Data Cleaning Process

The Client Enrollment, Client Services, and Client Pre-Post data files were merged utilizing R version 4.3.2 within the RStudio programming software. An analytical sample of 65 responses was identified for inclusion in this report following the removal of three respondents who were either missing a pre- or post-survey.

Data Manipulation

A final dataset was condensed and manipulated to identify variables of need and to create new variables to assist with analyses. Specifically, pertinent demographic variables and Parenting Ladder responses remained from the original data files. The following demographic variables were created:

  • Region - This variable was created to represent the three geographic regions of the seven cities within Ventura County, California included in the sample. Due to my unfamiliarity with the region, I consulted the AI chatbot ChatGPT to explore ways to categorize the seven cities into three regions. Ultimately, the cities of Oxnard, Port Hueneme, and Ventura were coded as Western Ventura County, the city of Santa Paula as Central Ventura County, and Moorpark, Newbury Park, and Simi Valley as Eastern Ventura County. An alternative to grouping cities by geographic region within the county would be to utilize census data to identify socioeconomic similarities between cities, such as educational attainment, median income level, and ethnicity.
  • CompletedSessions and CompletedSessions2 - These two variables were created to represent categorical levels of the number of sessions completed, as measured by the TotalNumberofServices variable. The CompletedSessions variable was divided into four levels using the standard deviation of the TotalNumberofServices variable, which includes the levels of 1-30 Sessions, 31-60 Sessions, 61-90 Sessions, and 91 Sessions or more. To create more balance on this variable, CompletedSessions2 was created with two levels designed more closely around the mean of 34.5 sessions completed, resulting in the levels of 1-30 Sessions and 31 Sessions or more.
  • Change variables were created per survey item to represent the numeric difference from pre- to post-survey.

Question 1: Pre-Post Change for Scaled Items 1-16 and the Three Frequency Questions

Overall Change Summary

On average, all items experienced positive changed from pre- to post-survey, suggesting general increases across all domains of the Parenting Ladder survey. The average item increase ranged from 0.62 (Item 6) to 1.82 (Item 1). The domains of knowledge and confidence experienced the largest increases from pre- to post-survey, with the knowledge items generating average increases of 1.82 and 1.51, respectively. The three confidence items yielded average increases of 1.09, 1.4, and 1.58. Items representing parent perception of having a positive parent-child relationship (Items 8 and 9), additionally experienced similar increases from pre- to post with change values of 1.38 and 1.22 respectively. The domain representing parent perception of their ability to engage in parenting practices (Items 10 and 11) generated inconsistent positive change from pre- to post. Item 10, Know fun activities to help my child learn, yielded an average increase of 1.58, while Item 11, Play with my child frequently, only generated an average increase of 0.82 from pre- to post. The social connection domain (Items 12 and 13) increased from pre- to post, but Item 12 only increased on average of 0.85, and Item 13 only increased an average of 0.71. Items 14-16 all generated average increases around 1, with Item 14 producing an average increase of 1.09, Item 15 an average increase of 1.04, and Item 16 an average increase of 1. Of note, Items 14 and 15 specifically represent the Connections to resources domain. The frequency items produced an average increase ranging from 0.87 (Frequency Item 2) to 1.02 (Frequency Item 1). The average change in Frequency Item 3 (0.89) was similar to Frequency Item 2.

Of note, items representing the knowledge, confidence, and social connection domains experienced a larger amount of growth from pre- to post-survey as these items were ranked lower at the pre-survey, allowing there to be a larger range for growth. Compared to Item 6, for example, the ability to keep one’s child safe and healthy, produced the highest pre-survey rating of all items, lowering its capacity for growth at the post-survey. Moreover, pre-survey results suggest that parents feel more confident in their ability to keep their child physically safe but feel less knowledgeable in their child’s growth and development and less confident in their parenting abilities prior to completing the parenting program.

Item 1

Perceived knowledge of how my child is growing and developing significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 2

Perceived knowledge of what behavior is typical at this age significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 3

Perceived confidence in oneself as a parent significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 4

Perceived confidence in the ability to help one’s child learn significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 5

Perceived confidence in the ability to set limits for one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 6

Perceived ability to keep one’s child safe and healthy significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 7

Perceived ability to recognize when one’s child is upset significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 8

Perceived ability to find positive ways to guide and discipline one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 9

Perceived ability to listen to one’s child and understand their feelings significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 10

Perceived knowledge of fun activities to help one’s child learn significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 11

Perceived frequency of playing with one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 12

The number of families an individual can depend on for support significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 13

Perceived strength of connections to other families significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 14

Perceived ability to obtain services that are needed for one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 15

Perceived ability to obtain services that are needed for oneself significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item 16

Perceived ability to manage the day-to-day stress of being a parent significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Frequency Item 1

The frequency (days per week) of reading or looking at picture books with one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Frequency Item 2

The frequency (days per week) of talking or sharing stories with one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Frequency Item 3

The frequency (days per week) of singing with one’s child significantly increased from pre- to post-survey.

Mean Summary

Box Plot

Item test-retest reliability

To evaluate the test-retest reliability of survey items, Pearson Correlation values were calculated using SPSS. Evidence of test-retest reliability was considered present if Pearson Correlation coefficients per item were equal to or greater than 0.7. The table below displays correlation coefficients for all Parenting Ladder and Frequency items. Two items from the Parenting Ladder survey, items 12 and 13, and one frequency item generated Pearson Correlation values exceeding the 0.7 threshold, suggesting these items established test-retest reliability. The frequency item to generate test-retest reliability was the item prompting respondents to report the frequency with which they read or look at picture books with their children. The two Parenting Ladder items to establish test-retest reliability were the two questions from the social connection and support domain.

Question 2: Are there subgroups that experienced more (or less) change than others?

To explore differences in the magnitude of change between different subgroups, the collapsed variable of completed sessions with levels of 1-30 Sessions and 31 Sessions or more was compared across the items representing knowledge obtained before and after the parenting program (Items 1 and 2).

The tabs below display an interesting and diverging trend between the knowledge domain questions when comparing responses dependent upon the number of sessions attended. Two linear models were calculated to evaluate how the number of sessions predicted pre- to post- change for items 1 and 2. The item 1 model determined that individuals who attended 1-30 sessions obtained more growth at the post-survey for the statement: “Knowledge of how my child is growing and developing”. Conversely, the item 2 model for the statement “Knowledge of what behavior is typical at this age” identified more growth at the post-survey for individuals who completed 31 or more sessions. One explanation for this discrepancy is that individuals who participated in fewer sessions perceived an increase in knowledge about their own child, while knowledge for child development, in this case behavior, more greatly increased with attending a higher number of sessions.

Item 1 Linear Model

## 
## Call:
## lm(formula = Item1Change ~ CompletedSessions2, data = pl_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -3.0000 -0.9118  0.0000  1.0000  3.1765 
## 
## Coefficients:
##                                       Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                             2.0000     0.3297   6.067  5.1e-07 ***
## CompletedSessions231 Sessions or more  -0.1765     0.4993  -0.353    0.726    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 1.546 on 37 degrees of freedom
##   (26 observations deleted due to missingness)
## Multiple R-squared:  0.003364,   Adjusted R-squared:  -0.02357 
## F-statistic: 0.1249 on 1 and 37 DF,  p-value: 0.7258

Item 1 Summary

Item 1 Plot

Item 2 Linear Model

## 
## Call:
## lm(formula = Item2Change ~ CompletedSessions2, data = pl_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2.4545 -1.1390 -0.4545  0.5455  3.5455 
## 
## Coefficients:
##                                       Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                             1.4545     0.3211   4.530 5.96e-05 ***
## CompletedSessions231 Sessions or more   0.3690     0.4863   0.759    0.453    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 1.506 on 37 degrees of freedom
##   (26 observations deleted due to missingness)
## Multiple R-squared:  0.01532,    Adjusted R-squared:  -0.01129 
## F-statistic: 0.5756 on 1 and 37 DF,  p-value: 0.4528

Item 2 Summary

Item 2 Plot

Question 3: Conclusion of Parenting Ladder instrument

Overall, the Parenting Ladder survey appears to be a valid measurement to evaluate parent perception of their abilities prior to, and after participating in the parent training. I think a useful next step to evaluate this measurement would be to analyze a larger sample of data and to also consider utilizing an exploratory factor analysis or pathway analysis to explore how the items from each domain reliability correlate to each other.

My one critique of the measurement, however, has more to do with the methodology in which the data was collected rather than the content and face validity of the instrument. The pre- and post-survey data appear to have been collected simultaneously at the final session of the training, where the participants were prompted to complete their “post” scores before reporting their “pre-training” scores. I believe this introduces a response bias and indicates that results should be interpreted with caution. I think a more valid way to analyze pre- and post-survey data of this instrument would be to collect pre-survey data prior to the first session and the post-survey data after the completion of the final session. Now, that is the ideal scenario, and contextual research within a natural environment should always be dependent upon what is most feasible for the stakeholders collecting the data and participants in the research process.