C-A-B The outcome for assessing the effect of each treatment was the duration of the drug, in hours. 1. Read in the data and make pictures to illustrate the possible effects. Immediately after reading in the data, we redefine group as the new variable sequence to be able to keep track of the order of treatments in each group. Moreover, we calculate some summary statistics to get an impression of the data: data a1; infile " URL firstobs=2; input id group treatment$ period duration; if group=1 then sequence= ABC ; if group=2 then sequence= BCA ; if group=3 then sequence= CAB ; proc means nway N mean stddev data=a1; class sequence treatment period; var duration; output out=average mean=mean_duration; 1

2 which provides us with the output Analysis Variable : duration N sequence treatment period Obs N Mean Std Dev ABC A B C BCA A B C CAB A B C Note the design: For each group=sequence, we have only 3 of the 9 possible combinations of treatment*period. Next, we make two panels of series plots: one where we plot against treatment and one against period, both subdivided according to sequence (group). proc sort data=a1; by sequence id treatment; proc sgpanel noautolegend data=a1; panelby sequence / rows=1; series Y=duration X=treatment / group=id; proc sort data=a1; by sequence id period; proc sgpanel noautolegend data=a1; panelby sequence / rows=1; series Y=duration X=period / group=id; 2

3 There seems to be a reasonable consistent tendency in the treatment effect, but no obvious consistent pattern in the period effect. If we average over individuals, we get proc sort data=average; by sequence period; proc sgplot data=average; series Y=mean_duration X=period / group=sequence; proc sort data=average; by sequence treatment; 3

4 proc sgplot data=average; series Y=mean_duration X=treatment / group=sequence; From the graph on the left hand side, we clearly see an effect of treatment, although the red sequence (BCA) behaves differently from the other two for treatment A (last treatment) where the duration is somewhat prolonged. Could it be a carry-over effect? Treatment C is seen to have the largest effect, and this is perhaps not very surprising due to the nature of this (sustained-release), followed by the high-dose B. The right hand graph is not that clear, probably because the treatment effects obscure the possible effect of period. Of course, we hope that there is no effect of period at all. Perhaps it is clearer, if we do not divide according to sequence, but rather according to either treatment or period, i.e. proc sort data=average; by treatment period; proc sgplot data=average; series Y=mean_duration X=period / group=treatment; 4

5 proc sort data=average; by period treatment; proc sgplot data=average; series Y=mean_duration X=treatment / group=period; Note, however, that in these plots, the lines do not correspond to time effects for any individual, since the groups are mixed up. In the left hand figure, the pattern for period 1 is clearly to say that C is the best treatment, followed by B and then A. The same conclusion holds for the two other periods, although the pattern here is not as clear. In the right hand side picture, we see a tendency for higher values in later periods, corresponding to a sort of cumulative effect over time, i.e. a plausible carry-over effect. 2. Which factors could determine the duration of the drug? And which of these can be considered systematic? And random? We have three factors that may have an influence on the outcome: The treatment, the period and the subject. The two first of these must be 5

6 considered systematic, whereas the subject may be considered random. We might also think of a possible interaction between treatment and period, in the form of a carry-over effect. Actually, this concept is a bit tricky here, with three periods and three treatments, since we may in principle have a different carry-over effect from treatment A to treatment B than e.g. from treatment A to treatment C. To complicate matters, we may even have a different carry-over effect, if a treatment is given after two other treatments. At first, we disregard possible carry-over effects. 3. Test equality of mean durations from the three formulations/treatments, taking into account a possible period effect. Also, take the correlation between measurements on the same individual into account by including a random effect of individual. We can analyze the data in a mixed model, with systematic effect of the factors period and treatment, and a random effect of subject id: proc mixed data=a1; class sequence id period treatment; model duration = period treatment / ddfm=kr s cl; random intercept / subject=id v vcorr; estimate 100 mg vs 50 mg treatment / cl; which gives us Class Level Information Class Levels Values sequence 3 ABC BCA CAB id period treatment 3 A B C Dimensions Covariance Parameters 2 Columns in X 7 Columns in Z Per Subject 1 Subjects 12 Max Obs Per Subject 3 6

7 Number of Observations Number of Observations Read 36 Number of Observations Used 36 Number of Observations Not Used 0 Estimated V Matrix for id Estimated V Correlation Matrix for id Covariance Parameter Estimates Cov Parm Subject Estimate Intercept id Residual Fit Statistics -2 Res Log Likelihood 49.7 AIC (smaller is better) 53.7 AICC (smaller is better) 54.2 BIC (smaller is better) 54.7 Solution for Fixed Effects Effect treatment period Estimate Error DF t Value Pr > t Intercept <.0001 period period period treatment A <.0001 treatment B treatment C Effect treatment period Alpha Lower Upper Intercept period period period 3... treatment A treatment B treatment C... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F period treatment <

8 Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg Label Lower Upper 100 mg vs 50 mg We note a highly significant difference between the treatments (P < ), but unfortunately also a highly significant increase in duration with time (period, P = ). We also note, that the correlation between measurements taken on the same individual is This correlation is assumed to be the same no matter which two observations we consider, i.e. between all three possible pairs of observations. Note that the analysis above could have been made with a fixed subject effect instead: proc glm data=a1; class id treatment period; model duration = id treatment period / solution; giving instead the output The GLM Procedure Class Level Information Class Levels Values id treatment 3 A B C period Dependent Variable: duration Sum of Source DF Squares Mean Square F Value Pr > F Model <.0001 Error Corrected Total R-Square Coeff Var Root MSE duration Mean Source DF Type III SS Mean Square F Value Pr > F id

9 treatment <.0001 period Parameter Estimate Error t Value Pr > t Intercept B <.0001 id B id B id B id B id B id B id B id B id B id B id B id B... treatment A B <.0001 treatment B B treatment C B... period B period B period B... We note that estimates, standard errors and P-values for treatment and period are exactly the same as in the mixed model above. So why bother with a mixed model? Several reasons actually: Evaluation of cross-over effects, possible covariates for subjects, quantification of subject differences, possible extensions to other covariance patterns, ease of reading output,... (a) What is the estimated difference in duration between 100 mg and 50 mg tablets? Remember to quantify with confidence intervals. Since this particular comparison (which is treatment B vs. A) is not one shown by default, we need either to reparametrize to another reference level, or to make an estimate statement. The latter is the easiest, and it was also included in the code above: estimate 100 mg vs 50 mg treatment / cl; which creates the additional output: Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg

10 Label Lower Upper 100 mg vs capsule i.e. an estimated prolonged duration of 0.58 hours, with confidence interval CI=(-1.00, -0.37), P= (b) Do we see any effect of period? Yes, as already noticed, we see a highly significant increase in duration with time (period, P = ), as seen from the pattern repeated below: Effect treatment period Estimate Error DF t Value Pr > t Intercept <.0001 period period period treatment A <.0001 treatment B treatment C Solution for Fixed Effects Effect treatment period Alpha Lower Upper Intercept period period period 3... treatment A treatment B treatment C... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F period treatment <.0001 These estimates tell us that period 1 gives an expected duration of 0.55 hours less than period 3, whereas period 2 has a duration between these two, and closest to period 3. This could be cautiously interpreted as a sort of cumulative effect over time, i.e. a carry-over effect, just as we saw in a previous picture. 10

11 4. What can we say about the heterogeneity among the subjects? When subjects are treated as fixed effects in the GLM-analysis, we get comparisons of all subjects to the reference subject (the last one in the ad hoc chosen ordering), as well as an overall test for identity of all subjects. This is not interesting, since we do not focus on these particular subjects but rather want to generalize to all subjects (in some specified population). We therefore turn to the mixed model, and use the quantification of between-subject variation: Covariance Parameter Estimates Cov Parm Subject Estimate Intercept id Residual This allows us to calculate typical differences in the duration of a drug (in hours), either between (hypothetical) repetitive measurements of the same treatment in the same period for the same subject, or between measurements of the same treatment in the same period for a different subject: Typical differences (95% Prediction Intervals): for measurements on the same subject ± = ±1.045 for measurements on the different subjects ±2 2 ( ) = ±1.35 The size of these heterogeneity can of course be seen already in the spaghetti plots from question 1, but here, we are able to quantify them. 5. Make two (erroneously) simple tests for the treatment effect, and discuss the possible pitfalls of these analyses: 11

12 A one-way analysis of variance In a one-way anova, we disregard the period effect and we do not take advantage of the pairing (i.e. that all subjects receive all treatments). If the design had been unbalanced, we could get bias due to the confounding between periods and treatments, but in a balanced design such as this, we only loose power. title wrong analysis ; proc glm data=a1; class treatment; model duration = treatment / solution clparm; estimate 100 mg vs 50 mg treatment ; with output wrong analysis The GLM Procedure Dependent Variable: duration Sum of Source DF Squares Mean Square F Value Pr > F Model <.0001 Error Corrected Total R-Square Coeff Var Root MSE duration Mean Source DF Type III SS Mean Square F Value Pr > F treatment <.0001 Parameter Estimate Error t Value Pr > t 100 mg vs 50 mg Parameter 95% Confidence Limits 100 mg vs 50 mg Parameter Estimate Error t Value Pr > t Intercept B <.0001 treatment A B <.0001 treatment B B treatment C B... Note that the estimated difference between treatment A and B is still as in the analyses above, but that the standard error of 12

13 the estimate has increased to instead of the correct level of above. A two-way analysis of variance in treatment and subject. In this analysis, we take the pairing into account, so if there is no period effect, this analysis would be OK. title wrong analysis ; proc glm data=a1; class id treatment; model duration = id treatment / solution clparm; estimate 100 mg vs 50 mg treatment ; with output wrong analysis The GLM Procedure Dependent Variable: duration Sum of Source DF Squares Mean Square F Value Pr > F Model Error Corrected Total R-Square Coeff Var Root MSE duration Mean Source DF Type III SS Mean Square F Value Pr > F id treatment <.0001 Parameter Estimate Error t Value Pr > t 100 mg vs 50 mg Parameter 95% Confidence Limits 100 mg vs 50 mg Parameter Estimate Error t Value Pr > t Intercept B <.0001 id B id B id B id B id B id B

14 id B id B id B id B id B id B... treatment A B <.0001 treatment B B treatment C B... Note once more that the estimated difference between treatment A and B is still as in all the previous analyses, and that the standard error is closer to the correct one than in the oneway anova, namely 0.187, pretty close to the one we got from the mixed model analysis (0.151). This reflects the limited effect of the periods (which was not included in this model). We could have produced a another wrong analysis by carrying out a two-way analysis in treatment and period, but this would again not have taken the correlation into account and would therefore be seriously inefficient, just like the one-way ANOVA presented above. 6. With the previously used variance component model, we implicitly assume a compound symmetry covariance structure, but since time/period (and different treatments over time) is involved, this may not be the best choice. Estimate the correlation structure between the three measurements of duration for each subject, by specifying it as unstructured. Does it look like a compound structure? Note, that the use of an unstructured covariance matrix requires the definition of the order of the observations, i.e. the definition of the time scale. It seems most natural to take this to mean periods, but note that this implies that e.g. period 1 and 2 are correlated to the same extent for all subjects, even though this implies different treatment pairs! We specify this by adding period as the repeated factor: proc mixed data=a1; class sequence id period treatment; model duration = period treatment / ddfm=kr s cl; 14

15 repeated period / type=un subject=id r rcorr; and get the output The Mixed Procedure Model Information Data Set Dependent Variable Covariance Structure Subject Effect Estimation Method Residual Variance Method Fixed Effects SE Method Degrees of Freedom Method WORK.A1 duration Unstructured id REML None Kenward-Roger Kenward-Roger Class Level Information Class Levels Values sequence 3 ABC BCA CAB id period treatment 3 A B C Dimensions Covariance Parameters 6 Columns in X 7 Columns in Z 0 Subjects 12 Max Obs Per Subject 3 Number of Observations Number of Observations Read 36 Number of Observations Used 36 Number of Observations Not Used 0 Estimated R Matrix for id Estimated R Correlation Matrix for id Fit Statistics -2 Res Log Likelihood 46.6 AIC (smaller is better)

16 AICC (smaller is better) 62.1 BIC (smaller is better) 61.5 Null Model Likelihood Ratio Test DF Chi-Square Pr > ChiSq Solution for Fixed Effects Effect treatment period Estimate Error DF t Value Pr > t Intercept <.0001 period period period treatment A <.0001 treatment B treatment C Effect treatment period Alpha Lower Upper Intercept period period period 3... treatment A treatment B treatment C... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F period treatment <.0001 Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg Label Lower Upper 100 mg vs 50 mg We note from the above estimate of the correlation structure, that period 2 and 3 are less correlated than the others. But remember, that the sequence of treatments is not the same in the three periods for all individuals. Note also, that the estimates for treatment comparisons have changed, compared to those we got from the CS-analysis in question 3. The estimates for period comparisons, however, remain unchanged. 16

17 We may specify the covariance structure in the order of the treatments instead, i.e. with the code proc mixed data=a1; class sequence id period treatment; model duration = period treatment / ddfm=kr s cl; repeated treatment / type=un subject=id r rcorr; estimate 100 mg vs 50 mg treatment / cl; which yields another output (because it is a different model, specifying a different covariance structure): The Mixed Procedure Model Information Data Set Dependent Variable Covariance Structure Subject Effect Estimation Method Residual Variance Method Fixed Effects SE Method Degrees of Freedom Method WORK.A1 duration Unstructured id REML None Kenward-Roger Kenward-Roger Class Level Information Class Levels Values sequence 3 ABC BCA CAB id period treatment 3 A B C Dimensions Covariance Parameters 6 Columns in X 7 Columns in Z 0 Subjects 12 Max Obs Per Subject 3 Number of Observations Number of Observations Read 36 Number of Observations Used 36 Number of Observations Not Used 0 Estimated R Matrix for id

18 Estimated R Correlation Matrix for id Fit Statistics -2 Res Log Likelihood 42.1 AIC (smaller is better) 54.1 AICC (smaller is better) 57.6 BIC (smaller is better) 57.0 Solution for Fixed Effects Effect treatment period Estimate Error DF t Value Pr > t Intercept <.0001 period period period treatment A <.0001 treatment B treatment C Effect treatment period Alpha Lower Upper Intercept period period period 3... treatment A treatment B treatment C... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F period treatment Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg Label Lower Upper 100 mg vs 50 mg We note from the above estimate of the correlation structure, that treatment A and B are less correlated than the others. This will be reflected in a larger standard error for this comparison. 18

19 Note also, that the estimates for the period comparisons have now changed, compared to those, we got from the CS-analysis in question 3. The estimates for the treatment comparisons, however, remain unchanged. The two different models above do not agree closely with respect to the estimated effects, and the standard errors also differ, because the assumptions concerning the covariance structure differ. Our predictions from the latter model is shown in the figure below: 19

20 We can summarize our results so far as follows: Number of Model B vs. A -2 log L parameters CS random (0.151) id UN treatment (0.168) order UN period (0.160) order 7. Does the correlation structure seem to depend upon the sequence of administration of the different formulations? We can fit different correlation structures for each sequence group by including group=sequence in the repeated-statement. Note that this gives the same model, no matter whether we specify period or treatment to be the time We sort the data before the fitting to make 20

21 clear how the correlations are to be interpreted. Furthermore, since the covariance structure now depends upon the group, we need to print 3 different structures. We choose to print those of patients number 1, 5 and 9 (one from each group): proc sort data=a1; by sequence id period; title UN period, group=sequence ; proc mixed data=a1; class sequence id period treatment; model duration = period treatment / ddfm=kr s cl; repeated period / type=un subject=id group=sequence r=1,5,9 rcorr=1,5,9; proc sort data=a1; by sequence id treatment; title UN treatment, group=sequence ; proc mixed data=a1; class sequence id period treatment; model duration = period treatment / ddfm=kr s cl; repeated treatment / type=un subject=id group=sequence r=1,5,9 rcorr=1,5,9; The output from the first of these models (with period as time) yields the output (only some of it is shown): UN period, group=sequence Dimensions Covariance Parameters 18 Columns in X 7 Columns in Z 0 Subjects 12 Max Obs Per Subject 3 Estimated R Matrix for id

22 Estimated R Correlation Matrix for id Estimated R Matrix for id Estimated R Correlation Matrix for id Estimated R Matrix for id Estimated R Correlation Matrix for id Fit Statistics -2 Res Log Likelihood 32.6 AIC (smaller is better) 68.6 AICC (smaller is better) BIC (smaller is better) 77.3 Solution for Fixed Effects Effect treatment period Estimate Error DF t Value Pr > t Intercept <.0001 period period period treatment A treatment B treatment C Effect treatment period Alpha Lower Upper Intercept period period period 3... treatment A treatment B treatment C... 22

23 Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F period treatment We have to be careful when interpreting the above covariance/correlation estimates, since their meaning depends on the sequence. For id=5, the sequence is BCA, and the correlation matrix therefore tells us, that Corr(A,C)< 0 and that Corr(A,B) is small, but positive. On the other hand, for id=9, the sequence is CAB, and the correlation matrix therefore tells us, that Corr(A,B)< 0. Since the model with treatment as time is the very same model, the only difference in the output will be the order of the elements in the covariance/correlation, and therefore these are the only output, we present: UN treatment, group=sequence The Mixed Procedure Estimated R Matrix for id Estimated R Correlation Matrix for id Estimated R Matrix for id Estimated R Correlation Matrix for id

24 Estimated R Matrix for id Estimated R Correlation Matrix for id Note that in this output, all covariance/correlation matrices are in the order ABC. Comparing this model to the two simpler previous models yields Number of Model B vs. A -2 log L parameters UN separate (0.172) groups UN treatment (0.168) order UN period (0.160) order CS random (0.151) id Testing equality of covariance structures in the three groups (the simpler alternative being either an unstructured covariance for treatments, or an unstructured covariance for periods), we get Treatment-order: 2 log Q = = 9.5 χ 2 (12) P = 0.66 Period-order: 2 log Q = = 14.0 χ 2 (12) P = 0.30 i.e. no indication of a sequence-dependent covariance structure (possibly because the data set is so small). There is even no convincingly argument 24

25 against the simple compound symmetry covariance structure (P=0.10). Comparison of CS to UN-treatment: 2 log Q = = 7.6 χ 2 (4) P = Make an analysis with an appropriate correlation structure, and estimate again the difference in duration between 100 mg and 50 mg tablets, with confidence intervals. If we have to choose between the two sequence-invariant covariance structures, we choose the one with the smallest 2 log L, i.e. the one, where the covariance is determined by the treatment, not the period. We have previously seen the estimate for the difference between 100 mg and 50 mg in this model, namely: (0.168), P= Since all of the covariance structures have agreed upon the significance of this difference, it seems safe to conclude from this investigation regardless of the small sample size. 9. Can we detect any evidence of (one or more) carry-over effects in these data? Since drug C is a sustained-release formulation, we will in advance expect this to show the strongest signs of a carry-over effect, followed by the highdose tablet B. The easy choice to look for a carry-over effect would be just to include an interaction between treatment and period. This is however not very wise: In a two-period cross-over study (with two treatments), this is usually done, but it is important to point out that interaction can arise due to other effects than carry-over. In a three-period design as this, it is however possible to distinguish between the two by defining new variables to model the potential effect of carry over directly. Since we have a three-period design, there may be different carry-over effects depending upon the drug used immediately before, and we therefore define the variable previous1 to hold this information. Note, that the value 25

26 N means that there is no previous treatment. Likewise, we define previous2 to hold information about the treatnent given two periods before. This is of course only different from N for treatments given in period 3. previous1= N ; if group=1 and period=2 then previous1= A ; if group=1 and period=3 then previous1= B ; if group=2 and period=2 then previous1= B ; if group=2 and period=3 then previous1= C ; if group=3 and period=2 then previous1= C ; if group=3 and period=3 then previous1= A ; previous2= N ; if group=1 and period=3 then previous2= A ; if group=2 and period=3 then previous2= B ; if group=3 and period=3 then previous2= C ; With these two variables (previous1 and previous2) as covariates, we must exclude period from the analysis because of confounding, and we therefor make the analysis title specific carry-over ; proc mixed data=a1; class sequence id period treatment previous1 previous2; model duration = treatment previous1 previous2 / outpm=predm s cl; repeated treatment / type=un subject=id r rcorr; yielding the output specific carry-over Model Information Data Set Dependent Variable Covariance Structure Subject Effect Estimation Method Degrees of Freedom Method WORK.A1 duration Unstructured id REML Between-Within 26

27 Class Level Information Class Levels Values sequence 3 ABC BCA CAB id period treatment 3 A B C previous 4 A B C N previous1 4 A B C N previous2 4 A B C N Dimensions Covariance Parameters 6 Columns in X 12 Columns in Z 0 Subjects 12 Max Obs Per Subject 3 Estimated R Matrix for id Estimated R Correlation Matrix for id Solution for Fixed Effects Effect treatment previous1 previous2 Estimate Error DF Intercept treatment A treatment B treatment C 0.. previous1 A previous1 B previous1 C previous1 N 0.. previous2 A previous2 B previous2 C previous2 N 0.. Effect treatment previous1 previous2 t Value Pr > t Alpha Intercept 9.58 < treatment A treatment B treatment C... previous1 A previous1 B previous1 C previous1 N... previous2 A previous2 B previous2 C

28 previous2 N... Effect treatment previous1 previous2 Lower Upper Intercept treatment A treatment B treatment C.. previous1 A previous1 B previous1 C previous1 N.. previous2 A previous2 B previous2 C previous2 N.. Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F treatment previous previous Our predictions from this model is once again identical to the averages, so it is not shown here. We note, that the treatment given in the period immediately before seems to have an effect on the duration on the treatment to follow (P=0.046): Treatments A and C seems to linger on for a while. The treatment given two periods before does not seem to have any influence (P=0.41). If we omit this from the analysis, we have the model: title specific carry-over, only previous1 ; proc mixed data=a1; class sequence id period treatment previous1 previous2; model duration = treatment previous1 / outpm=predm s cl ; repeated treatment / type=un subject=id r rcorr; estimate 100 mg vs 50 mg treatment / cl; which yields the output specific carry-over, only previous1 The Mixed Procedure Class Level Information Class Levels Values 28

29 sequence 3 ABC BCA CAB id period treatment 3 A B C previous1 4 A B C N Dimensions Covariance Parameters 6 Columns in X 8 Columns in Z 0 Subjects 12 Max Obs Per Subject 3 Estimated R Matrix for id Estimated R Correlation Matrix for id Solution for Fixed Effects Effect treatment previous1 Estimate Error DF t Value Intercept treatment A treatment B treatment C 0... previous1 A previous1 B previous1 C previous1 N 0... Effect treatment previous1 Pr > t Alpha Lower Upper Intercept < treatment A treatment B treatment C.... previous1 A previous1 B previous1 C previous1 N.... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F treatment previous Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg

30 Label Lower Upper 100 mg vs 50 mg Based on this output, our conclusion seems to be that treatment C has a carry-over effect, since it prolongs the duration of the treatment to follow by an estimated 0.87 hours, CI=(0.26, 1.49). Our predictions from this reduced model is shown in the figure below: The carry-over effect of A and B are estimated to be quite similar, and not significant, so it is tempting to exclude these (note that we ought to make a formal test for excluding them both simultaneously, a test which would give P=0.11). We do this by defining the new covariate previousc= N ; if group=2 and period=3 then previousc= C ; if group=3 and period=2 then previousc= C ; and using the new model title specific carry-over, only previousc ; proc mixed data=a1; 30

31 class sequence id period treatment previousc previous2; model duration = treatment previousc / outpm=predm s cl ; repeated treatment / type=un subject=id r rcorr; estimate 100 mg vs 50 mg treatment / cl; from which we get the output specific carry-over, only previousc Solution for Fixed Effects previous Effect treatment C Estimate Error DF t Value Pr > t Intercept <.0001 treatment A <.0001 treatment B treatment C previousc C previousc N previous Effect treatment C Alpha Lower Upper Intercept treatment A treatment B treatment C... previousc C previousc N... Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F treatment <.0001 previousc Estimates Label Estimate Error DF t Value Pr > t Alpha 100 mg vs 50 mg Label Lower Upper 100 mg vs 50 mg Based on this, our conclusion is that treatment C definitely shows sign of a carry-over effect, since it prolongs the duration of the treatment to follow by an estimated 0.94 hours, CI=(0.36, 1.51), P= Note also the very important fact, that the estimated effects of the drugs change when we add this carry-over effect to the model! 31

Introduction to SAS proc mixed

Introduction to SAS proc mixed Faculty of Health Sciences Introduction to SAS proc mixed Analysis of repeated measurements, 2017 Julie Forman Department of Biostatistics, University of Copenhagen Outline Data in wide and long format

More information

Introduction to SAS proc mixed

Introduction to SAS proc mixed Faculty of Health Sciences Introduction to SAS proc mixed Analysis of repeated measurements, 2017 Julie Forman Department of Biostatistics, University of Copenhagen 2 / 28 Preparing data for analysis The

More information

ANOVA Longitudinal Models for the Practice Effects Data: via GLM

ANOVA Longitudinal Models for the Practice Effects Data: via GLM Psyc 943 Lecture 25 page 1 ANOVA Longitudinal Models for the Practice Effects Data: via GLM Model 1. Saturated Means Model for Session, E-only Variances Model (BP) Variances Model: NO correlation, EQUAL

More information

Covariance Structure Approach to Within-Cases

Covariance Structure Approach to Within-Cases Covariance Structure Approach to Within-Cases Remember how the data file grapefruit1.data looks: Store sales1 sales2 sales3 1 62.1 61.3 60.8 2 58.2 57.9 55.1 3 51.6 49.2 46.2 4 53.7 51.5 48.3 5 61.4 58.7

More information

Subject-specific observed profiles of log(fev1) vs age First 50 subjects in Six Cities Study

Subject-specific observed profiles of log(fev1) vs age First 50 subjects in Six Cities Study Subject-specific observed profiles of log(fev1) vs age First 50 subjects in Six Cities Study 1.4 0.0-6 7 8 9 10 11 12 13 14 15 16 17 18 19 age Model 1: A simple broken stick model with knot at 14 fit with

More information

SAS Syntax and Output for Data Manipulation:

SAS Syntax and Output for Data Manipulation: CLP 944 Example 5 page 1 Practice with Fixed and Random Effects of Time in Modeling Within-Person Change The models for this example come from Hoffman (2015) chapter 5. We will be examining the extent

More information

over Time line for the means). Specifically, & covariances) just a fixed variance instead. PROC MIXED: to 1000 is default) list models with TYPE=VC */

over Time line for the means). Specifically, & covariances) just a fixed variance instead. PROC MIXED: to 1000 is default) list models with TYPE=VC */ CLP 944 Example 4 page 1 Within-Personn Fluctuation in Symptom Severity over Time These data come from a study of weekly fluctuation in psoriasis severity. There was no intervention and no real reason

More information

Models for longitudinal data

Models for longitudinal data Faculty of Health Sciences Contents Models for longitudinal data Analysis of repeated measurements, NFA 016 Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics, University of Copenhagen

More information

Topic 17 - Single Factor Analysis of Variance. Outline. One-way ANOVA. The Data / Notation. One way ANOVA Cell means model Factor effects model

Topic 17 - Single Factor Analysis of Variance. Outline. One-way ANOVA. The Data / Notation. One way ANOVA Cell means model Factor effects model Topic 17 - Single Factor Analysis of Variance - Fall 2013 One way ANOVA Cell means model Factor effects model Outline Topic 17 2 One-way ANOVA Response variable Y is continuous Explanatory variable is

More information

Correlated data. Repeated measurements over time. Typical set-up for repeated measurements. Traditional presentation of data

Correlated data. Repeated measurements over time. Typical set-up for repeated measurements. Traditional presentation of data Faculty of Health Sciences Repeated measurements over time Correlated data NFA, May 22, 2014 Longitudinal measurements Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics University of

More information

Multi-factor analysis of variance

Multi-factor analysis of variance Faculty of Health Sciences Outline Multi-factor analysis of variance Basic statistics for experimental researchers 2015 Two-way ANOVA and interaction Mathed samples ANOVA Random vs systematic variation

More information

MIXED MODELS FOR REPEATED (LONGITUDINAL) DATA PART 2 DAVID C. HOWELL 4/1/2010

MIXED MODELS FOR REPEATED (LONGITUDINAL) DATA PART 2 DAVID C. HOWELL 4/1/2010 MIXED MODELS FOR REPEATED (LONGITUDINAL) DATA PART 2 DAVID C. HOWELL 4/1/2010 Part 1 of this document can be found at http://www.uvm.edu/~dhowell/methods/supplements/mixed Models for Repeated Measures1.pdf

More information

Analysis of Longitudinal Data: Comparison Between PROC GLM and PROC MIXED. Maribeth Johnson Medical College of Georgia Augusta, GA

Analysis of Longitudinal Data: Comparison Between PROC GLM and PROC MIXED. Maribeth Johnson Medical College of Georgia Augusta, GA Analysis of Longitudinal Data: Comparison Between PROC GLM and PROC MIXED Maribeth Johnson Medical College of Georgia Augusta, GA Overview Introduction to longitudinal data Describe the data for examples

More information

SAS Code for Data Manipulation: SPSS Code for Data Manipulation: STATA Code for Data Manipulation: Psyc 945 Example 1 page 1

SAS Code for Data Manipulation: SPSS Code for Data Manipulation: STATA Code for Data Manipulation: Psyc 945 Example 1 page 1 Psyc 945 Example page Example : Unconditional Models for Change in Number Match 3 Response Time (complete data, syntax, and output available for SAS, SPSS, and STATA electronically) These data come from

More information

Topic 20: Single Factor Analysis of Variance

Topic 20: Single Factor Analysis of Variance Topic 20: Single Factor Analysis of Variance Outline Single factor Analysis of Variance One set of treatments Cell means model Factor effects model Link to linear regression using indicator explanatory

More information

Correlated data. Longitudinal data. Typical set-up for repeated measurements. Examples from literature, I. Faculty of Health Sciences

Correlated data. Longitudinal data. Typical set-up for repeated measurements. Examples from literature, I. Faculty of Health Sciences Faculty of Health Sciences Longitudinal data Correlated data Longitudinal measurements Outline Designs Models for the mean Covariance patterns Lene Theil Skovgaard November 27, 2015 Random regression Baseline

More information

dm'log;clear;output;clear'; options ps=512 ls=99 nocenter nodate nonumber nolabel FORMCHAR=" = -/\<>*"; ODS LISTING;

dm'log;clear;output;clear'; options ps=512 ls=99 nocenter nodate nonumber nolabel FORMCHAR= = -/\<>*; ODS LISTING; dm'log;clear;output;clear'; options ps=512 ls=99 nocenter nodate nonumber nolabel FORMCHAR=" ---- + ---+= -/\*"; ODS LISTING; *** Table 23.2 ********************************************; *** Moore, David

More information

Chapter 11. Analysis of Variance (One-Way)

Chapter 11. Analysis of Variance (One-Way) Chapter 11 Analysis of Variance (One-Way) We now develop a statistical procedure for comparing the means of two or more groups, known as analysis of variance or ANOVA. These groups might be the result

More information

Odor attraction CRD Page 1

Odor attraction CRD Page 1 Odor attraction CRD Page 1 dm'log;clear;output;clear'; options ps=512 ls=99 nocenter nodate nonumber nolabel FORMCHAR=" ---- + ---+= -/\*"; ODS LISTING; *** Table 23.2 ********************************************;

More information

UNIVERSITY OF TORONTO. Faculty of Arts and Science APRIL 2010 EXAMINATIONS STA 303 H1S / STA 1002 HS. Duration - 3 hours. Aids Allowed: Calculator

UNIVERSITY OF TORONTO. Faculty of Arts and Science APRIL 2010 EXAMINATIONS STA 303 H1S / STA 1002 HS. Duration - 3 hours. Aids Allowed: Calculator UNIVERSITY OF TORONTO Faculty of Arts and Science APRIL 2010 EXAMINATIONS STA 303 H1S / STA 1002 HS Duration - 3 hours Aids Allowed: Calculator LAST NAME: FIRST NAME: STUDENT NUMBER: There are 27 pages

More information

Topic 25 - One-Way Random Effects Models. Outline. Random Effects vs Fixed Effects. Data for One-way Random Effects Model. One-way Random effects

Topic 25 - One-Way Random Effects Models. Outline. Random Effects vs Fixed Effects. Data for One-way Random Effects Model. One-way Random effects Topic 5 - One-Way Random Effects Models One-way Random effects Outline Model Variance component estimation - Fall 013 Confidence intervals Topic 5 Random Effects vs Fixed Effects Consider factor with numerous

More information

SAS Syntax and Output for Data Manipulation: CLDP 944 Example 3a page 1

SAS Syntax and Output for Data Manipulation: CLDP 944 Example 3a page 1 CLDP 944 Example 3a page 1 From Between-Person to Within-Person Models for Longitudinal Data The models for this example come from Hoffman (2015) chapter 3 example 3a. We will be examining the extent to

More information

WITHIN-PARTICIPANT EXPERIMENTAL DESIGNS

WITHIN-PARTICIPANT EXPERIMENTAL DESIGNS 1 WITHIN-PARTICIPANT EXPERIMENTAL DESIGNS I. Single-factor designs: the model is: yij i j ij ij where: yij score for person j under treatment level i (i = 1,..., I; j = 1,..., n) overall mean βi treatment

More information

Analysis of variance and regression. May 13, 2008

Analysis of variance and regression. May 13, 2008 Analysis of variance and regression May 13, 2008 Repeated measurements over time Presentation of data Traditional ways of analysis Variance component model (the dogs revisited) Random regression Baseline

More information

Statistics for exp. medical researchers Regression and Correlation

Statistics for exp. medical researchers Regression and Correlation Faculty of Health Sciences Regression analysis Statistics for exp. medical researchers Regression and Correlation Lene Theil Skovgaard Sept. 28, 2015 Linear regression, Estimation and Testing Confidence

More information

Faculty of Health Sciences. Correlated data. Variance component models. Lene Theil Skovgaard & Julie Lyng Forman.

Faculty of Health Sciences. Correlated data. Variance component models. Lene Theil Skovgaard & Julie Lyng Forman. Faculty of Health Sciences Correlated data Variance component models Lene Theil Skovgaard & Julie Lyng Forman November 27, 2018 1 / 84 Overview One-way anova with random variation The rabbit example Hierarchical

More information

Correlated data. Overview. Example: Swelling due to vaccine. Variance component models. Faculty of Health Sciences. Variance component models

Correlated data. Overview. Example: Swelling due to vaccine. Variance component models. Faculty of Health Sciences. Variance component models Faculty of Health Sciences Overview Correlated data Variance component models One-way anova with random variation The rabbit example Hierarchical models with several levels Random regression Lene Theil

More information

Correlated data. Overview. Variance component models. Terminology for correlated measurements. Faculty of Health Sciences. Variance component models

Correlated data. Overview. Variance component models. Terminology for correlated measurements. Faculty of Health Sciences. Variance component models Faculty of Health Sciences Overview Correlated data Variance component models Lene Theil Skovgaard & Julie Lyng Forman November 29, 2016 One-way anova with random variation The rabbit example Hierarchical

More information

Testing Indirect Effects for Lower Level Mediation Models in SAS PROC MIXED

Testing Indirect Effects for Lower Level Mediation Models in SAS PROC MIXED Testing Indirect Effects for Lower Level Mediation Models in SAS PROC MIXED Here we provide syntax for fitting the lower-level mediation model using the MIXED procedure in SAS as well as a sas macro, IndTest.sas

More information

Lecture 4. Random Effects in Completely Randomized Design

Lecture 4. Random Effects in Completely Randomized Design Lecture 4. Random Effects in Completely Randomized Design Montgomery: 3.9, 13.1 and 13.7 1 Lecture 4 Page 1 Random Effects vs Fixed Effects Consider factor with numerous possible levels Want to draw inference

More information

This is a Randomized Block Design (RBD) with a single factor treatment arrangement (2 levels) which are fixed.

This is a Randomized Block Design (RBD) with a single factor treatment arrangement (2 levels) which are fixed. EXST3201 Chapter 13c Geaghan Fall 2005: Page 1 Linear Models Y ij = µ + βi + τ j + βτij + εijk This is a Randomized Block Design (RBD) with a single factor treatment arrangement (2 levels) which are fixed.

More information

STAT 5200 Handout #23. Repeated Measures Example (Ch. 16)

STAT 5200 Handout #23. Repeated Measures Example (Ch. 16) Motivating Example: Glucose STAT 500 Handout #3 Repeated Measures Example (Ch. 16) An experiment is conducted to evaluate the effects of three diets on the serum glucose levels of human subjects. Twelve

More information

Faculty of Health Sciences. Correlated data. Variance component models. Lene Theil Skovgaard & Julie Lyng Forman.

Faculty of Health Sciences. Correlated data. Variance component models. Lene Theil Skovgaard & Julie Lyng Forman. Faculty of Health Sciences Correlated data Variance component models Lene Theil Skovgaard & Julie Lyng Forman November 28, 2017 1 / 96 Overview One-way anova with random variation The rabbit example Hierarchical

More information

Correlated data. Overview. Variance component models. Terminology for correlated measurements. Faculty of Health Sciences. Variance component models

Correlated data. Overview. Variance component models. Terminology for correlated measurements. Faculty of Health Sciences. Variance component models Faculty of Health Sciences Overview Correlated data Variance component models Lene Theil Skovgaard & Julie Lyng Forman November 28, 2017 One-way anova with random variation The rabbit example Hierarchical

More information

General Linear Model (Chapter 4)

General Linear Model (Chapter 4) General Linear Model (Chapter 4) Outcome variable is considered continuous Simple linear regression Scatterplots OLS is BLUE under basic assumptions MSE estimates residual variance testing regression coefficients

More information

Regression: Main Ideas Setting: Quantitative outcome with a quantitative explanatory variable. Example, cont.

Regression: Main Ideas Setting: Quantitative outcome with a quantitative explanatory variable. Example, cont. TCELL 9/4/205 36-309/749 Experimental Design for Behavioral and Social Sciences Simple Regression Example Male black wheatear birds carry stones to the nest as a form of sexual display. Soler et al. wanted

More information

Variance component models part I

Variance component models part I Faculty of Health Sciences Variance component models part I Analysis of repeated measurements, 30th November 2012 Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics, University of Copenhagen

More information

Review of CLDP 944: Multilevel Models for Longitudinal Data

Review of CLDP 944: Multilevel Models for Longitudinal Data Review of CLDP 944: Multilevel Models for Longitudinal Data Topics: Review of general MLM concepts and terminology Model comparisons and significance testing Fixed and random effects of time Significance

More information

36-309/749 Experimental Design for Behavioral and Social Sciences. Sep. 22, 2015 Lecture 4: Linear Regression

36-309/749 Experimental Design for Behavioral and Social Sciences. Sep. 22, 2015 Lecture 4: Linear Regression 36-309/749 Experimental Design for Behavioral and Social Sciences Sep. 22, 2015 Lecture 4: Linear Regression TCELL Simple Regression Example Male black wheatear birds carry stones to the nest as a form

More information

Introduction to Within-Person Analysis and RM ANOVA

Introduction to Within-Person Analysis and RM ANOVA Introduction to Within-Person Analysis and RM ANOVA Today s Class: From between-person to within-person ANOVAs for longitudinal data Variance model comparisons using 2 LL CLP 944: Lecture 3 1 The Two Sides

More information

Outline. Topic 20 - Diagnostics and Remedies. Residuals. Overview. Diagnostics Plots Residual checks Formal Tests. STAT Fall 2013

Outline. Topic 20 - Diagnostics and Remedies. Residuals. Overview. Diagnostics Plots Residual checks Formal Tests. STAT Fall 2013 Topic 20 - Diagnostics and Remedies - Fall 2013 Diagnostics Plots Residual checks Formal Tests Remedial Measures Outline Topic 20 2 General assumptions Overview Normally distributed error terms Independent

More information

Topic 23: Diagnostics and Remedies

Topic 23: Diagnostics and Remedies Topic 23: Diagnostics and Remedies Outline Diagnostics residual checks ANOVA remedial measures Diagnostics Overview We will take the diagnostics and remedial measures that we learned for regression and

More information

Introduction to Random Effects of Time and Model Estimation

Introduction to Random Effects of Time and Model Estimation Introduction to Random Effects of Time and Model Estimation Today s Class: The Big Picture Multilevel model notation Fixed vs. random effects of time Random intercept vs. random slope models How MLM =

More information

Analysis of variance and regression. November 22, 2007

Analysis of variance and regression. November 22, 2007 Analysis of variance and regression November 22, 2007 Parametrisations: Choice of parameters Comparison of models Test for linearity Linear splines Lene Theil Skovgaard, Dept. of Biostatistics, Institute

More information

Contrasting Marginal and Mixed Effects Models Recall: two approaches to handling dependence in Generalized Linear Models:

Contrasting Marginal and Mixed Effects Models Recall: two approaches to handling dependence in Generalized Linear Models: Contrasting Marginal and Mixed Effects Models Recall: two approaches to handling dependence in Generalized Linear Models: Marginal models: based on the consequences of dependence on estimating model parameters.

More information

T-test: means of Spock's judge versus all other judges 1 12:10 Wednesday, January 5, judge1 N Mean Std Dev Std Err Minimum Maximum

T-test: means of Spock's judge versus all other judges 1 12:10 Wednesday, January 5, judge1 N Mean Std Dev Std Err Minimum Maximum T-test: means of Spock's judge versus all other judges 1 The TTEST Procedure Variable: pcwomen judge1 N Mean Std Dev Std Err Minimum Maximum OTHER 37 29.4919 7.4308 1.2216 16.5000 48.9000 SPOCKS 9 14.6222

More information

Repeated Measures Modeling With PROC MIXED E. Barry Moser, Louisiana State University, Baton Rouge, LA

Repeated Measures Modeling With PROC MIXED E. Barry Moser, Louisiana State University, Baton Rouge, LA Paper 188-29 Repeated Measures Modeling With PROC MIXED E. Barry Moser, Louisiana State University, Baton Rouge, LA ABSTRACT PROC MIXED provides a very flexible environment in which to model many types

More information

PLS205 Lab 2 January 15, Laboratory Topic 3

PLS205 Lab 2 January 15, Laboratory Topic 3 PLS205 Lab 2 January 15, 2015 Laboratory Topic 3 General format of ANOVA in SAS Testing the assumption of homogeneity of variances by "/hovtest" by ANOVA of squared residuals Proc Power for ANOVA One-way

More information

Repeated Measures Data

Repeated Measures Data Repeated Measures Data Mixed Models Lecture Notes By Dr. Hanford page 1 Data where subjects are measured repeatedly over time - predetermined intervals (weekly) - uncontrolled variable intervals between

More information

Repeated Measures Design. Advertising Sales Example

Repeated Measures Design. Advertising Sales Example STAT:5201 Anaylsis/Applied Statistic II Repeated Measures Design Advertising Sales Example A company is interested in comparing the success of two different advertising campaigns. It has 10 test markets,

More information

UNIVERSITY OF TORONTO Faculty of Arts and Science

UNIVERSITY OF TORONTO Faculty of Arts and Science UNIVERSITY OF TORONTO Faculty of Arts and Science December 2013 Final Examination STA442H1F/2101HF Methods of Applied Statistics Jerry Brunner Duration - 3 hours Aids: Calculator Model(s): Any calculator

More information

Topic 32: Two-Way Mixed Effects Model

Topic 32: Two-Way Mixed Effects Model Topic 3: Two-Way Mixed Effects Model Outline Two-way mixed models Three-way mixed models Data for two-way design Y is the response variable Factor A with levels i = 1 to a Factor B with levels j = 1 to

More information

Correlated data. Variance component models. Example: Evaluate vaccine. Traditional assumption so far. Faculty of Health Sciences

Correlated data. Variance component models. Example: Evaluate vaccine. Traditional assumption so far. Faculty of Health Sciences Faculty of Health Sciences Variance component models Definitions and motivation Correlated data Variance component models, I Lene Theil Skovgaard November 29, 2013 One-way anova with random variation The

More information

Variance component models

Variance component models Faculty of Health Sciences Variance component models Analysis of repeated measurements, NFA 2016 Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics, University of Copenhagen Topics for

More information

Outline. Topic 19 - Inference. The Cell Means Model. Estimates. Inference for Means Differences in cell means Contrasts. STAT Fall 2013

Outline. Topic 19 - Inference. The Cell Means Model. Estimates. Inference for Means Differences in cell means Contrasts. STAT Fall 2013 Topic 19 - Inference - Fall 2013 Outline Inference for Means Differences in cell means Contrasts Multiplicity Topic 19 2 The Cell Means Model Expressed numerically Y ij = µ i + ε ij where µ i is the theoretical

More information

Descriptions of post-hoc tests

Descriptions of post-hoc tests Experimental Statistics II Page 81 Descriptions of post-hoc tests Post-hoc or Post-ANOVA tests! Once you have found out some treatment(s) are different, how do you determine which one(s) are different?

More information

Random Intercept Models

Random Intercept Models Random Intercept Models Edps/Psych/Soc 589 Carolyn J. Anderson Department of Educational Psychology c Board of Trustees, University of Illinois Spring 2019 Outline A very simple case of a random intercept

More information

Independence (Null) Baseline Model: Item means and variances, but NO covariances

Independence (Null) Baseline Model: Item means and variances, but NO covariances CFA Example Using Forgiveness of Situations (N = 1103) using SAS MIXED (See Example 4 for corresponding Mplus syntax and output) SAS Code to Read in Mplus Data: * Import data from Mplus, becomes var1-var23

More information

Laboratory Topics 4 & 5

Laboratory Topics 4 & 5 PLS205 Lab 3 January 23, 2014 Orthogonal contrasts Class comparisons in SAS Trend analysis in SAS Multiple mean comparisons Laboratory Topics 4 & 5 Orthogonal contrasts Planned, single degree-of-freedom

More information

Analyzing the Behavior of Rats by Repeated Measurements

Analyzing the Behavior of Rats by Repeated Measurements Georgia State University ScholarWorks @ Georgia State University Mathematics Theses Department of Mathematics and Statistics 5-3-007 Analyzing the Behavior of Rats by Repeated Measurements Kenita A. Hall

More information

Random Coefficients Model Examples

Random Coefficients Model Examples Random Coefficients Model Examples STAT:5201 Week 15 - Lecture 2 1 / 26 Each subject (or experimental unit) has multiple measurements (this could be over time, or it could be multiple measurements on a

More information

WISE Regression/Correlation Interactive Lab. Introduction to the WISE Correlation/Regression Applet

WISE Regression/Correlation Interactive Lab. Introduction to the WISE Correlation/Regression Applet WISE Regression/Correlation Interactive Lab Introduction to the WISE Correlation/Regression Applet This tutorial focuses on the logic of regression analysis with special attention given to variance components.

More information

Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED.

Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED. Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED. Maribeth Johnson, Medical College of Georgia, Augusta, GA ABSTRACT Longitudinal data refers to datasets with multiple measurements

More information

Time-Invariant Predictors in Longitudinal Models

Time-Invariant Predictors in Longitudinal Models Time-Invariant Predictors in Longitudinal Models Today s Class (or 3): Summary of steps in building unconditional models for time What happens to missing predictors Effects of time-invariant predictors

More information

" M A #M B. Standard deviation of the population (Greek lowercase letter sigma) σ 2

 M A #M B. Standard deviation of the population (Greek lowercase letter sigma) σ 2 Notation and Equations for Final Exam Symbol Definition X The variable we measure in a scientific study n The size of the sample N The size of the population M The mean of the sample µ The mean of the

More information

Describing Within-Person Change over Time

Describing Within-Person Change over Time Describing Within-Person Change over Time Topics: Multilevel modeling notation and terminology Fixed and random effects of linear time Predicted variances and covariances from random slopes Dependency

More information

Step 2: Select Analyze, Mixed Models, and Linear.

Step 2: Select Analyze, Mixed Models, and Linear. Example 1a. 20 employees were given a mood questionnaire on Monday, Wednesday and again on Friday. The data will be first be analyzed using a Covariance Pattern model. Step 1: Copy Example1.sav data file

More information

Faculty of Health Sciences. Correlated data. Count variables. Lene Theil Skovgaard & Julie Lyng Forman. December 6, 2016

Faculty of Health Sciences. Correlated data. Count variables. Lene Theil Skovgaard & Julie Lyng Forman. December 6, 2016 Faculty of Health Sciences Correlated data Count variables Lene Theil Skovgaard & Julie Lyng Forman December 6, 2016 1 / 76 Modeling count outcomes Outline The Poisson distribution for counts Poisson models,

More information

Describing Change over Time: Adding Linear Trends

Describing Change over Time: Adding Linear Trends Describing Change over Time: Adding Linear Trends Longitudinal Data Analysis Workshop Section 7 University of Georgia: Institute for Interdisciplinary Research in Education and Human Development Section

More information

An Introduction to Multilevel Models. PSYC 943 (930): Fundamentals of Multivariate Modeling Lecture 25: December 7, 2012

An Introduction to Multilevel Models. PSYC 943 (930): Fundamentals of Multivariate Modeling Lecture 25: December 7, 2012 An Introduction to Multilevel Models PSYC 943 (930): Fundamentals of Multivariate Modeling Lecture 25: December 7, 2012 Today s Class Concepts in Longitudinal Modeling Between-Person vs. +Within-Person

More information

Describing Within-Person Fluctuation over Time using Alternative Covariance Structures

Describing Within-Person Fluctuation over Time using Alternative Covariance Structures Describing Within-Person Fluctuation over Time using Alternative Covariance Structures Today s Class: The Big Picture ACS models using the R matrix only Introducing the G, Z, and V matrices ACS models

More information

Lab 11. Multilevel Models. Description of Data

Lab 11. Multilevel Models. Description of Data Lab 11 Multilevel Models Henian Chen, M.D., Ph.D. Description of Data MULTILEVEL.TXT is clustered data for 386 women distributed across 40 groups. ID: 386 women, id from 1 to 386, individual level (level

More information

Supplemental Materials. In the main text, we recommend graphing physiological values for individual dyad

Supplemental Materials. In the main text, we recommend graphing physiological values for individual dyad 1 Supplemental Materials Graphing Values for Individual Dyad Members over Time In the main text, we recommend graphing physiological values for individual dyad members over time to aid in the decision

More information

Faculty of Health Sciences. Correlated data. More about LMMs. Lene Theil Skovgaard. December 4, / 104

Faculty of Health Sciences. Correlated data. More about LMMs. Lene Theil Skovgaard. December 4, / 104 Faculty of Health Sciences Correlated data More about LMMs Lene Theil Skovgaard December 4, 2015 1 / 104 Further topics Model check and diagnostics Cross-over studies Paired T-tests with missing values

More information

ST 732, Midterm Solutions Spring 2019

ST 732, Midterm Solutions Spring 2019 ST 732, Midterm Solutions Spring 2019 Please sign the following pledge certifying that the work on this test is your own: I have neither given nor received aid on this test. Signature: Printed Name: There

More information

6. Multiple regression - PROC GLM

6. Multiple regression - PROC GLM Use of SAS - November 2016 6. Multiple regression - PROC GLM Karl Bang Christensen Department of Biostatistics, University of Copenhagen. http://biostat.ku.dk/~kach/sas2016/ kach@biostat.ku.dk, tel: 35327491

More information

Swabs, revisited. The families were subdivided into 3 groups according to the factor crowding, which describes the space available for the household.

Swabs, revisited. The families were subdivided into 3 groups according to the factor crowding, which describes the space available for the household. Swabs, revisited 18 families with 3 children each (in well defined age intervals) were followed over a certain period of time, during which repeated swabs were taken. The variable swabs indicates how many

More information

SAS Commands. General Plan. Output. Construct scatterplot / interaction plot. Run full model

SAS Commands. General Plan. Output. Construct scatterplot / interaction plot. Run full model Topic 23 - Unequal Replication Data Model Outline - Fall 2013 Parameter Estimates Inference Topic 23 2 Example Page 954 Data for Two Factor ANOVA Y is the response variable Factor A has levels i = 1, 2,...,

More information

Statistical Inference: The Marginal Model

Statistical Inference: The Marginal Model Statistical Inference: The Marginal Model Edps/Psych/Stat 587 Carolyn J. Anderson Department of Educational Psychology c Board of Trustees, University of Illinois Fall 2017 Outline Inference for fixed

More information

Models for binary data

Models for binary data Faculty of Health Sciences Models for binary data Analysis of repeated measurements 2015 Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics, University of Copenhagen 1 / 63 Program for

More information

Stat 5303 (Oehlert): Randomized Complete Blocks 1

Stat 5303 (Oehlert): Randomized Complete Blocks 1 Stat 5303 (Oehlert): Randomized Complete Blocks 1 > library(stat5303libs);library(cfcdae);library(lme4) > immer Loc Var Y1 Y2 1 UF M 81.0 80.7 2 UF S 105.4 82.3 3 UF V 119.7 80.4 4 UF T 109.7 87.2 5 UF

More information

Time-Invariant Predictors in Longitudinal Models

Time-Invariant Predictors in Longitudinal Models Time-Invariant Predictors in Longitudinal Models Topics: What happens to missing predictors Effects of time-invariant predictors Fixed vs. systematically varying vs. random effects Model building strategies

More information

STAT 350. Assignment 4

STAT 350. Assignment 4 STAT 350 Assignment 4 1. For the Mileage data in assignment 3 conduct a residual analysis and report your findings. I used the full model for this since my answers to assignment 3 suggested we needed the

More information

You can specify the response in the form of a single variable or in the form of a ratio of two variables denoted events/trials.

You can specify the response in the form of a single variable or in the form of a ratio of two variables denoted events/trials. The GENMOD Procedure MODEL Statement MODEL response = < effects > < /options > ; MODEL events/trials = < effects > < /options > ; You can specify the response in the form of a single variable or in the

More information

Section 9c. Propensity scores. Controlling for bias & confounding in observational studies

Section 9c. Propensity scores. Controlling for bias & confounding in observational studies Section 9c Propensity scores Controlling for bias & confounding in observational studies 1 Logistic regression and propensity scores Consider comparing an outcome in two treatment groups: A vs B. In a

More information

Linear Combinations of Group Means

Linear Combinations of Group Means Linear Combinations of Group Means Look at the handicap example on p. 150 of the text. proc means data=mth567.disability; class handicap; var score; proc sort data=mth567.disability; by handicap; proc

More information

Exam details. Final Review Session. Things to Review

Exam details. Final Review Session. Things to Review Exam details Final Review Session Short answer, similar to book problems Formulae and tables will be given You CAN use a calculator Date and Time: Dec. 7, 006, 1-1:30 pm Location: Osborne Centre, Unit

More information

Time Invariant Predictors in Longitudinal Models

Time Invariant Predictors in Longitudinal Models Time Invariant Predictors in Longitudinal Models Longitudinal Data Analysis Workshop Section 9 University of Georgia: Institute for Interdisciplinary Research in Education and Human Development Section

More information

Repeated Measures ANOVA Multivariate ANOVA and Their Relationship to Linear Mixed Models

Repeated Measures ANOVA Multivariate ANOVA and Their Relationship to Linear Mixed Models Repeated Measures ANOVA Multivariate ANOVA and Their Relationship to Linear Mixed Models EPSY 905: Multivariate Analysis Spring 2016 Lecture #12 April 20, 2016 EPSY 905: RM ANOVA, MANOVA, and Mixed Models

More information

A brief introduction to mixed models

A brief introduction to mixed models A brief introduction to mixed models University of Gothenburg Gothenburg April 6, 2017 Outline An introduction to mixed models based on a few examples: Definition of standard mixed models. Parameter estimation.

More information

Introduction to Crossover Trials

Introduction to Crossover Trials Introduction to Crossover Trials Stat 6500 Tutorial Project Isaac Blackhurst A crossover trial is a type of randomized control trial. It has advantages over other designed experiments because, under certain

More information

Workshop 9.3a: Randomized block designs

Workshop 9.3a: Randomized block designs -1- Workshop 93a: Randomized block designs Murray Logan November 23, 16 Table of contents 1 Randomized Block (RCB) designs 1 2 Worked Examples 12 1 Randomized Block (RCB) designs 11 RCB design Simple Randomized

More information

Time-Invariant Predictors in Longitudinal Models

Time-Invariant Predictors in Longitudinal Models Time-Invariant Predictors in Longitudinal Models Topics: Summary of building unconditional models for time Missing predictors in MLM Effects of time-invariant predictors Fixed, systematically varying,

More information

STAT 705 Generalized linear mixed models

STAT 705 Generalized linear mixed models STAT 705 Generalized linear mixed models Timothy Hanson Department of Statistics, University of South Carolina Stat 705: Data Analysis II 1 / 24 Generalized Linear Mixed Models We have considered random

More information

Booklet of Code and Output for STAC32 Final Exam

Booklet of Code and Output for STAC32 Final Exam Booklet of Code and Output for STAC32 Final Exam December 8, 2014 List of Figures in this document by page: List of Figures 1 Popcorn data............................. 2 2 MDs by city, with normal quantile

More information

Statistics for exp. medical researchers Comparison of groups, T-tests and ANOVA

Statistics for exp. medical researchers Comparison of groups, T-tests and ANOVA Faculty of Health Sciences Outline Statistics for exp. medical researchers Comparison of groups, T-tests and ANOVA Lene Theil Skovgaard Sept. 14, 2015 Paired comparisons: tests and confidence intervals

More information

More about linear mixed models

More about linear mixed models Faculty of Health Sciences Contents More about linear mixed models Analysis of repeated measurements, NFA 2016 Julie Lyng Forman & Lene Theil Skovgaard Department of Biostatistics, University of Copenhagen

More information

A (Brief) Introduction to Crossed Random Effects Models for Repeated Measures Data

A (Brief) Introduction to Crossed Random Effects Models for Repeated Measures Data A (Brief) Introduction to Crossed Random Effects Models for Repeated Measures Data Today s Class: Review of concepts in multivariate data Introduction to random intercepts Crossed random effects models

More information

ssh tap sas913, sas

ssh tap sas913, sas B. Kedem, STAT 430 SAS Examples SAS8 ===================== ssh xyz@glue.umd.edu, tap sas913, sas https://www.statlab.umd.edu/sasdoc/sashtml/onldoc.htm Multiple Regression ====================== 0. Show

More information

17. Example SAS Commands for Analysis of a Classic Split-Plot Experiment 17. 1

17. Example SAS Commands for Analysis of a Classic Split-Plot Experiment 17. 1 17 Example SAS Commands for Analysis of a Classic SplitPlot Experiment 17 1 DELIMITED options nocenter nonumber nodate ls80; Format SCREEN OUTPUT proc import datafile"c:\data\simulatedsplitplotdatatxt"

More information