Take Away Paper: Set Analysis

Design

For this assessment, you chose from the following two scenarios. More information about both can be found on the Discovering Statistics Canvas site and the Developmental Psychology Canvas site.

Scenario 1: Vocabulary, Reading, & Screen time

Building on the work of Rosslund et al. (2024), you investigated the following research question: does the frequency of shared book reading and the duration of daily screen time predict children’s expressive vocabulary?

Scenario 2: Emotion Regulation

Building on the work of Bariola et al. (2012), you investigated the following research question: do the mother’s and father’s tendency towards emotional suppression predict their child’s?

Important Information

Below are the results you must use in order to complete your lab report for Developmental Psychology. You MUST use these results, regardless of the results you obtained via your analysis for the TAP.

Differences From TAP Results

If you have different results, or chose a different model, in your TAP assessment than below,

DON’T PANIC!

These are not the only “right answers” for the TAP assessment. The TAP marking will look at (among other things) the evidence you have used and the justification you provided for your model - remember that we have ‘researcher degrees of freedom’ when conducting an analysis, and multiple different analyses can be equally justifiable. If your analysis does not match the one below, it does not mean you will get a bad mark on the TAP!

Set Analysis

Choose the relevant scenario tab below for the results you must use in your Developmental Psychology lab report.

Scenario 1: Codebook

  • ppt_id the participant/family ID
  • child_gender: the child’s gender
  • child_age: the child’s age in days
  • mum_age: the child’s mother’s age in years
  • dad_age: the child’s father’s age in years
  • n_siblings: the number of siblings
  • outdoor_exploration: frequency engaging in outdoor activities, scored from 0-5 (never, once a month, once a week, several times a week, once a day, several times a day)
  • sensory_play: frequency participating in play involving sensory materials, scored from 0-5 (never, once a month, once a week, several times a week, once a day, several times a day)
  • book_reading: frequency engaging in shared reading activities, scored from 0-5 (never, once a month, once a week, several times a week, once a day, several times a day)
  • screen_time: infants’ daily screen time, scored from 0-5 (none, under 1 hour, 1 hour, 2 hours, 3 hours, 4 hours)
  • vocab_score: infants’ expressive vocabulary (the number of words their infant produced)

Scenario 1: Participant Descriptives

The total number of children was 252.

Age breakdown per participant is given below. Note that child age was measured in days. Parent age is measured in years.

Person Mean SD Min Max
Child 731.70 5.32 715 742
Mother 31.81 3.74 25 39
Father 34.68 5.04 25 46
Child Gender N Percent
Female 142 56.35
Male 110 43.65

Scenario 1: Summary Statistics

Variable Mean SD Min Max 95% CI lower 95% CI upper
Shared Book Reading 3.92 1.15 0 5 3.78 4.06
Screen Time 1.31 1.08 0 5 1.17 1.44
Vocab Score 347.21 175.35 8 720 325.46 368.97

Scenario 1: Model-Building

Model 1: Shared reading predicts expressive vocabulary.

Model Parameters:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 43.3850 33.7700 1.2847 0.2 -23.1249 109.8949
book_reading 77.4949 8.2641 9.3774 < .001 61.2189 93.7710

Model Fit:

r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.2602 0.2573 151.1244 87.9347 < .001 1 -1621.13 3248.261 3258.849 5709643 250 252

Model 2: Both shared reading and screen time predicts expressive vocabulary

Model Parameters:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 152.5885 37.4739 4.0719 < .001 78.7822 226.3947
book_reading 65.6919 8.0934 8.1168 < .001 49.7518 81.6321
screen_time -48.2003 8.6690 -5.5601 < .001 -65.2743 -31.1263

Model Fit:

r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.3419 0.3366 142.821 64.6854 < .001 2 -1606.385 3220.769 3234.887 5079061 249 252

Model Comparison:

Res.Df RSS Df Sum of Sq F Pr(>F)
250 5709643 NA NA NA NA
249 5079061 1 630582.3 30.9142 < .001

Scenario 1: Checking Assumptions

Diagnostic Plots (for the better model):

Casewise Diagnostics:

The following table gives the number (#) and percent (perc) of cases at the 5% level (with absolute values greater than or equal to 1.96) and at the 1% level (with absolute values greater than or equal to 2.5), as well as the number of cases with standardised residuals with absolute values greater than 3.

# Cases (5%) Perc Cases (5%) # Cases (1%) Perc Cases (1%) # Outliers (>3)
13 5.16 2 0.79 0

Other Information:

The max value of the standardised residuals (absolute value) was 2.64, and the max value of Cook’s distance was 0.02. Robust models (robust parameters, robust standard errors, and bootstrapped models) showed the same pattern of results as the original unadjusted model.

Scenario 1: Final Unadjusted Model

Model Parameters and Fit:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 152.5885 37.4739 4.0719 < .001 78.7822 226.3947
book_reading 65.6919 8.0934 8.1168 < .001 49.7518 81.6321
screen_time -48.2003 8.6690 -5.5601 < .001 -65.2743 -31.1263
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.3419 0.3366 142.821 64.6854 < .001 2 -1606.385 3220.769 3234.887 5079061 249 252

Standardised Bs:

Parameter Coefficient SE CI CI_low CI_high t df_error p
(Intercept) 0.000 0.051 0.95 -0.101 0.101 0.000 249 1
book_reading 0.432 0.053 0.95 0.327 0.537 8.117 249 < .001
screen_time -0.296 0.053 0.95 -0.401 -0.191 -5.560 249 < .001

Scenario 1: Codebook

  • ppt_id the participant/family ID
  • child_gender: the child’s gender
  • child_age: the child’s age in years
  • mum_age: the child’s mother’s age in years
  • dad_age: the child’s father’s age in years
  • dad_supp: the father’s expressive suppression score (min 4, max 20)
  • dad_reapp: the father’s cognitive reappraisal score (min 6, max 30)
  • mum_supp: the mother’s expressive suppression score (min 4, max 20)
  • mum_reapp: the mother’s cognitive reappraisal score (min 6, max 30)
  • child_supp: the child’s expressive suppression score (min 4, max 20)
  • child_reapp: the child’s cognitive reappraisal score (min 6, max 30)

Scenario 2: Participant Descriptives

The total number of children was 252.

Age breakdown per participant is given below. Note that both child age and parent age was measured in years.

Person Mean SD Min Max
Child 14.91 2.29 9 19
Mother 40.99 3.83 30 51
Father 42.91 3.95 31 52
Child Gender N Percent
Female 146 57.94
Male 106 42.06

Scenario 2: Summary Statistics

Variable Mean SD Min Max 95% CI lower 95% CI upper
Mother's Suppression 7.90 3.51 4 20 7.46 8.33
Father's Suppression 11.30 3.60 4 20 10.86 11.75
Child's Suppression 14.14 4.21 4 20 13.62 14.67

Scenario 2: Model-Building

Model A: Mother’s expressive suppression predicts child’s suppression.

Model Parameters:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 10.1229 0.5936 17.0539 < .001 8.9538 11.2919
mum_supp 0.5091 0.0687 7.4095 < .001 0.3738 0.6444

Model Fit:

r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1801 0.1768 3.8224 54.9002 < .001 1 -694.471 1394.942 1405.53 3652.718 250 252

Model B: Both mother’s and father’s expressive suppression predict child’s suppression.

Model Parameters:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 8.1953 0.8279 9.8984 < .001 6.5646 9.8259
mum_supp 0.4213 0.0725 5.8074 < .001 0.2784 0.5641
dad_supp 0.2319 0.0708 3.2754 < .001 0.0925 0.3714

Model Fit:

r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.2139 0.2076 3.7501 33.8824 < .001 2 -689.1559 1386.312 1400.43 3501.839 249 252

Model Comparison:

Res.Df RSS Df Sum of Sq F Pr(>F)
250 3652.718 NA NA NA NA
249 3501.839 1 150.8789 10.7283 < .001

Scenario 2: Checking Assumptions

Diagnostic Plots (for the better model):

Casewise Diagnostics:

The following table gives the number (#) and percent (perc) of cases at the 5% level (with absolute values greater than or equal to 1.96) and at the 1% level (with absolute values greater than or equal to 2.5), as well as the number of cases with standardised residuals with absolute values greater than 3.

# Cases (5%) Perc Cases (5%) # Cases (1%) Perc Cases (1%) # Outliers (>3)
14 5.56 4 1.59 0

Other Information:

The max value of the standardised residuals (absolute value) was 2.82, and the max value of Cook’s distance was 0.04. Robust models (robust parameters, robust standard errors, and bootstrapped models) showed the same pattern of results as the original unadjusted model.

Scenario 2: Final Unadjusted Model

Model Parameters and Fit:

term estimate std.error statistic p.value conf.low conf.high
(Intercept) 8.1953 0.8279 9.8984 < .001 6.5646 9.8259
mum_supp 0.4213 0.0725 5.8074 < .001 0.2784 0.5641
dad_supp 0.2319 0.0708 3.2754 < .001 0.0925 0.3714
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.2139 0.2076 3.7501 33.8824 < .001 2 -689.1559 1386.312 1400.43 3501.839 249 252

Standardised Bs:

Parameter Coefficient SE CI CI_low CI_high t df_error p
(Intercept) 0.000 0.056 0.95 -0.110 0.110 0.000 249 1
mum_supp 0.351 0.060 0.95 0.232 0.470 5.807 249 < .001
dad_supp 0.198 0.060 0.95 0.079 0.317 3.275 249 < .001