5 Synthesis

Alright, here I visualize the findings from the analysis section. I’ll create a summary figure for each of the three blocks.

For that, I’ll extract the model estimates of the effects from each model first.

model1 <- read_rds(here("models", "model1.rds"))
model2 <- read_rds(here("models", "model2.rds"))
model3 <- readRDS(here("models", "model3.rds"))
model4 <- read_rds(here("models", "model4.rds"))
model5 <- read_rds(here("models", "model5.rds"))
model6 <- readRDS(here("models", "model6.rds"))
model7 <- readRDS(here("models", "model7.rds"))
model8 <- read_rds(here("models", "model8.rds"))
model9 <- read_rds(here("models", "model9.rds"))

First we get all estimates.

estimates <- 
  tibble(
    block = as.character(),
    model = as.character(),
    predictor = as.character(),
    estimate = as.double(),
    ci_low = as.double(),
    ci_high = as.double(),
    outcome = as.character()
  )

models <- 
  paste0(
    "model",
    1:9
  )

for (model in models) {
  m <- get(model)
  
  dat <- 
    fixef(m) %>% # get fixed effects
    as_tibble(., rownames = "predictor") %>% # turn to tibble and save predictor variable 
    filter(predictor != "Intercept") %>%  # remove intercept
    select(-Est.Error) %>%  # drop variable
    mutate( # add the outcome by turning formula into character and selecting the first section
      outcome = as.character(m$formula)[1] %>% 
        str_replace(., "\n", "") %>% # remove line breaks
        str_remove(., "\\ (.*)") # select the outcome name
    ) %>% 
    rename(
      estimate = Estimate,
      ci_low = `Q2.5`,
      ci_high = `Q97.5`
    ) %>% 
    mutate(
      model = model, # original name
      # block of the RQ
      block = case_when(
        parse_number(model) %in% 1:3 ~ "Personality on smartphone use",
        parse_number(model) %in% 4:6 ~ "Experiences on smartphone use",
        parse_number(model)%in% 7:9 ~ "Smartphone use on well-being"
      )
    )
  
  # bind them all into a tibble
  estimates <-
    bind_rows(
      estimates,
      dat
    )
}

Next, let’s make some cosmetic changes.

estimates <- 
  estimates %>% 
  # get odds for estimates of the gamma models
  mutate(
    across(
      estimate:ci_high,
      ~ case_when(
        str_detect(outcome, "social_media") ~ exp(.x),
        TRUE ~ .x
      )
    )
  ) %>% 
  # create variable that indicates within or between for models on the state level and whether predictors are state or trait
  mutate(
    between_within = case_when(
      str_detect(predictor, "between") ~ "between",
      str_detect(predictor, "within") ~ "within",
      TRUE ~ NA_character_
    ),
    trait_state = case_when(
      block != "Personality on smartphone use" ~ "state",
      TRUE ~ "trait"
    )
  ) %>% 
  # clean up level names
  mutate(
    across(
      c(predictor, outcome),
      ~ case_when(
        str_detect(.x, "objective") ~ "Objective",
        str_detect(.x, "subjective") ~ "Subjective",
        str_detect(.x, "_c$|between|within") ~ str_to_title(str_remove(.x, "_.*")),
        str_detect(.x, "well") ~ "Well-being",
        TRUE ~ str_to_title(.x)
      )
    )
  ) %>% 
  # give a value for the line that signifies no effect
  mutate(
    line = if_else(outcome %in% c("Objective", "Subjective"), 1, 0)
  )

5.1 Numerical model estimates

In the paper, we don’t present numerical estimates because we ran that many models and believe a figure is more adequate to give readers a good idea of our data. Here, we show all estimates from the analysis section in one place.

RQ	Estimate	Predictor	Outcome	Lower 95% CI	Upper 95% CI	Predictor between or within	Predictor trait or state
Personality on smartphone use	1.0012650	Openness	Objective	0.8561624	1.1682906	NA	trait
Personality on smartphone use	0.9928173	Conscientiousness	Objective	0.8486894	1.1599741	NA	trait
Personality on smartphone use	1.1031411	Extraversion	Objective	0.9634127	1.2664122	NA	trait
Personality on smartphone use	0.9353134	Agreeableness	Objective	0.8010626	1.0910298	NA	trait
Personality on smartphone use	1.1220985	Neuroticism	Objective	0.9498516	1.3186005	NA	trait
Personality on smartphone use	0.9088681	Competence	Objective	0.7874181	1.0541874	NA	trait
Personality on smartphone use	0.9915935	Relatedness	Objective	0.8797513	1.1192422	NA	trait
Personality on smartphone use	0.9639844	Autonomy	Objective	0.8472285	1.0954242	NA	trait
Personality on smartphone use	0.9564898	Openness	Subjective	0.8142978	1.1263943	NA	trait
Personality on smartphone use	0.9270839	Conscientiousness	Subjective	0.7891877	1.0857180	NA	trait
Personality on smartphone use	1.0599241	Extraversion	Subjective	0.9218669	1.2184596	NA	trait
Personality on smartphone use	0.9607391	Agreeableness	Subjective	0.8240315	1.1259502	NA	trait
Personality on smartphone use	1.0572651	Neuroticism	Subjective	0.9030043	1.2374448	NA	trait
Personality on smartphone use	0.9560225	Competence	Subjective	0.8268558	1.1044105	NA	trait
Personality on smartphone use	0.9797460	Relatedness	Subjective	0.8650780	1.1079526	NA	trait
Personality on smartphone use	0.9098677	Autonomy	Subjective	0.7988148	1.0413136	NA	trait
Personality on smartphone use	-16.8059530	Openness	Error	-42.6156187	9.7989453	NA	trait
Personality on smartphone use	-10.9245239	Conscientiousness	Error	-36.6795446	15.1357417	NA	trait
Personality on smartphone use	-3.4464908	Extraversion	Error	-22.7416868	16.0504478	NA	trait
Personality on smartphone use	-8.8490298	Agreeableness	Error	-34.3206265	16.6875642	NA	trait
Personality on smartphone use	-21.8877449	Neuroticism	Error	-48.2662733	4.5817418	NA	trait
Personality on smartphone use	11.3090672	Competence	Error	-10.3417279	32.5748817	NA	trait
Personality on smartphone use	-0.7915929	Relatedness	Error	-18.8548668	16.6994099	NA	trait
Personality on smartphone use	-2.6627363	Autonomy	Error	-21.2450482	16.3289309	NA	trait
Experiences on smartphone use	0.9782329	Autonomy	Objective	0.8430962	1.1319678	between	state
Experiences on smartphone use	1.0082515	Competence	Objective	0.8776508	1.1577488	between	state
Experiences on smartphone use	0.9921151	Relatedness	Objective	0.8634949	1.1467975	between	state
Experiences on smartphone use	0.9655522	Satisfied	Objective	0.8352040	1.1195041	between	state
Experiences on smartphone use	1.0466593	Boring	Objective	0.9346933	1.1708973	between	state
Experiences on smartphone use	1.0051358	Stressful	Objective	0.9087241	1.1110273	between	state
Experiences on smartphone use	1.0760257	Enjoyable	Objective	0.9300967	1.2443189	between	state
Experiences on smartphone use	0.9936441	Autonomy	Objective	0.8884132	1.1105737	within	state
Experiences on smartphone use	0.9428482	Competence	Objective	0.8471665	1.0517166	within	state
Experiences on smartphone use	1.0119016	Relatedness	Objective	0.8983982	1.1367532	within	state
Experiences on smartphone use	1.0155899	Satisfied	Objective	0.9214768	1.1197386	within	state
Experiences on smartphone use	0.9896022	Boring	Objective	0.9177769	1.0666320	within	state
Experiences on smartphone use	0.9808938	Stressful	Objective	0.9154929	1.0519858	within	state
Experiences on smartphone use	1.0304687	Enjoyable	Objective	0.9404241	1.1300332	within	state
Experiences on smartphone use	0.9300347	Autonomy	Subjective	0.8007163	1.0780199	between	state
Experiences on smartphone use	0.9110952	Competence	Subjective	0.7933090	1.0511974	between	state
Experiences on smartphone use	1.0069004	Relatedness	Subjective	0.8719371	1.1594090	between	state
Experiences on smartphone use	0.9078853	Satisfied	Subjective	0.7824334	1.0543062	between	state
Experiences on smartphone use	1.0990411	Boring	Subjective	0.9801975	1.2311815	between	state
Experiences on smartphone use	0.9792680	Stressful	Subjective	0.8864074	1.0833942	between	state
Experiences on smartphone use	1.1221182	Enjoyable	Subjective	0.9673823	1.2974436	between	state
Experiences on smartphone use	0.9961638	Autonomy	Subjective	0.8881633	1.1178577	within	state
Experiences on smartphone use	0.9622879	Competence	Subjective	0.8586757	1.0773853	within	state
Experiences on smartphone use	0.9959550	Relatedness	Subjective	0.8811687	1.1249255	within	state
Experiences on smartphone use	0.9738658	Satisfied	Subjective	0.8838940	1.0750254	within	state
Experiences on smartphone use	1.0204638	Boring	Subjective	0.9446895	1.1036047	within	state
Experiences on smartphone use	0.9762666	Stressful	Subjective	0.9079664	1.0490854	within	state
Experiences on smartphone use	1.0417190	Enjoyable	Subjective	0.9432723	1.1486231	within	state
Experiences on smartphone use	-3.0436910	Autonomy	Error	-24.8136383	18.6602039	between	state
Experiences on smartphone use	-9.9727821	Competence	Error	-29.5460975	9.6907612	between	state
Experiences on smartphone use	5.5548291	Relatedness	Error	-13.2576986	24.3494590	between	state
Experiences on smartphone use	-13.0003871	Satisfied	Error	-35.4784744	9.2554794	between	state
Experiences on smartphone use	4.9723429	Boring	Error	-8.7595478	18.2959361	between	state
Experiences on smartphone use	-7.5493568	Stressful	Error	-18.9017106	3.9205180	between	state
Experiences on smartphone use	6.8753807	Enjoyable	Error	-14.5185437	28.8908163	between	state
Experiences on smartphone use	-0.2548933	Autonomy	Error	-6.8289289	6.2160828	within	state
Experiences on smartphone use	-0.3476491	Competence	Error	-6.5695029	5.9160020	within	state
Experiences on smartphone use	0.4799661	Relatedness	Error	-8.2108490	8.6383165	within	state
Experiences on smartphone use	-6.0554989	Satisfied	Error	-12.1136872	0.0585069	within	state
Experiences on smartphone use	2.7004403	Boring	Error	-1.1717077	6.7490586	within	state
Experiences on smartphone use	-0.4921211	Stressful	Error	-4.3427572	3.4020066	within	state
Experiences on smartphone use	2.6270650	Enjoyable	Error	-2.3714624	7.6912586	within	state
Smartphone use on well-being	0.0065712	Objective	Well-being	-0.0732838	0.0848080	between	state
Smartphone use on well-being	-0.0167289	Objective	Well-being	-0.0903201	0.0577650	within	state
Smartphone use on well-being	-0.0128997	Subjective	Well-being	-0.0824097	0.0562883	between	state
Smartphone use on well-being	-0.0280263	Subjective	Well-being	-0.0791005	0.0238904	within	state
Smartphone use on well-being	-0.0232013	Error	Well-being	-0.1618117	0.1150030	between	state
Smartphone use on well-being	-0.0284695	Error	Well-being	-0.1469967	0.0776612	within	state

5.2 Personality on smartphone use (trait)

Okay, next I create the first summary figure. We see that none of the trait underlying true trait relations (conditional on our model) exclude zero. Neuroticism and openness might be related to less error in estimating smartphone use, but even their credible interval includes zero.

# function for breaks
my_breaks <- 
  function(x) {
    if (max(x) > 5){
      c(-50, 0, 50)
    } else {
    c(0.8, 1, 1.2)
    }
  }

# function for limits
my_limits <- 
  function(x) {
    if (max(x) > 5){
      c(-50, 50)
    } else {
    c(0.65, 1.35)
    }
  }

# plot
ggplot(
    estimates %>% 
      filter(block == "Personality on smartphone use") %>% 
      mutate(
        outcome = as.factor(outcome),
        outcome = fct_recode(
          outcome,
          "Objective (min)" = "Objective",
          "Subjective (min)" = "Subjective",
          "Accuracy (%)" = "Error"
        ),
        
        # reorder predictors
        predictor = as.factor(predictor),
        predictor = fct_rev(fct_relevel(
          predictor,
          "Autonomy",
          "Competence",
          "Relatedness",
          "Agreeableness",
          "Conscientiousness",
          "Extraversion",
          "Neuroticism",
          "Openness"
        )
      )),
  aes(
    x = estimate,
    y = predictor
  )
) +
  geom_vline(
    aes(
      xintercept = line
    ),
    color = "#999999",
    linetype = "dashed"
  ) +
  geom_pointrange(
    aes(
      xmin = ci_low,
      xmax = ci_high
    )#,
    # position = position_dodge2(.6, reverse = TRUE)
  ) +
  facet_wrap(
    ~ outcome,
    scales = "free_x"
  ) +
  scale_x_continuous(breaks = my_breaks, limits = my_limits) +
  theme(
    axis.title.y = element_blank(),
    axis.title.x = element_blank(),
    axis.ticks.y = element_blank(),
    axis.line.y = element_blank(),
    strip.background.x = element_blank(),
    strip.background.y = element_blank(),
    text = element_text(family = "Corbel")
  ) -> figure4

figure4

ggsave(
  here("figures", "figure4.tiff"),
  plot = figure4,
  width = 21 * 0.8,
  height = 29.7 * 0.4,
  units = "cm",
  dpi = 300
)

5.3 Experiences on smartphone use (state)

I’ll move on to the next block investigating state relations between experiences and smartphone use.

ggplot(
    estimates %>% 
      filter(block == "Experiences on smartphone use") %>% 
      mutate(
        outcome = as.factor(outcome),
        outcome = fct_recode(
          outcome,
          "Objective (min)" = "Objective",
          "Subjective (min)" = "Subjective",
          "Accuracy (%)" = "Error"
        ),
        predictor = as.factor(predictor),
        predictor = fct_rev(fct_relevel(
          predictor,
          "Autonomy",
          "Competence",
          "Relatedness",
          "Boring",
          "Enjoyable",
          "Satisfied",
          "Stressful"
        ))
      ),
  aes(
    x = estimate,
    y = predictor,
    color = between_within
  )
) +
  geom_vline(
    aes(
      xintercept = line
    ),
    color = "#999999",
    linetype = "dashed"
  ) +
  geom_pointrange(
    aes(
      xmin = ci_low,
      xmax = ci_high
    ),
    position = position_dodge2(.6, reverse = TRUE)
  ) +
  facet_wrap(
    ~ outcome,
    scales = "free_x"
  ) +
  scale_x_continuous(breaks = my_breaks, limits = my_limits) +
  scale_color_manual(values = c("#009E73", "#D55E00")) +
  scale_fill_manual(values = c("#009E73", "#D55E00")) +
  theme(
    axis.title.y = element_blank(),
    axis.title.x = element_blank(),
    axis.ticks.y = element_blank(),
    axis.line.y = element_blank(),
    strip.background.x = element_blank(),
    strip.background.y = element_blank(),
    legend.title = element_blank(),
    legend.position = "bottom",
    text = element_text(family = "Corbel")
  ) -> figure5

figure5

ggsave(
  here("figures", "figure5.tiff"),
  plot = figure5,
  width = 21 * 0.8,
  height = 29.7 * 0.4,
  units = "cm",
  dpi = 300
)

5.4 Smartphone use on well-being (state)

I’ll move on to the next block investigating state relations between smartphone use and well-being.

ggplot(
    estimates %>% 
      filter(block == "Smartphone use on well-being") %>% 
      mutate(
        predictor = as.factor(predictor),
        predictor = fct_recode(
          predictor,
          "Objective (h)" = "Objective",
          "Subjective (h)" = "Subjective",
          "Accuracy (%, standardized)" = "Error"
        )
      ),
  aes(
    x = estimate,
    y = predictor,
    color = between_within
  )
) +
  geom_vline(
    aes(
      xintercept = line
    ),
    color = "#999999",
    linetype = "dashed"
  ) +
  geom_pointrange(
    aes(
      xmin = ci_low,
      xmax = ci_high
    ),
    position = position_dodge2(.6, reverse = TRUE)
  ) +
  scale_x_continuous(breaks = c(-0.2, 0, 0.2), limits = c(-0.21, 0.21)) +
  scale_color_manual(values = c("#009E73", "#D55E00")) +
  scale_fill_manual(values = c("#009E73", "#D55E00")) +
  theme(
    axis.title.y = element_blank(),
    axis.title.x = element_blank(),
    axis.ticks.y = element_blank(),
    axis.line.y = element_blank(),
    strip.background.x = element_blank(),
    strip.background.y = element_blank(),
    legend.title = element_blank(),
    legend.position = "bottom",
    text = element_text(family = "Corbel")
  ) -> figure6

figure6

ggsave(
  here("figures", "figure6.tiff"),
  plot = figure6,
  width = 21 * 0.8,
  height = 29.7 * 0.4,
  units = "cm",
  dpi = 300
)