O1A Report

Power Analysis

1 Load data

# Read
file = "DATE O1A.xlsx"
Date <- readxl::read_xlsx(file, sheet = "date", skip = 5, col_names = FALSE)

1.1 Processing data

# Clean
varnames <- c("Nr_crt", "ID", "Nume_Prenume", "Zi", "Vas_stres_pre", "Vas_bine_pre",
             sprintf("Stais_pre_%01d", seq(1,20)),
             "SOP",
             "IOS_mama", "IOS_tata", "IOS_iubit", "IOS_prieten", "IOS_personalitate",
             "Vas_rel_global", "Vas_rel_arousal",
             "CRQ_1", "CRQ_2", "CRQ_3", "CRQ_4", "CRQ_5", "CRQ_6",
             "Vas_stres_post", "Vas_bine_post",
             sprintf("Stais_post_%01d", seq(1,20))
)
names(Date) <- varnames   # nume noi
Date <- Date[-c(1:2),]    # scoatem randurile cu numele precedente
Date$Nr_crt <- 1:nrow(Date)   # era gol, asa ca numerotam randurile ca sa avem acesta variabila 
# Process NAs
Date <- Date %>% 
    mutate(NA_per_row = rowSums(is.na(.)))     # count NAs by row
Date <- Date %>% 
    filter(NA_per_row < 20)         # arbitrary cutoff for NAs on columns ... it is normal to have 4 NAs for all columns

1.2 Compute new variables

# Compute new variables 
Conditie <- Date %>% 
    select(Nr_crt, ID, IOS_mama, IOS_tata, IOS_iubit, IOS_prieten, IOS_personalitate) %>% 
    gather(type, value, -c(Nr_crt, ID)) %>% 
    mutate(Conditie = ifelse(!is.na(value), type, NA) ) %>%
    mutate(Conditie = str_replace(Conditie, "IOS_", "")) %>%
    arrange(Nr_crt) %>%
    select(Conditie) %>% na.omit() 
Date$Conditie <- Conditie$Conditie     # tidyverse returns tibble, must do this
IOS <- Date %>% 
  mutate(IOS = coalesce(IOS_mama, IOS_tata, IOS_iubit, IOS_prieten, IOS_personalitate)) %>%
  select(IOS)
Date$IOS <- IOS$IOS   # tidyverse returns tibble, must do this
rm(Conditie, IOS)    # remove 2 tibbles
# Scoring Stai   (convert numeric - VAS)
itemiVAS <- c(5, 6, 41, 42)
itemiStaiS_pre <- 7:26
itemiStaiS_post <- 43:62
ReversedItems <- c(1,2,5,8,10,11,15,16,19,20)
Date <- Date %>%                 
  replace(Date == "na", NA) %>%        # scimbam codarea cu na a Doinei
  mutate_at(vars(itemiStaiS_pre), funs(as.numeric)) %>%        # facem coloanele numerice pt STAI
  mutate_at(vars(itemiStaiS_post), funs(as.numeric)) %>% 
  mutate_at(vars(itemiVAS), funs(as.numeric))
Date[ ,itemiStaiS_pre[ReversedItems]] = 5 - Date[ ,itemiStaiS_pre[ReversedItems]]
Date[ ,itemiStaiS_post[ReversedItems]] = 5 - Date[ ,itemiStaiS_post[ReversedItems]]
Date$StaiS_pre = rowSums(Date[ ,itemiStaiS_pre], na.rm=T ) * NA ^ (rowSums(!is.na(Date[ ,itemiStaiS_pre])) == 0)
Date$StaiS_post = rowSums(Date[ ,itemiStaiS_post], na.rm=T ) * NA ^ (rowSums(!is.na(Date[ ,itemiStaiS_post])) == 0)

2 STAI-Y

2.1 Plots with p values

## STAI plot 
Staimelt <- Date[, c("ID", "Conditie", "StaiS_pre","StaiS_post")] %>% 
  gather("StaiS_pre", "StaiS_post", key = "Stai_cond", value = "value") %>% 
  mutate_at(vars(c(1, 2,3)), funs(as.factor)) %>% 
  mutate(Stai_cond = factor(Stai_cond, levels = c("StaiS_pre","StaiS_post"))) # %>%    # change factor order for plot pre, post
ggplot(Staimelt, aes(x = Stai_cond, y = value)) +
  geom_boxplot() +
  facet_wrap(~Conditie) +
  ggpubr::stat_compare_means(method = "t.test", paired = TRUE, comparisons = list(c("StaiS_pre","StaiS_post")))

2.2 t tests

## STAI tables
Date %>% 
  group_by(Conditie) %>% 
  do(broom::tidy(t.test(.$StaiS_pre, 
                        .$StaiS_post, 
                        mu = 0, 
                        alt = "two.sided", 
                        paired = TRUE, 
                        conf.level = 0.95))) %>%
  knitr::kable(digits = 2)

Conditie	estimate	statistic	p.value	parameter	conf.low	conf.high	method	alternative
iubit	0.71	0.48	0.64	27	-2.34	3.77	Paired t-test	two.sided
mama	0.09	0.09	0.93	33	-1.98	2.16	Paired t-test	two.sided
personalitate	1.44	1.48	0.15	35	-0.53	3.42	Paired t-test	two.sided
prieten	3.27	3.96	0.00	36	1.59	4.95	Paired t-test	two.sided
tata	-0.69	-0.91	0.37	34	-2.21	0.84	Paired t-test	two.sided

2.3 Smaple size (n) for Power = .8, Sig level = .05, computed delta

stai_table <- 
Date %>% 
  group_by(Conditie) %>% 
  dplyr::summarize(mean_pre = mean(StaiS_pre, na.rm=TRUE),
                   mean_post = mean(StaiS_post, na.rm=TRUE),
                   mean_diff = mean(StaiS_pre, na.rm=TRUE) - mean(StaiS_post, na.rm=TRUE),   # m_pre - m_post
                   sd_pooled = sd(c(StaiS_pre, StaiS_post), na.rm=TRUE),                     # pooled sd
                   n_pre = sum(!is.na(StaiS_pre)),                                           # n 
                   n_post = sum(!is.na(StaiS_post)))                                         # n 
stai_power <- 
stai_table  %>% 
  rowwise()  %>%
  do(broom::tidy(power.t.test(power = .8, 
                              delta =  .$mean_diff,
                              sd =   .$sd_pooled,
                              sig.level = .05,
                              type = "paired",
                              alternative = "two.sided")))                                # determine sample size requirement
  
knitr::kable(cbind(stai_table, stai_power),
    digits = 2)

Conditie	mean_pre	mean_post	mean_diff	sd_pooled	n_pre	n_post	n	delta	sd	sig.level	power
iubit	30.82	30.11	0.71	9.74	28	28	1462.52	0.71	9.74	0.05	0.8
mama	28.79	28.71	0.09	7.92	34	34	63270.54	0.09	7.92	0.05	0.8
personalitate	30.44	29.00	1.44	8.05	36	36	245.79	1.44	8.05	0.05	0.8
prieten	32.95	29.68	3.27	9.11	37	37	62.87	3.27	9.11	0.05	0.8
tata	29.54	30.23	-0.69	7.95	35	35	1056.96	0.69	7.95	0.05	0.8

3 VAS Stres

3.1 Plots with p values

## Vas Stres plot 
Vasstresmelt <- Date[, c("ID", "Conditie", "Vas_stres_pre","Vas_stres_post")] %>% 
  gather("Vas_stres_pre","Vas_stres_post", key = "Vas_stres_cond", value = "value") %>% 
  mutate_at(vars(c(1,2,3)), funs(as.factor)) %>% 
  mutate_at(vars(c(4)), funs(as.numeric)) %>% 
  mutate(Vas_stres_cond = factor(Vas_stres_cond, levels = c("Vas_stres_pre","Vas_stres_post"))) # change factor order for plot pre, post
ggplot(Vasstresmelt, aes(x = Vas_stres_cond, y = value)) +
  geom_boxplot() +
  facet_wrap(~Conditie) +
  ggpubr::stat_compare_means(method = "t.test", paired = TRUE, comparisons = list(c("Vas_stres_pre","Vas_stres_post")))

3.2 t tests

## Vas Stres tables
Date %>% 
  group_by(Conditie) %>% 
  do(broom::tidy(t.test(.$Vas_stres_pre, 
                        .$Vas_stres_post, 
                        mu = 0, 
                        alt = "two.sided", 
                        paired = TRUE, 
                        conf.level = 0.95))) %>%
  knitr::kable(digits = 2)

Conditie	estimate	statistic	p.value	parameter	conf.low	conf.high	method	alternative
iubit	1.54	0.52	0.60	27	-4.47	7.55	Paired t-test	two.sided
mama	-1.53	-0.62	0.54	33	-6.56	3.50	Paired t-test	two.sided
personalitate	0.06	0.03	0.97	35	-3.51	3.62	Paired t-test	two.sided
prieten	3.65	1.78	0.08	36	-0.51	7.81	Paired t-test	two.sided
tata	-2.83	-1.43	0.16	34	-6.84	1.18	Paired t-test	two.sided

3.3 Smaple size (n) for Power = .8, Sig level = .05, computed delta

VasS_table <- 
Date %>% 
  group_by(Conditie) %>% 
  dplyr::summarize(mean_pre = mean(Vas_stres_pre, na.rm=TRUE),
                   mean_post = mean(Vas_stres_post, na.rm=TRUE),
                   mean_diff = mean(Vas_stres_pre, na.rm=TRUE) - mean(Vas_stres_post, na.rm=TRUE),   # m_pre - m_post
                   sd_pooled = sd(c(Vas_stres_pre, Vas_stres_post), na.rm=TRUE),                     # pooled sd
                   n_pre = sum(!is.na(Vas_stres_pre)),                                           # n 
                   n_post = sum(!is.na(Vas_stres_post)))                                        # n 
VasS_power <- 
VasS_table  %>%  
  rowwise()  %>%                   
  do(broom::tidy(power.t.test(power = .8, 
                              delta =  .$mean_diff,
                              sd =   .$sd_pooled,
                              sig.level = .05,
                              type = "paired",
                              alternative = "two.sided")))                              # determine sample size requirement
knitr::kable(cbind(VasS_table, VasS_power),
    digits = 2)

Conditie	mean_pre	mean_post	mean_diff	sd_pooled	n_pre	n_post	n	delta	sd	sig.level	power
iubit	18.79	17.25	1.54	20.46	28	28	1395.03	1.54	20.46	0.05	0.8
mama	13.59	15.12	-1.53	15.89	34	34	848.99	1.53	15.89	0.05	0.8
personalitate	17.42	17.36	0.06	14.81	36	36	557933.90	0.06	14.81	0.05	0.8
prieten	24.65	21.00	3.65	19.48	37	37	225.70	3.65	19.48	0.05	0.8
tata	17.80	20.63	-2.83	19.25	35	35	365.58	2.83	19.25	0.05	0.8

4 VAS Wellbeing

4.1 Plots with p values

## Vas Stres plot 
Vasbinemelt <- Date[, c("ID", "Conditie", "Vas_bine_pre","Vas_bine_post")] %>% 
  gather("Vas_bine_pre","Vas_bine_post", key = "Vas_stres_cond", value = "value") %>% 
  mutate_at(vars(c(1,2,3)), funs(as.factor)) %>% 
  mutate_at(vars(c(4)), funs(as.numeric)) %>% 
  mutate(Vas_stres_cond = factor(Vas_stres_cond, levels = c("Vas_bine_pre","Vas_bine_post"))) # change factor order for plot pre, post
ggplot(Vasbinemelt, aes(x = Vas_stres_cond, y = value)) +
  geom_boxplot() +
  facet_wrap(~Conditie) +
  ggpubr::stat_compare_means(method = "t.test", paired = TRUE, comparisons = list(c("Vas_bine_pre","Vas_bine_post")))

4.2 t tests

## Vas Stres tables
Date %>% 
  group_by(Conditie) %>% 
  do(broom::tidy(t.test(.$Vas_bine_pre, 
                        .$Vas_bine_post, 
                        mu = 0, 
                        alt = "two.sided", 
                        paired = TRUE, 
                        conf.level = 0.95))) %>%
  knitr::kable(digits = 2)

Conditie	estimate	statistic	p.value	parameter	conf.low	conf.high	method	alternative
iubit	-2.14	-0.70	0.49	27	-8.40	4.11	Paired t-test	two.sided
mama	-0.21	-0.08	0.94	33	-5.36	4.94	Paired t-test	two.sided
personalitate	-1.69	-0.77	0.44	35	-6.14	2.76	Paired t-test	two.sided
prieten	-5.81	-3.19	0.00	36	-9.51	-2.12	Paired t-test	two.sided
tata	-2.29	-0.65	0.52	34	-9.48	4.91	Paired t-test	two.sided

4.3 Smaple size (n) for Power = .8, Sig level = .05, computed delta

VasB_table <- 
Date %>% 
  group_by(Conditie) %>% 
  dplyr::summarize(mean_pre = mean(Vas_bine_pre, na.rm=TRUE),
                   mean_post = mean(Vas_bine_post, na.rm=TRUE),
                   mean_diff = mean(Vas_bine_pre, na.rm=TRUE) - mean(Vas_bine_post, na.rm=TRUE),   # m_pre - m_post
                   sd_pooled = sd(c(Vas_bine_pre, Vas_bine_post), na.rm=TRUE),                     # pooled sd
                   n_pre = sum(!is.na(Vas_bine_pre)),                                           # n 
                   n_post = sum(!is.na(Vas_bine_post)))                                     # n 
VasB_power <-
VasB_table  %>%  
  rowwise()  %>%                   
  do(broom::tidy(power.t.test(power = .8, 
                              delta =  .$mean_diff,
                              sd =   .$sd_pooled,
                              sig.level = .05,
                              type = "paired",
                              alternative = "two.sided")))                               # determine sample size requirement
knitr::kable(cbind(VasB_table, VasB_power),
    digits = 2)

Conditie	mean_pre	mean_post	mean_diff	sd_pooled	n_pre	n_post	n	delta	sd	sig.level	power
iubit	75.14	77.29	-2.14	21.37	28	28	782.41	2.14	21.37	0.05	0.8
mama	78.44	78.65	-0.21	17.01	34	34	53603.98	0.21	17.01	0.05	0.8
personalitate	74.00	75.69	-1.69	20.20	36	36	1117.79	1.69	20.20	0.05	0.8
prieten	68.86	74.68	-5.81	17.28	37	37	71.37	5.81	17.28	0.05	0.8
tata	71.14	73.43	-2.29	19.14	35	35	552.39	2.29	19.14	0.05	0.8

5 Session Info

R version 3.5.2 (2018-12-20)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=Romanian_Romania.1250  LC_CTYPE=Romanian_Romania.1250    LC_MONETARY=Romanian_Romania.1250
[4] LC_NUMERIC=C                      LC_TIME=Romanian_Romania.1250    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] bindrcpp_0.2.2    ggpubr_0.2        magrittr_1.5      broom_0.5.1       papaja_0.1.0.9842 psych_1.8.10      forcats_0.3.0    
 [8] stringr_1.3.1     dplyr_0.7.8       purrr_0.2.5       readr_1.3.0       tidyr_0.8.2       tibble_1.4.2      ggplot2_3.1.0    
[15] tidyverse_1.2.1   pacman_0.5.0     

loaded via a namespace (and not attached):
 [1] tidyselect_0.2.5 xfun_0.4         haven_2.1.0      lattice_0.20-38  colorspace_1.3-2 generics_0.0.2   htmltools_0.3.6 
 [8] yaml_2.2.0       rlang_0.3.1      pillar_1.3.1     foreign_0.8-71   glue_1.3.0       withr_2.1.2      modelr_0.1.2    
[15] readxl_1.1.0     bindr_0.1.1      plyr_1.8.4       ggsignif_0.4.0   munsell_0.5.0    gtable_0.2.0     cellranger_1.1.0
[22] rvest_0.3.2      evaluate_0.12    labeling_0.3     knitr_1.21       parallel_3.5.2   highr_0.7        Rcpp_1.0.0      
[29] scales_1.0.0     backports_1.1.3  jsonlite_1.6     mnormt_1.5-5     digest_0.6.18    hms_0.4.2        stringi_1.2.4   
[36] grid_3.5.2       cli_1.0.1        tools_3.5.2      lazyeval_0.2.1   crayon_1.3.4     pkgconfig_2.0.2  xml2_1.2.0      
[43] lubridate_1.7.4  rmarkdown_1.11   assertthat_0.2.0 httr_1.4.0       rstudioapi_0.8   R6_2.3.0         nlme_3.1-137    
[50] compiler_3.5.2  

 

---

A work by Claudiu Papasteri

claudiu.papasteri@gmail.com